diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index a37e110..22c4542 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -53,7 +53,7 @@ jobs:
           enable-cache: true
 
       - name: Install dependencies
-        run: uv sync
+        run: uv sync --extra native
 
       - name: Cache core wasm fixture
         uses: actions/cache@v4
diff --git a/README.md b/README.md
index c3323fa..4747b30 100644
--- a/README.md
+++ b/README.md
@@ -1,122 +1,100 @@
 # chonkle
 
-Chonkle is a demonstrator, built to explore WebAssembly (Wasm) as a codec delivery mechanism for chunked array formats like Zarr and COG. In these formats, each chunk is encoded by a sequence (i.e., a pipeline) of codecs applied in order — bytes, compression, prediction filters, etc. Chonkle pipelines can mix standard Python codecs (via [numcodecs](https://numcodecs.readthedocs.io/)) with custom Wasm codecs that run at near-native speed inside a sandbox — portable, safe, and free from platform-specific build tooling. See [WASM.md](docs/WASM.md) for details on how Wasm codecs work. The library is functional but should be treated as a learning artifact; our understanding of Wasm is still evolving.
+A Python host for Wasm codec pipelines. Pipelines are directed acyclic graphs (DAGs) of codec steps defined in JSON. The orchestrator parses the DAG, validates wiring against codec signatures, and executes the pipeline via [Wasmtime](https://wasmtime.dev/).
 
-## Install
+Status: proof of concept.
 
-Requires [uv](https://docs.astral.sh/uv/).
+## Codec backends
 
-```bash
-uv sync
-```
-
-## How codec pipelines work
-
-Each chunk has a sidecar JSON file that describes the codec pipeline used to encode it:
-
-```json
-{
-  "codecs": [
-    {
-      "name": "bytes",
-      "type": "numcodecs",
-      "configuration": {
-        "endian": "little",
-        "data_type": "uint16",
-        "shape": [1024, 1024]
-      }
-    },
-    {
-      "name": "tiff_predictor_2",
-      "type": "wasm",
-      "uri": "file://path/to/tiff-predictor-2-c.wasm",
-      "configuration": {
-        "bytes_per_sample": 2,
-        "width": 1024
-      }
-    },
-    {
-      "name": "zlib",
-      "type": "numcodecs",
-      "configuration": {
-        "level": 9
-      }
-    }
-  ]
-}
-```
+chonkle supports three codec backends. Each implements the same `Codec` ABC (`call(direction, port_map)` and `signature()`), so backends can be mixed freely within a single pipeline.
 
-**Encoding** applies codecs in forward order (top to bottom): array → bytes → tiff_predictor_2 (Wasm) → zlib → compressed bytes.
+**Component Model Wasm** — `.wasm` components implementing the `chonkle:codec/transform@0.1.0` WIT interface. Any language with a Component Model toolchain (Rust, C, Python via componentize-py) can produce a conforming component. Data transfer uses the canonical ABI. The Wasmtime sandbox isolates each component from the host.
 
-**Decoding** applies codecs in reverse order, unwinding the encoding.
+**Core Wasm** — wasm32-wasi reactor modules using a binary port-map wire format via `Memory.read`/`Memory.write`. When consecutive pipeline steps are both core wasm, data transfers between their linear memories use `ctypes.memmove` (single-copy, no serialization round-trip).
 
-Each codec entry has:
+**Native (numcodecs)** — Python codecs from the [numcodecs](https://numcodecs.readthedocs.io/) library. No Wasm overhead. `numcodecs` and `numpy` are optional dependencies, imported lazily. Adding a new numcodecs codec requires only adding a signature file.
 
-- `"name"` — codec identifier (for numcodecs lookup and human readability)
-- `"type"` — `"numcodecs"` or `"wasm"`
-- `"configuration"` — codec-specific parameters
-- `"uri"` — (Wasm only) URI of the `.wasm` module: `file://`, `https://`, or `oci://`
+The `Resolver` selects among available implementations using a configurable backend preference list. The default preference order is `["native", "core", "component"]`.
 
-Python and Wasm codec steps can be freely mixed in any order. For information on how Wasm codecs work, see [WASM.md](docs/WASM.md).
+## Usage
 
-## Python API
+### CLI
 
-```python
-from chonkle import decode, encode, get_codecs
-```
+```bash
+# Run a pipeline
+chonkle run pipeline.json --input bytes=chunk.bin --output bytes=out.bin
 
-### Load codec specs
+# With resolver options
+chonkle run pipeline.json --input bytes=chunk.bin \
+  --direction decode \
+  --codec-store ./codec/ \
+  --preference core,component,native \
+  --override zlib=zlib-rs \
+  --source zlib=https://example.com/zlib.wasm
 
-`get_codecs` extracts the codec spec list from a pipeline JSON file or a dict:
+# List installed codecs
+chonkle codecs
 
-```python
-from pathlib import Path
-from chonkle import get_codecs
+# Show details for a specific codec
+chonkle codecs zlib
 
-codecs = get_codecs(Path("pipeline.json"))       # from a file
-codecs = get_codecs({"codecs": [...]})           # from a dict
+# Embed a signature into a .wasm binary (build-time tool)
+chonkle embed-signature codec.wasm signature.json
 ```
 
-### Decode
-
-`decode` applies a codec pipeline in reverse order to raw bytes, returning a numpy array:
+### Python API
 
 ```python
-from chonkle import decode, get_codecs
+from chonkle.pipeline import prepare
+from chonkle.executor import run
 
-codecs = get_codecs(Path("chunks/0.json"))
-arr = decode(Path("chunks/0").read_bytes(), codecs)
+prepared = prepare("pipeline.json", direction="decode")
+outputs = run(prepared, {"bytes": chunk_bytes})
 ```
 
-### Encode
+## Format drivers
 
-`encode` applies a codec pipeline in forward order, returning encoded bytes:
+The executor is format-agnostic. It accepts a pipeline DAG and chunk data, runs the codecs, and returns the result. It has no knowledge of Zarr, Parquet, COG, ORC, or any other file format.
 
-```python
-from chonkle import encode
+A **format driver** is the layer above the executor that bridges a specific file format and the pipeline executor. It reads format-specific metadata, translates it into a pipeline DAG, supplies metadata-derived inputs, and manages chunk I/O. Format drivers are outside the scope of this repository.
 
-encoded = encode(arr, codecs)
-```
+## Documentation
 
-## Demo
+- [docs/OVERVIEW.md](docs/OVERVIEW.md) — Architecture, design rationale, and execution model
+- [docs/reference/PIPELINE_SCHEMA.md](docs/reference/PIPELINE_SCHEMA.md) — Pipeline JSON schema
+- [docs/reference/codec-contract/](docs/reference/codec-contract/) — Codec interface specs (Component Model, Core Wasm, Native)
+- [docs/reference/CODEC_RESOLUTION.md](docs/reference/CODEC_RESOLUTION.md) — Codec resolution chain and backend preference
 
-See [demo/](demo/) for a Jupyter notebook demonstrating the full pipeline with a real Sentinel-2 COG tile.
+See [docs/README.md](docs/README.md) for the full index.
 
-## CLI
+## Development
 
-A `chonkle` CLI is available for interactive use; run `chonkle --help` for usage.
+- **Package manager**: uv
+- **Build backend**: hatchling
+- **Python**: >= 3.13
+- **Linting/formatting**: ruff
+- **Type checking**: mypy
+- **Testing**: pytest
+- **Pre-commit**: ruff check, ruff format, mypy, yaml/toml validation
+- **CI**: GitHub Actions (lint on 3.14, test on 3.13 and 3.14)
 
-## Configuration
+```bash
+# Install dependencies
+uv sync
 
-Wasm codecs downloaded from HTTPS or OCI sources are cached locally to avoid redundant network requests.
+# Include native (numcodecs) backend
+uv sync --extra native
 
-| Variable | Description | Default |
-| --- | --- | --- |
-| `CHONKLE_CACHE_DIR` | Override the Wasm module cache directory | `$TMPDIR/chonkle/wasm/` |
-| `CHONKLE_FORCE_DOWNLOAD` | Set to `1` to re-download cached Wasm modules, bypassing the local cache. Primarily useful for testing and development | unset |
+# Run tests
+uv run pytest
 
-`$TMPDIR` is the OS temporary directory (e.g. `/tmp` on Linux, `/var/folders/...` on macOS). Run `echo $TMPDIR` to see the value on your system.
+# Run linter
+uv run ruff check
+
+# Network tests (downloads codecs from OCI registries)
+uv run pytest --run-network
+```
 
 ## Acknowledgements
 
-Partially supported by NASA-IMPACT VEDA project
+Partially supported by NASA-IMPACT VEDA project.
diff --git a/bench/README.md b/bench/README.md
new file mode 100644
index 0000000..4c3f578
--- /dev/null
+++ b/bench/README.md
@@ -0,0 +1,59 @@
+# bench/
+
+Three benchmarking tools for different aspects of the chonkle stack.
+
+## rust-host/
+
+Standalone Rust crate. Uses `wasmtime-rs 41` typed bindings generated by
+`component::bindgen!`. `Vec<u8>` arguments are lowered via the compiled
+canonical ABI path — no per-element Python interpreter overhead.
+
+```
+cd bench/rust-host
+cargo build --release
+cargo run --release
+```
+
+## python-host/
+
+Minimal Python script. Uses `wasmtime-py 41` raw `Func` call, bypassing the
+chonkle executor entirely. Structurally identical to the Rust binary so the
+comparison isolates the host language, not chonkle overhead.
+
+Uses PEP 723 inline script metadata to declare `wasmtime==41.*` as a
+dependency. Does not depend on the root chonkle project.
+
+```
+uv run bench/python-host/time_abi_raw.py
+```
+
+## chonkle-host/
+
+Drives the chonkle executor directly to investigate per-step `fn()` timing
+across codec types (zlib, predictor2, identity) and data sizes. This is the
+script used to generate the data in `docs/decisions/CANONICAL_ABI_PERF.md`.
+
+Uses PEP 723 inline script metadata with `chonkle` as a local path dependency
+(`path = "../.."`). Requires built codec `.wasm` files.
+
+```
+uv run bench/chonkle-host/time_codec.py
+```
+
+---
+
+## Interpreting the output
+
+`rust-host` and `python-host` print `[TIMING]` lines in the same format:
+
+```
+[TIMING] identity.wasm decode: fn=<s>s  in=<B>B out=<B>B abi_total=<B>B throughput=<MB/s>MB/s  (<label> run <n>/3, <host>)
+```
+
+`throughput` counts both input and output bytes (total ABI traffic). Run-to-run
+variance in the Rust numbers is likely CPU frequency scaling or allocator
+effects, not JIT warm-up — wasmtime compiles the WASM once at load time, before
+the timed loop starts.
+
+See [docs/decisions/CANONICAL_ABI_PERF.md](../docs/decisions/CANONICAL_ABI_PERF.md)
+for the full investigation and measured results.
diff --git a/bench/chonkle-host/time_codec.py b/bench/chonkle-host/time_codec.py
new file mode 100644
index 0000000..7d2fc60
--- /dev/null
+++ b/bench/chonkle-host/time_codec.py
@@ -0,0 +1,207 @@
+# ruff: noqa: T201
+# /// script
+# requires-python = ">=3.13"
+# dependencies = ["chonkle"]
+#
+# [tool.uv.sources]
+# chonkle = { path = "../.." }
+# ///
+"""Investigate fn() call timing vs data size for COG WASM codecs.
+
+Run with:
+    uv run bench/chonkle-host/time_codec.py
+
+Produces per-step timing for multiple input sizes to determine whether
+the 2s per call is fixed overhead or scales with data volume.
+"""
+
+from __future__ import annotations
+
+import logging
+import zlib
+from pathlib import Path
+
+from chonkle.executor import run
+from chonkle.pipeline import prepare
+from chonkle.resolver import Resolver
+
+logging.basicConfig(level=logging.WARNING, format="%(message)s")
+
+REPO_ROOT = Path(__file__).parent.parent.parent
+CODEC_DIR = REPO_ROOT / "codec"
+CHUNKS_DIR = REPO_ROOT / "tests" / "fixtures" / "chunks"
+
+RESOLVER = Resolver(
+    paths={
+        "zlib": CODEC_DIR / "zlib-rs" / "zlib.wasm",
+        "tiff-predictor-2": CODEC_DIR / "tiff-predictor-2-c" / "tiff-predictor-2.wasm",
+        "identity": CODEC_DIR / "identity-c" / "identity.wasm",
+    }
+)
+
+
+def _make_zlib_only_pipeline() -> dict:
+    return {
+        "codec_id": "zlib-only",
+        "direction": "decode",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "constants": {"level": {"type": "int", "value": 9}},
+        "outputs": {"bytes": "zlib.bytes"},
+        "steps": {
+            "zlib": {
+                "codec_id": "zlib",
+                "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+            }
+        },
+    }
+
+
+def _make_pred2_only_pipeline(width: int) -> dict:
+    return {
+        "codec_id": "pred2-only",
+        "direction": "decode",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "constants": {
+            "bytes_per_sample": {"type": "int", "value": 2},
+            "width": {"type": "int", "value": width},
+        },
+        "outputs": {"bytes": "predictor2.bytes"},
+        "steps": {
+            "predictor2": {
+                "codec_id": "tiff-predictor-2",
+                "inputs": {
+                    "bytes": "input.bytes",
+                    "bytes_per_sample": "constant.bytes_per_sample",
+                    "width": "constant.width",
+                },
+            }
+        },
+    }
+
+
+def _make_identity_pipeline() -> dict:
+    return {
+        "codec_id": "identity",
+        "direction": "decode",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "constants": {},
+        "outputs": {"bytes": "identity.bytes"},
+        "steps": {
+            "identity": {
+                "codec_id": "identity",
+                "inputs": {"bytes": "input.bytes"},
+            }
+        },
+    }
+
+
+def run_identity(data: bytes, label: str) -> bytes:
+    print(f"\n--- identity: {label} ({len(data):,} bytes) ---")
+    prepared = prepare(_make_identity_pipeline(), "decode", resolver=RESOLVER)
+    result = run(prepared, {"bytes": data})
+    return result["bytes"]
+
+
+def _make_zlib_encode_pipeline() -> dict:
+    return {
+        "codec_id": "zlib-encode-only",
+        "direction": "encode",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "constants": {"level": {"type": "int", "value": 9}},
+        "outputs": {"bytes": "zlib.bytes"},
+        "steps": {
+            "zlib": {
+                "codec_id": "zlib",
+                "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+            }
+        },
+    }
+
+
+def run_zlib(data: bytes, label: str) -> bytes:
+    print(f"\n--- zlib decode: {label} ({len(data):,} bytes compressed) ---")
+    prepared = prepare(_make_zlib_only_pipeline(), "decode", resolver=RESOLVER)
+    result = run(prepared, {"bytes": data})
+    return result["bytes"]
+
+
+def run_zlib_encode(data: bytes, label: str) -> bytes:
+    print(f"\n--- zlib encode: {label} ({len(data):,} bytes raw) ---")
+    prepared = prepare(_make_zlib_encode_pipeline(), "encode", resolver=RESOLVER)
+    result = run(prepared, {"bytes": data})
+    return result["bytes"]
+
+
+def run_pred2(data: bytes, width: int, label: str) -> bytes:
+    print(f"\n--- predictor2 decode: {label} ({len(data):,} bytes) width={width} ---")
+    prepared = prepare(_make_pred2_only_pipeline(width), "decode", resolver=RESOLVER)
+    result = run(prepared, {"bytes": data})
+    return result["bytes"]
+
+
+def main() -> None:
+    print("=" * 70)
+    print("CODEC TIMING INVESTIGATION")
+    print("Each [TIMING] line shows from_file / store+linker / instantiate / fn")
+    print("=" * 70)
+
+    # --- zlib at various sizes ---
+    tiny_raw = bytes(range(64))
+    small_raw = bytes(range(256)) * 4  # 1 KB
+    medium_raw = bytes(range(256)) * 256  # 64 KB
+    large_raw = bytes(range(256)) * 4096  # 1 MB
+
+    tiny_z = zlib.compress(tiny_raw, level=9)
+    small_z = zlib.compress(small_raw, level=9)
+    medium_z = zlib.compress(medium_raw, level=9)
+    large_z = zlib.compress(large_raw, level=9)
+    real_z = (CHUNKS_DIR / "cog-chunk-0").read_bytes()
+
+    run_zlib(tiny_z, "64 B raw")
+    run_zlib(small_z, "1 KB raw")
+    run_zlib(medium_z, "64 KB raw")
+    run_zlib(large_z, "1 MB raw")
+    run_zlib(real_z, "real COG chunk → 2 MB raw")
+
+    # --- identity at various sizes (ABI isolation: pure memcpy in WASM) ---
+    identity_wasm = CODEC_DIR / "identity-c" / "identity.wasm"
+    if identity_wasm.exists():
+        run_identity(bytes(64), "64 B")
+        run_identity(bytes(1024), "1 KB")
+        run_identity(bytes(65536), "64 KB")
+        run_identity(bytes(1048576), "1 MB")
+        run_identity(bytes(2097152), "2 MB")
+    else:
+        print(f"\n[identity codec not found — skipping: {identity_wasm}]")
+
+    # --- predictor2 at various sizes ---
+    # width=4 means each row is 4 samples → 8 bytes; rows = len/8
+    run_pred2(bytes(8), width=4, label="8 B (1 row)")
+    run_pred2(bytes(1024), width=4, label="1 KB")
+    run_pred2(bytes(65536), width=256, label="64 KB, width=256")
+    run_pred2(bytes(1048576), width=512, label="1 MB, width=512")
+    repeat = 3
+    for i in range(repeat):
+        run_pred2(
+            bytes(2097152),
+            width=1024,
+            label=f"2 MB (real tile size), width=1024 [run {i + 1}/{repeat}]",
+        )
+
+    # --- lowering vs lifting asymmetry (Experiment 4) ---
+    # Same ~2 MB total ABI traffic, but split differs: one is mostly lowering
+    # (large input, tiny output) and the other is mostly lifting (tiny input,
+    # large output).
+    zeros_2mb = bytes(2 * 1024 * 1024)
+    compressed_zeros = zlib.compress(zeros_2mb, level=9)
+    # lowering-heavy: 2 MB in → tiny out
+    run_zlib_encode(zeros_2mb, "2 MB zeros → tiny (lowering heavy)")
+    # lifting-heavy: tiny in → 2 MB out
+    run_zlib(compressed_zeros, "tiny → 2 MB zeros (lifting heavy)")
+
+    print("\n" + "=" * 70)
+    print("DONE")
+
+
+if __name__ == "__main__":
+    main()
diff --git a/bench/python-host/time_abi_raw.py b/bench/python-host/time_abi_raw.py
new file mode 100644
index 0000000..fbe19c4
--- /dev/null
+++ b/bench/python-host/time_abi_raw.py
@@ -0,0 +1,63 @@
+# ruff: noqa: T201
+# /// script
+# requires-python = ">=3.13"
+# dependencies = ["wasmtime==41.*"]
+# ///
+"""Minimal wasmtime-py direct call — bypasses chonkle executor.
+
+Mirrors bench/rust-host/src/main.rs for apples-to-apples comparison.
+Run with:
+    uv run bench/python-host/time_abi_raw.py
+"""
+
+from __future__ import annotations
+
+import time
+from pathlib import Path
+
+import wasmtime
+import wasmtime.component
+
+REPO_ROOT = Path(__file__).parent.parent.parent
+IDENTITY_WASM = REPO_ROOT / "codec" / "identity-c" / "identity.wasm"
+IFACE = "chonkle:codec/transform@0.1.0"
+
+engine = wasmtime.Engine()
+component = wasmtime.component.Component.from_file(engine, str(IDENTITY_WASM))
+
+store = wasmtime.Store(engine)
+store.set_wasi(wasmtime.WasiConfig())
+linker = wasmtime.component.Linker(engine)
+linker.add_wasip2()
+instance = linker.instantiate(store, component)
+
+iface_idx = instance.get_export_index(store, IFACE)
+if iface_idx is None:
+    raise RuntimeError(f"interface {IFACE!r} not found in component")
+fn_idx = instance.get_export_index(store, "decode", iface_idx)
+if fn_idx is None:
+    raise RuntimeError(f"'decode' not found in {IFACE!r}")
+fn = instance.get_func(store, fn_idx)
+if fn is None:
+    raise RuntimeError("'decode' export is not a function")
+
+# Warm up
+fn(store, [("bytes", bytes(64))])
+fn.post_return(store)
+
+for size, label in [(1 << 20, "1 MB"), (2 << 20, "2 MB")]:
+    data = bytes(size)
+    port_map = [("bytes", data)]
+    for run in range(1, 4):
+        t0 = time.perf_counter()
+        result = fn(store, port_map)
+        elapsed = time.perf_counter() - t0
+        fn.post_return(store)
+        out = sum(len(v) for _, v in result)
+        total = size + out
+        throughput = total / elapsed / 1_048_576
+        print(
+            f"[TIMING] identity.wasm decode: fn={elapsed:.3f}s  in={size}B "
+            f"out={out}B abi_total={total}B throughput={throughput:.1f}MB/s  "
+            f"({label} run {run}/3, Python raw)"
+        )
diff --git a/bench/python-host/uv.lock b/bench/python-host/uv.lock
new file mode 100644
index 0000000..b458d33
--- /dev/null
+++ b/bench/python-host/uv.lock
@@ -0,0 +1,33 @@
+version = 1
+revision = 3
+requires-python = ">=3.13"
+
+[[package]]
+name = "python-host-bench"
+version = "0.1.0"
+source = { virtual = "." }
+dependencies = [
+    { name = "wasmtime" },
+]
+
+[package.metadata]
+requires-dist = [{ name = "wasmtime", specifier = "==41.*" }]
+
+[[package]]
+name = "wasmtime"
+version = "41.0.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/be/68/6dc0e7156f883afe0129dd89e4031c8d1163131794ba6ce9e454a09168ad/wasmtime-41.0.0.tar.gz", hash = "sha256:fc2aaacf3ba794eac8baeb739939b2f7903e12d6b78edddc0b7f3ac3a9af6dfc", size = 117354, upload-time = "2026-01-20T18:18:00.565Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/31/f9/f6aef5de536d12652d97cf162f124cbdd642150c7da61ffa7863272cdab7/wasmtime-41.0.0-py3-none-android_26_arm64_v8a.whl", hash = "sha256:f5a6e237b5b94188ef9867926b447f779f540c729c92e4d91cc946f2bee7c282", size = 6837018, upload-time = "2026-01-20T18:17:41.489Z" },
+    { url = "https://files.pythonhosted.org/packages/04/b9/42ec977972b2dcc8c61e3a40644d24d229b41fba151410644e44e35e6eb1/wasmtime-41.0.0-py3-none-android_26_x86_64.whl", hash = "sha256:4a3e33d0d3cf49062eaa231f748f54af991e89e9a795c5ab9d4f0eee85736e4c", size = 7654957, upload-time = "2026-01-20T18:17:43.285Z" },
+    { url = "https://files.pythonhosted.org/packages/18/ca/6cce49b03c35c7fecb4437fd98990c64694a5e0024f9279bef0ddef000f7/wasmtime-41.0.0-py3-none-any.whl", hash = "sha256:5f6721406a6cd186d11f34e6d4991c4d536387b0c577d09a56bd93b8a3cf10c2", size = 6325757, upload-time = "2026-01-20T18:17:44.789Z" },
+    { url = "https://files.pythonhosted.org/packages/a0/16/d91cb80322cc7ae10bfa5db8cea4e0b9bb112f0c100b4486783ab16c1c22/wasmtime-41.0.0-py3-none-macosx_10_13_x86_64.whl", hash = "sha256:2107360212fce33ed2adcfc33b7e75ed7136380a17d3ed598a5bab376dcf9e1b", size = 7471888, upload-time = "2026-01-20T18:17:46.185Z" },
+    { url = "https://files.pythonhosted.org/packages/bc/f0/dcc80973d2ec58a1978b838887ccbd84d56900cf66dec5fb730bec3bd081/wasmtime-41.0.0-py3-none-macosx_11_0_arm64.whl", hash = "sha256:f475df32ce9bfec4f6d0e124a49ca4a89e2ee71ccca460677f5237b1c8ee92ae", size = 6507285, upload-time = "2026-01-20T18:17:48.138Z" },
+    { url = "https://files.pythonhosted.org/packages/bd/df/0867edd9ec26eb2e5eee7674a55f82c23ec27dd1d38d2d401f0e308eb920/wasmtime-41.0.0-py3-none-manylinux1_x86_64.whl", hash = "sha256:ad7e866430313eb2ee07c85811e524344884489d00896f3b2246b65553fe322c", size = 7732024, upload-time = "2026-01-20T18:17:50.207Z" },
+    { url = "https://files.pythonhosted.org/packages/bb/48/b748a2e70478feabc5c876d90e90a39f4aba35378f5ee822f607e8f29c69/wasmtime-41.0.0-py3-none-manylinux2014_aarch64.whl", hash = "sha256:e0ea44584f60dcfa620af82d4fc2589248bcf64a93905b54ac3144242113b48a", size = 6800017, upload-time = "2026-01-20T18:17:51.7Z" },
+    { url = "https://files.pythonhosted.org/packages/14/29/43656c3a464d437d62421de16f2de2db645647bab0a0153deea30bfdade4/wasmtime-41.0.0-py3-none-musllinux_1_2_aarch64.whl", hash = "sha256:1dabb20a2751f01b835095013426a76091bd0bdb36ca9bcfc49c910b78347438", size = 6840763, upload-time = "2026-01-20T18:17:53.125Z" },
+    { url = "https://files.pythonhosted.org/packages/9f/09/4608b65fa35ce5fc1479e138293a1166b4ea817cfa9a79f019ab6d7013d8/wasmtime-41.0.0-py3-none-musllinux_1_2_x86_64.whl", hash = "sha256:d9627dfc5625b4947ea35c819561da358838fe76f65bda8ffe01ce34df8b32b1", size = 7754016, upload-time = "2026-01-20T18:17:55.346Z" },
+    { url = "https://files.pythonhosted.org/packages/f4/9d/236bb367270579e4f628fb7b04fe93541151df7953006f3766607fc667c9/wasmtime-41.0.0-py3-none-win_amd64.whl", hash = "sha256:4f29171d73b71f232b6fe86cba77526fee84139f1590071af5facba401b0c9eb", size = 6325764, upload-time = "2026-01-20T18:17:57.034Z" },
+    { url = "https://files.pythonhosted.org/packages/4a/4a/bba9c0368c377250ab24fd005a7a1e9076121778c1e83b1bcc092ab84f86/wasmtime-41.0.0-py3-none-win_arm64.whl", hash = "sha256:0c4bcaba055e78fc161f497b85f39f1d35d475f0341b1e0259fa0a4b49e223e8", size = 5392238, upload-time = "2026-01-20T18:17:59.052Z" },
+]
diff --git a/bench/rust-host/.gitignore b/bench/rust-host/.gitignore
new file mode 100644
index 0000000..ea8c4bf
--- /dev/null
+++ b/bench/rust-host/.gitignore
@@ -0,0 +1 @@
+/target
diff --git a/bench/rust-host/Cargo.lock b/bench/rust-host/Cargo.lock
new file mode 100644
index 0000000..d142961
--- /dev/null
+++ b/bench/rust-host/Cargo.lock
@@ -0,0 +1,2763 @@
+# This file is automatically @generated by Cargo.
+# It is not intended for manual editing.
+version = 4
+
+[[package]]
+name = "addr2line"
+version = "0.25.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1b5d307320b3181d6d7954e663bd7c774a838b8220fe0593c86d9fb09f498b4b"
+dependencies = [
+ "gimli",
+]
+
+[[package]]
+name = "allocator-api2"
+version = "0.2.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "683d7910e743518b0e34f1186f92494becacb047c7b6bf616c96772180fef923"
+
+[[package]]
+name = "ambient-authority"
+version = "0.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e9d4ee0d472d1cd2e28c97dfa124b3d8d992e10eb0a035f33f5d12e3a177ba3b"
+
+[[package]]
+name = "android_system_properties"
+version = "0.1.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "819e7219dbd41043ac279b19830f2efc897156490d7fd6ea916720117ee66311"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "anyhow"
+version = "1.0.102"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7f202df86484c868dbad7eaa557ef785d5c66295e41b460ef922eca0723b842c"
+
+[[package]]
+name = "arbitrary"
+version = "1.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c3d036a3c4ab069c7b410a2ce876bd74808d2d0888a82667669f8e783a898bf1"
+
+[[package]]
+name = "async-trait"
+version = "0.1.89"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9035ad2d096bed7955a320ee7e2230574d28fd3c3a0f186cbea1ff3c7eed5dbb"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "base64"
+version = "0.22.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "72b3254f16251a8381aa12e40e3c4d2f0199f8c6508fbecb9d91f575e0fbb8c6"
+
+[[package]]
+name = "bitflags"
+version = "2.11.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "843867be96c8daad0d758b57df9392b6d8d271134fce549de6ce169ff98a92af"
+
+[[package]]
+name = "bitmaps"
+version = "2.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "031043d04099746d8db04daf1fa424b2bc8bd69d92b25962dcde24da39ab64a2"
+dependencies = [
+ "typenum",
+]
+
+[[package]]
+name = "block-buffer"
+version = "0.10.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3078c7629b62d3f0439517fa394996acacc5cbc91c5a20d8c658e77abd503a71"
+dependencies = [
+ "generic-array",
+]
+
+[[package]]
+name = "bumpalo"
+version = "3.20.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5d20789868f4b01b2f2caec9f5c4e0213b41e3e5702a50157d699ae31ced2fcb"
+dependencies = [
+ "allocator-api2",
+]
+
+[[package]]
+name = "bytes"
+version = "1.11.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e748733b7cbc798e1434b6ac524f0c1ff2ab456fe201501e6497c8417a4fc33"
+
+[[package]]
+name = "cap-fs-ext"
+version = "3.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d5528f85b1e134ae811704e41ef80930f56e795923f866813255bc342cc20654"
+dependencies = [
+ "cap-primitives",
+ "cap-std",
+ "io-lifetimes",
+ "windows-sys 0.59.0",
+]
+
+[[package]]
+name = "cap-net-ext"
+version = "3.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "20a158160765c6a7d0d8c072a53d772e4cb243f38b04bfcf6b4939cfbe7482e7"
+dependencies = [
+ "cap-primitives",
+ "cap-std",
+ "rustix 1.1.4",
+ "smallvec",
+]
+
+[[package]]
+name = "cap-primitives"
+version = "3.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6cf3aea8a5081171859ef57bc1606b1df6999df4f1110f8eef68b30098d1d3a"
+dependencies = [
+ "ambient-authority",
+ "fs-set-times",
+ "io-extras",
+ "io-lifetimes",
+ "ipnet",
+ "maybe-owned",
+ "rustix 1.1.4",
+ "rustix-linux-procfs",
+ "windows-sys 0.59.0",
+ "winx",
+]
+
+[[package]]
+name = "cap-rand"
+version = "3.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d8144c22e24bbcf26ade86cb6501a0916c46b7e4787abdb0045a467eb1645a1d"
+dependencies = [
+ "ambient-authority",
+ "rand",
+]
+
+[[package]]
+name = "cap-std"
+version = "3.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6dc3090992a735d23219de5c204927163d922f42f575a0189b005c62d37549a"
+dependencies = [
+ "cap-primitives",
+ "io-extras",
+ "io-lifetimes",
+ "rustix 1.1.4",
+]
+
+[[package]]
+name = "cap-time-ext"
+version = "3.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "def102506ce40c11710a9b16e614af0cde8e76ae51b1f48c04b8d79f4b671a80"
+dependencies = [
+ "ambient-authority",
+ "cap-primitives",
+ "iana-time-zone",
+ "once_cell",
+ "rustix 1.1.4",
+ "winx",
+]
+
+[[package]]
+name = "cc"
+version = "1.2.57"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7a0dd1ca384932ff3641c8718a02769f1698e7563dc6974ffd03346116310423"
+dependencies = [
+ "find-msvc-tools",
+ "jobserver",
+ "libc",
+ "shlex",
+]
+
+[[package]]
+name = "cfg-if"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
+
+[[package]]
+name = "cobs"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0fa961b519f0b462e3a3b4a34b64d119eeaca1d59af726fe450bbba07a9fc0a1"
+dependencies = [
+ "thiserror 2.0.18",
+]
+
+[[package]]
+name = "core-foundation-sys"
+version = "0.8.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "773648b94d0e5d620f64f280777445740e61fe701025087ec8b57f45c791888b"
+
+[[package]]
+name = "cpp_demangle"
+version = "0.4.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f2bb79cb74d735044c972aae58ed0aaa9a837e85b01106a54c39e42e97f62253"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
+name = "cpufeatures"
+version = "0.2.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "59ed5838eebb26a2bb2e58f6d5b5316989ae9d08bab10e0e6d103e656d1b0280"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "cranelift-assembler-x64"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "50a04121a197fde2fe896f8e7cac9812fc41ed6ee9c63e1906090f9f497845f6"
+dependencies = [
+ "cranelift-assembler-x64-meta",
+]
+
+[[package]]
+name = "cranelift-assembler-x64-meta"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a09e699a94f477303820fb2167024f091543d6240783a2d3b01a3f21c42bc744"
+dependencies = [
+ "cranelift-srcgen",
+]
+
+[[package]]
+name = "cranelift-bforest"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f07732c662a9755529e332d86f8c5842171f6e98ba4d5976a178043dad838654"
+dependencies = [
+ "cranelift-entity",
+]
+
+[[package]]
+name = "cranelift-bitset"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "18391da761cf362a06def7a7cf11474d79e55801dd34c2e9ba105b33dc0aef88"
+dependencies = [
+ "serde",
+ "serde_derive",
+]
+
+[[package]]
+name = "cranelift-codegen"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b3a09b3042c69810d255aef59ddc3b3e4c0644d1d90ecfd6e3837798cc88a3c"
+dependencies = [
+ "bumpalo",
+ "cranelift-assembler-x64",
+ "cranelift-bforest",
+ "cranelift-bitset",
+ "cranelift-codegen-meta",
+ "cranelift-codegen-shared",
+ "cranelift-control",
+ "cranelift-entity",
+ "cranelift-isle",
+ "gimli",
+ "hashbrown 0.15.5",
+ "log",
+ "pulley-interpreter",
+ "regalloc2",
+ "rustc-hash",
+ "serde",
+ "smallvec",
+ "target-lexicon",
+ "wasmtime-internal-math",
+]
+
+[[package]]
+name = "cranelift-codegen-meta"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "75817926ec812241889208d1b190cadb7fedded4592a4bb01b8524babb9e4849"
+dependencies = [
+ "cranelift-assembler-x64-meta",
+ "cranelift-codegen-shared",
+ "cranelift-srcgen",
+ "heck",
+ "pulley-interpreter",
+]
+
+[[package]]
+name = "cranelift-codegen-shared"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "859158f87a59476476eda3884d883c32e08a143cf3d315095533b362a3250a63"
+
+[[package]]
+name = "cranelift-control"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "03b65a9aec442d715cbf54d14548b8f395476c09cef7abe03e104a378291ab88"
+dependencies = [
+ "arbitrary",
+]
+
+[[package]]
+name = "cranelift-entity"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8334c99a7e86060c24028732efd23bac84585770dcb752329c69f135d64f2fc1"
+dependencies = [
+ "cranelift-bitset",
+ "serde",
+ "serde_derive",
+]
+
+[[package]]
+name = "cranelift-frontend"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "43ac6c095aa5b3e845d7ca3461e67e2b65249eb5401477a5ff9100369b745111"
+dependencies = [
+ "cranelift-codegen",
+ "log",
+ "smallvec",
+ "target-lexicon",
+]
+
+[[package]]
+name = "cranelift-isle"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "69d3d992870ed4f0f2e82e2175275cb3a123a46e9660c6558c46417b822c91fa"
+
+[[package]]
+name = "cranelift-native"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ee32e36beaf80f309edb535274cfe0349e1c5cf5799ba2d9f42e828285c6b52e"
+dependencies = [
+ "cranelift-codegen",
+ "libc",
+ "target-lexicon",
+]
+
+[[package]]
+name = "cranelift-srcgen"
+version = "0.128.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "903adeaf4938e60209a97b53a2e4326cd2d356aab9764a1934630204bae381c9"
+
+[[package]]
+name = "crc32fast"
+version = "1.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9481c1c90cbf2ac953f07c8d4a58aa3945c425b7185c9154d67a65e4230da511"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
+name = "crossbeam-deque"
+version = "0.8.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9dd111b7b7f7d55b72c0a6ae361660ee5853c9af73f70c3c2ef6858b950e2e51"
+dependencies = [
+ "crossbeam-epoch",
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "crossbeam-epoch"
+version = "0.9.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5b82ac4a3c2ca9c3460964f020e1402edd5753411d7737aa39c3714ad1b5420e"
+dependencies = [
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "crossbeam-utils"
+version = "0.8.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d0a5c400df2834b80a4c3327b3aad3a4c4cd4de0629063962b03235697506a28"
+
+[[package]]
+name = "crypto-common"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "78c8292055d1c1df0cce5d180393dc8cce0abec0a7102adb6c7b1eef6016d60a"
+dependencies = [
+ "generic-array",
+ "typenum",
+]
+
+[[package]]
+name = "debugid"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bef552e6f588e446098f6ba40d89ac146c8c7b64aade83c051ee00bb5d2bc18d"
+dependencies = [
+ "uuid",
+]
+
+[[package]]
+name = "digest"
+version = "0.10.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ed9a281f7bc9b7576e61468ba615a66a5c8cfdff42420a70aa82701a3b1e292"
+dependencies = [
+ "block-buffer",
+ "crypto-common",
+]
+
+[[package]]
+name = "directories-next"
+version = "2.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "339ee130d97a610ea5a5872d2bbb130fdf68884ff09d3028b81bec8a1ac23bbc"
+dependencies = [
+ "cfg-if",
+ "dirs-sys-next",
+]
+
+[[package]]
+name = "dirs-sys-next"
+version = "0.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4ebda144c4fe02d1f7ea1a7d9641b6fc6b580adcfa024ae48797ecdeb6825b4d"
+dependencies = [
+ "libc",
+ "redox_users",
+ "winapi",
+]
+
+[[package]]
+name = "displaydoc"
+version = "0.2.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "97369cbbc041bc366949bc74d34658d6cda5621039731c6310521892a3a20ae0"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "either"
+version = "1.15.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "48c757948c5ede0e46177b7add2e67155f70e33c07fea8284df6576da70b3719"
+
+[[package]]
+name = "embedded-io"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ef1a6892d9eef45c8fa6b9e0086428a2cca8491aca8f787c534a3d6d0bcb3ced"
+
+[[package]]
+name = "embedded-io"
+version = "0.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "edd0f118536f44f5ccd48bcb8b111bdc3de888b58c74639dfb034a357d0f206d"
+
+[[package]]
+name = "encoding_rs"
+version = "0.8.35"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "75030f3c4f45dafd7586dd6780965a8c7e8e285a5ecb86713e63a79c5b2766f3"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
+name = "equivalent"
+version = "1.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "877a4ace8713b0bcf2a4e7eec82529c029f1d0619886d18145fea96c3ffe5c0f"
+
+[[package]]
+name = "errno"
+version = "0.3.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
+dependencies = [
+ "libc",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "fallible-iterator"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2acce4a10f12dc2fb14a218589d4f1f62ef011b2d0cc4b3cb1bba8e94da14649"
+
+[[package]]
+name = "fastrand"
+version = "2.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "37909eebbb50d72f9059c3b6d82c0463f2ff062c9e95845c43a6c9c0355411be"
+
+[[package]]
+name = "fd-lock"
+version = "4.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0ce92ff622d6dadf7349484f42c93271a0d49b7cc4d466a936405bacbe10aa78"
+dependencies = [
+ "cfg-if",
+ "rustix 1.1.4",
+ "windows-sys 0.59.0",
+]
+
+[[package]]
+name = "find-msvc-tools"
+version = "0.1.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5baebc0774151f905a1a2cc41989300b1e6fbb29aff0ceffa1064fdd3088d582"
+
+[[package]]
+name = "fixedbitset"
+version = "0.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0ce7134b9999ecaf8bcd65542e436736ef32ddca1b3e06094cb6ec5755203b80"
+
+[[package]]
+name = "foldhash"
+version = "0.1.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d9c4f5dac5e15c24eb999c26181a6ca40b39fe946cbe4c263c7209467bc83af2"
+
+[[package]]
+name = "form_urlencoded"
+version = "1.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cb4cb245038516f5f85277875cdaa4f7d2c9a0fa0468de06ed190163b1581fcf"
+dependencies = [
+ "percent-encoding",
+]
+
+[[package]]
+name = "fs-set-times"
+version = "0.20.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "94e7099f6313ecacbe1256e8ff9d617b75d1bcb16a6fddef94866d225a01a14a"
+dependencies = [
+ "io-lifetimes",
+ "rustix 1.1.4",
+ "windows-sys 0.59.0",
+]
+
+[[package]]
+name = "futures"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8b147ee9d1f6d097cef9ce628cd2ee62288d963e16fb287bd9286455b241382d"
+dependencies = [
+ "futures-channel",
+ "futures-core",
+ "futures-io",
+ "futures-sink",
+ "futures-task",
+ "futures-util",
+]
+
+[[package]]
+name = "futures-channel"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "07bbe89c50d7a535e539b8c17bc0b49bdb77747034daa8087407d655f3f7cc1d"
+dependencies = [
+ "futures-core",
+ "futures-sink",
+]
+
+[[package]]
+name = "futures-core"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7e3450815272ef58cec6d564423f6e755e25379b217b0bc688e295ba24df6b1d"
+
+[[package]]
+name = "futures-io"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cecba35d7ad927e23624b22ad55235f2239cfa44fd10428eecbeba6d6a717718"
+
+[[package]]
+name = "futures-sink"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c39754e157331b013978ec91992bde1ac089843443c49cbc7f46150b0fad0893"
+
+[[package]]
+name = "futures-task"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "037711b3d59c33004d3856fbdc83b99d4ff37a24768fa1be9ce3538a1cde4393"
+
+[[package]]
+name = "futures-util"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "389ca41296e6190b48053de0321d02a77f32f8a5d2461dd38762c0593805c6d6"
+dependencies = [
+ "futures-channel",
+ "futures-core",
+ "futures-io",
+ "futures-sink",
+ "futures-task",
+ "memchr",
+ "pin-project-lite",
+ "slab",
+]
+
+[[package]]
+name = "fxprof-processed-profile"
+version = "0.8.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "25234f20a3ec0a962a61770cfe39ecf03cb529a6e474ad8cff025ed497eda557"
+dependencies = [
+ "bitflags",
+ "debugid",
+ "rustc-hash",
+ "serde",
+ "serde_derive",
+ "serde_json",
+]
+
+[[package]]
+name = "generic-array"
+version = "0.14.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "85649ca51fd72272d7821adaf274ad91c288277713d9c18820d8499a7ff69e9a"
+dependencies = [
+ "typenum",
+ "version_check",
+]
+
+[[package]]
+name = "getrandom"
+version = "0.2.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ff2abc00be7fca6ebc474524697ae276ad847ad0a6b3faa4bcb027e9a4614ad0"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "wasi",
+]
+
+[[package]]
+name = "getrandom"
+version = "0.3.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "899def5c37c4fd7b2664648c28120ecec138e4d395b459e5ca34f9cce2dd77fd"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "r-efi 5.3.0",
+ "wasip2",
+]
+
+[[package]]
+name = "getrandom"
+version = "0.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0de51e6874e94e7bf76d726fc5d13ba782deca734ff60d5bb2fb2607c7406555"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "r-efi 6.0.0",
+ "wasip2",
+ "wasip3",
+]
+
+[[package]]
+name = "gimli"
+version = "0.32.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e629b9b98ef3dd8afe6ca2bd0f89306cec16d43d907889945bc5d6687f2f13c7"
+dependencies = [
+ "fallible-iterator",
+ "indexmap",
+ "stable_deref_trait",
+]
+
+[[package]]
+name = "hashbrown"
+version = "0.15.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9229cfe53dfd69f0609a49f65461bd93001ea1ef889cd5529dd176593f5338a1"
+dependencies = [
+ "foldhash",
+ "serde",
+]
+
+[[package]]
+name = "hashbrown"
+version = "0.16.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "841d1cc9bed7f9236f321df977030373f4a4163ae1a7dbfe1a51a2c1a51d9100"
+
+[[package]]
+name = "heck"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea"
+
+[[package]]
+name = "iana-time-zone"
+version = "0.1.65"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e31bc9ad994ba00e440a8aa5c9ef0ec67d5cb5e5cb0cc7f8b744a35b389cc470"
+dependencies = [
+ "android_system_properties",
+ "core-foundation-sys",
+ "iana-time-zone-haiku",
+ "js-sys",
+ "log",
+ "wasm-bindgen",
+ "windows-core",
+]
+
+[[package]]
+name = "iana-time-zone-haiku"
+version = "0.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f31827a206f56af32e590ba56d5d2d085f558508192593743f16b2306495269f"
+dependencies = [
+ "cc",
+]
+
+[[package]]
+name = "icu_collections"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4c6b649701667bbe825c3b7e6388cb521c23d88644678e83c0c4d0a621a34b43"
+dependencies = [
+ "displaydoc",
+ "potential_utf",
+ "yoke",
+ "zerofrom",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_locale_core"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "edba7861004dd3714265b4db54a3c390e880ab658fec5f7db895fae2046b5bb6"
+dependencies = [
+ "displaydoc",
+ "litemap",
+ "tinystr",
+ "writeable",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_normalizer"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5f6c8828b67bf8908d82127b2054ea1b4427ff0230ee9141c54251934ab1b599"
+dependencies = [
+ "icu_collections",
+ "icu_normalizer_data",
+ "icu_properties",
+ "icu_provider",
+ "smallvec",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_normalizer_data"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7aedcccd01fc5fe81e6b489c15b247b8b0690feb23304303a9e560f37efc560a"
+
+[[package]]
+name = "icu_properties"
+version = "2.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "020bfc02fe870ec3a66d93e677ccca0562506e5872c650f893269e08615d74ec"
+dependencies = [
+ "icu_collections",
+ "icu_locale_core",
+ "icu_properties_data",
+ "icu_provider",
+ "zerotrie",
+ "zerovec",
+]
+
+[[package]]
+name = "icu_properties_data"
+version = "2.1.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "616c294cf8d725c6afcd8f55abc17c56464ef6211f9ed59cccffe534129c77af"
+
+[[package]]
+name = "icu_provider"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "85962cf0ce02e1e0a629cc34e7ca3e373ce20dda4c4d7294bbd0bf1fdb59e614"
+dependencies = [
+ "displaydoc",
+ "icu_locale_core",
+ "writeable",
+ "yoke",
+ "zerofrom",
+ "zerotrie",
+ "zerovec",
+]
+
+[[package]]
+name = "id-arena"
+version = "2.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3d3067d79b975e8844ca9eb072e16b31c3c1c36928edf9c6789548c524d0d954"
+
+[[package]]
+name = "idna"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3b0875f23caa03898994f6ddc501886a45c7d3d62d04d2d90788d47be1b1e4de"
+dependencies = [
+ "idna_adapter",
+ "smallvec",
+ "utf8_iter",
+]
+
+[[package]]
+name = "idna_adapter"
+version = "1.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3acae9609540aa318d1bc588455225fb2085b9ed0c4f6bd0d9d5bcd86f1a0344"
+dependencies = [
+ "icu_normalizer",
+ "icu_properties",
+]
+
+[[package]]
+name = "im-rc"
+version = "15.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "af1955a75fa080c677d3972822ec4bad316169ab1cfc6c257a942c2265dbe5fe"
+dependencies = [
+ "bitmaps",
+ "rand_core",
+ "rand_xoshiro",
+ "sized-chunks",
+ "typenum",
+ "version_check",
+]
+
+[[package]]
+name = "indexmap"
+version = "2.13.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7714e70437a7dc3ac8eb7e6f8df75fd8eb422675fc7678aff7364301092b1017"
+dependencies = [
+ "equivalent",
+ "hashbrown 0.16.1",
+ "serde",
+ "serde_core",
+]
+
+[[package]]
+name = "io-extras"
+version = "0.18.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2285ddfe3054097ef4b2fe909ef8c3bcd1ea52a8f0d274416caebeef39f04a65"
+dependencies = [
+ "io-lifetimes",
+ "windows-sys 0.59.0",
+]
+
+[[package]]
+name = "io-lifetimes"
+version = "2.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "06432fb54d3be7964ecd3649233cddf80db2832f47fec34c01f65b3d9d774983"
+
+[[package]]
+name = "ipnet"
+version = "2.12.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d98f6fed1fde3f8c21bc40a1abb88dd75e67924f9cffc3ef95607bad8017f8e2"
+
+[[package]]
+name = "itertools"
+version = "0.14.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2b192c782037fadd9cfa75548310488aabdbf3d2da73885b31bd0abd03351285"
+dependencies = [
+ "either",
+]
+
+[[package]]
+name = "itoa"
+version = "1.0.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "92ecc6618181def0457392ccd0ee51198e065e016d1d527a7ac1b6dc7c1f09d2"
+
+[[package]]
+name = "ittapi"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6b996fe614c41395cdaedf3cf408a9534851090959d90d54a535f675550b64b1"
+dependencies = [
+ "anyhow",
+ "ittapi-sys",
+ "log",
+]
+
+[[package]]
+name = "ittapi-sys"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "52f5385394064fa2c886205dba02598013ce83d3e92d33dbdc0c52fe0e7bf4fc"
+dependencies = [
+ "cc",
+]
+
+[[package]]
+name = "jobserver"
+version = "0.1.34"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9afb3de4395d6b3e67a780b6de64b51c978ecf11cb9a462c66be7d4ca9039d33"
+dependencies = [
+ "getrandom 0.3.4",
+ "libc",
+]
+
+[[package]]
+name = "js-sys"
+version = "0.3.91"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b49715b7073f385ba4bc528e5747d02e66cb39c6146efb66b781f131f0fb399c"
+dependencies = [
+ "once_cell",
+ "wasm-bindgen",
+]
+
+[[package]]
+name = "leb128"
+version = "0.2.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"
+
+[[package]]
+name = "leb128fmt"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "09edd9e8b54e49e587e4f6295a7d29c3ea94d469cb40ab8ca70b288248a81db2"
+
+[[package]]
+name = "libc"
+version = "0.2.183"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b5b646652bf6661599e1da8901b3b9522896f01e736bad5f723fe7a3a27f899d"
+
+[[package]]
+name = "libm"
+version = "0.2.16"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6d2cec3eae94f9f509c767b45932f1ada8350c4bdb85af2fcab4a3c14807981"
+
+[[package]]
+name = "libredox"
+version = "0.1.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1744e39d1d6a9948f4f388969627434e31128196de472883b39f148769bfe30a"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "linux-raw-sys"
+version = "0.4.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d26c52dbd32dccf2d10cac7725f8eae5296885fb5703b261f7d0a0739ec807ab"
+
+[[package]]
+name = "linux-raw-sys"
+version = "0.12.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "32a66949e030da00e8c7d4434b251670a91556f4144941d37452769c25d58a53"
+
+[[package]]
+name = "litemap"
+version = "0.8.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6373607a59f0be73a39b6fe456b8192fcc3585f602af20751600e974dd455e77"
+
+[[package]]
+name = "log"
+version = "0.4.29"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5e5032e24019045c762d3c0f28f5b6b8bbf38563a65908389bf7978758920897"
+
+[[package]]
+name = "mach2"
+version = "0.4.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d640282b302c0bb0a2a8e0233ead9035e3bed871f0b7e81fe4a1ec829765db44"
+dependencies = [
+ "libc",
+]
+
+[[package]]
+name = "maybe-owned"
+version = "0.3.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4facc753ae494aeb6e3c22f839b158aebd4f9270f55cd3c79906c45476c47ab4"
+
+[[package]]
+name = "memchr"
+version = "2.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f8ca58f447f06ed17d5fc4043ce1b10dd205e060fb3ce5b979b8ed8e59ff3f79"
+
+[[package]]
+name = "memfd"
+version = "0.6.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ad38eb12aea514a0466ea40a80fd8cc83637065948eb4a426e4aa46261175227"
+dependencies = [
+ "rustix 1.1.4",
+]
+
+[[package]]
+name = "mio"
+version = "1.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a69bcab0ad47271a0234d9422b131806bf3968021e5dc9328caf2d4cd58557fc"
+dependencies = [
+ "libc",
+ "wasi",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "object"
+version = "0.37.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ff76201f031d8863c38aa7f905eca4f53abbfa15f609db4277d44cd8938f33fe"
+dependencies = [
+ "crc32fast",
+ "hashbrown 0.15.5",
+ "indexmap",
+ "memchr",
+]
+
+[[package]]
+name = "once_cell"
+version = "1.21.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9f7c3e4beb33f85d45ae3e3a1792185706c8e16d043238c593331cc7cd313b50"
+
+[[package]]
+name = "percent-encoding"
+version = "2.3.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9b4f627cb1b25917193a259e49bdad08f671f8d9708acfd5fe0a8c1455d87220"
+
+[[package]]
+name = "petgraph"
+version = "0.6.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b4c5cc86750666a3ed20bdaf5ca2a0344f9c67674cae0515bec2da16fbaa47db"
+dependencies = [
+ "fixedbitset",
+ "indexmap",
+]
+
+[[package]]
+name = "pin-project-lite"
+version = "0.2.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a89322df9ebe1c1578d689c92318e070967d1042b512afbe49518723f4e6d5cd"
+
+[[package]]
+name = "pkg-config"
+version = "0.3.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7edddbd0b52d732b21ad9a5fab5c704c14cd949e5e9a1ec5929a24fded1b904c"
+
+[[package]]
+name = "postcard"
+version = "1.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6764c3b5dd454e283a30e6dfe78e9b31096d9e32036b5d1eaac7a6119ccb9a24"
+dependencies = [
+ "cobs",
+ "embedded-io 0.4.0",
+ "embedded-io 0.6.1",
+ "serde",
+]
+
+[[package]]
+name = "potential_utf"
+version = "0.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b73949432f5e2a09657003c25bca5e19a0e9c84f8058ca374f49e0ebe605af77"
+dependencies = [
+ "zerovec",
+]
+
+[[package]]
+name = "ppv-lite86"
+version = "0.2.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "85eae3c4ed2f50dcfe72643da4befc30deadb458a9b590d720cde2f2b1e97da9"
+dependencies = [
+ "zerocopy",
+]
+
+[[package]]
+name = "prettyplease"
+version = "0.2.37"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "479ca8adacdd7ce8f1fb39ce9ecccbfe93a3f1344b3d0d97f20bc0196208f62b"
+dependencies = [
+ "proc-macro2",
+ "syn",
+]
+
+[[package]]
+name = "proc-macro2"
+version = "1.0.106"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8fd00f0bb2e90d81d1044c2b32617f68fcb9fa3bb7640c23e9c748e53fb30934"
+dependencies = [
+ "unicode-ident",
+]
+
+[[package]]
+name = "pulley-interpreter"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e9812652c1feb63cf39f8780cecac154a32b22b3665806c733cd4072547233a4"
+dependencies = [
+ "cranelift-bitset",
+ "log",
+ "pulley-macros",
+ "wasmtime-internal-math",
+]
+
+[[package]]
+name = "pulley-macros"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "56000349b6896e3d44286eb9c330891237f40b27fd43c1ccc84547d0b463cb40"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "quote"
+version = "1.0.45"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "41f2619966050689382d2b44f664f4bc593e129785a36d6ee376ddf37259b924"
+dependencies = [
+ "proc-macro2",
+]
+
+[[package]]
+name = "r-efi"
+version = "5.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "69cdb34c158ceb288df11e18b4bd39de994f6657d83847bdffdbd7f346754b0f"
+
+[[package]]
+name = "r-efi"
+version = "6.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f8dcc9c7d52a811697d2151c701e0d08956f92b0e24136cf4cf27b57a6a0d9bf"
+
+[[package]]
+name = "rand"
+version = "0.8.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "34af8d1a0e25924bc5b7c43c079c942339d8f0a8b57c39049bef581b46327404"
+dependencies = [
+ "libc",
+ "rand_chacha",
+ "rand_core",
+]
+
+[[package]]
+name = "rand_chacha"
+version = "0.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e6c10a63a0fa32252be49d21e7709d4d4baf8d231c2dbce1eaa8141b9b127d88"
+dependencies = [
+ "ppv-lite86",
+ "rand_core",
+]
+
+[[package]]
+name = "rand_core"
+version = "0.6.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ec0be4795e2f6a28069bec0b5ff3e2ac9bafc99e6a9a7dc3547996c5c816922c"
+dependencies = [
+ "getrandom 0.2.17",
+]
+
+[[package]]
+name = "rand_xoshiro"
+version = "0.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6f97cdb2a36ed4183de61b2f824cc45c9f1037f28afe0a322e9fff4c108b5aaa"
+dependencies = [
+ "rand_core",
+]
+
+[[package]]
+name = "rayon"
+version = "1.11.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "368f01d005bf8fd9b1206fb6fa653e6c4a81ceb1466406b81792d87c5677a58f"
+dependencies = [
+ "either",
+ "rayon-core",
+]
+
+[[package]]
+name = "rayon-core"
+version = "1.13.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "22e18b0f0062d30d4230b2e85ff77fdfe4326feb054b9783a3460d8435c8ab91"
+dependencies = [
+ "crossbeam-deque",
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "redox_users"
+version = "0.4.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ba009ff324d1fc1b900bd1fdb31564febe58a8ccc8a6fdbb93b543d33b13ca43"
+dependencies = [
+ "getrandom 0.2.17",
+ "libredox",
+ "thiserror 1.0.69",
+]
+
+[[package]]
+name = "regalloc2"
+version = "0.13.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "08effbc1fa53aaebff69521a5c05640523fab037b34a4a2c109506bc938246fa"
+dependencies = [
+ "allocator-api2",
+ "bumpalo",
+ "hashbrown 0.15.5",
+ "log",
+ "rustc-hash",
+ "smallvec",
+]
+
+[[package]]
+name = "rust-host"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+ "wasmtime",
+ "wasmtime-wasi",
+]
+
+[[package]]
+name = "rustc-demangle"
+version = "0.1.27"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b50b8869d9fc858ce7266cce0194bd74df58b9d0e3f6df3a9fc8eb470d95c09d"
+
+[[package]]
+name = "rustc-hash"
+version = "2.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d"
+
+[[package]]
+name = "rustix"
+version = "0.38.44"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fdb5bc1ae2baa591800df16c9ca78619bf65c0488b41b96ccec5d11220d8c154"
+dependencies = [
+ "bitflags",
+ "errno",
+ "libc",
+ "linux-raw-sys 0.4.15",
+ "windows-sys 0.59.0",
+]
+
+[[package]]
+name = "rustix"
+version = "1.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6fe4565b9518b83ef4f91bb47ce29620ca828bd32cb7e408f0062e9930ba190"
+dependencies = [
+ "bitflags",
+ "errno",
+ "libc",
+ "linux-raw-sys 0.12.1",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "rustix-linux-procfs"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2fc84bf7e9aa16c4f2c758f27412dc9841341e16aa682d9c7ac308fe3ee12056"
+dependencies = [
+ "once_cell",
+ "rustix 1.1.4",
+]
+
+[[package]]
+name = "rustversion"
+version = "1.0.22"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b39cdef0fa800fc44525c84ccb54a029961a8215f9619753635a9c0d2538d46d"
+
+[[package]]
+name = "ryu"
+version = "1.0.23"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9774ba4a74de5f7b1c1451ed6cd5285a32eddb5cccb8cc655a4e50009e06477f"
+
+[[package]]
+name = "semver"
+version = "1.0.27"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d767eb0aabc880b29956c35734170f26ed551a859dbd361d140cdbeca61ab1e2"
+dependencies = [
+ "serde",
+ "serde_core",
+]
+
+[[package]]
+name = "serde"
+version = "1.0.228"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9a8e94ea7f378bd32cbbd37198a4a91436180c5bb472411e48b5ec2e2124ae9e"
+dependencies = [
+ "serde_core",
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_core"
+version = "1.0.228"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "41d385c7d4ca58e59fc732af25c3983b67ac852c1a25000afe1175de458b67ad"
+dependencies = [
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_derive"
+version = "1.0.228"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d540f220d3187173da220f885ab66608367b6574e925011a9353e4badda91d79"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "serde_json"
+version = "1.0.149"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "83fc039473c5595ace860d8c4fafa220ff474b3fc6bfdb4293327f1a37e94d86"
+dependencies = [
+ "itoa",
+ "memchr",
+ "serde",
+ "serde_core",
+ "zmij",
+]
+
+[[package]]
+name = "serde_spanned"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f8bbf91e5a4d6315eee45e704372590b30e260ee83af6639d64557f51b067776"
+dependencies = [
+ "serde_core",
+]
+
+[[package]]
+name = "serde_yaml"
+version = "0.9.34+deprecated"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6a8b1a1a2ebf674015cc02edccce75287f1a0130d394307b36743c2f5d504b47"
+dependencies = [
+ "indexmap",
+ "itoa",
+ "ryu",
+ "serde",
+ "unsafe-libyaml",
+]
+
+[[package]]
+name = "sha2"
+version = "0.10.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7507d819769d01a365ab707794a4084392c824f54a7a6a7862f8c3d0892b283"
+dependencies = [
+ "cfg-if",
+ "cpufeatures",
+ "digest",
+]
+
+[[package]]
+name = "shlex"
+version = "1.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0fda2ff0d084019ba4d7c6f371c95d8fd75ce3524c3cb8fb653a3023f6323e64"
+
+[[package]]
+name = "sized-chunks"
+version = "0.6.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "16d69225bde7a69b235da73377861095455d298f2b970996eec25ddbb42b3d1e"
+dependencies = [
+ "bitmaps",
+ "typenum",
+]
+
+[[package]]
+name = "slab"
+version = "0.4.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0c790de23124f9ab44544d7ac05d60440adc586479ce501c1d6d7da3cd8c9cf5"
+
+[[package]]
+name = "smallvec"
+version = "1.15.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "67b1b7a3b5fe4f1376887184045fcf45c69e92af734b7aaddc05fb777b6fbd03"
+dependencies = [
+ "serde",
+]
+
+[[package]]
+name = "socket2"
+version = "0.6.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3a766e1110788c36f4fa1c2b71b387a7815aa65f88ce0229841826633d93723e"
+dependencies = [
+ "libc",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "stable_deref_trait"
+version = "1.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6ce2be8dc25455e1f91df71bfa12ad37d7af1092ae736f3a6cd0e37bc7810596"
+
+[[package]]
+name = "syn"
+version = "2.0.117"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e665b8803e7b1d2a727f4023456bbbbe74da67099c585258af0ad9c5013b9b99"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
+]
+
+[[package]]
+name = "synstructure"
+version = "0.13.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "728a70f3dbaf5bab7f0c4b1ac8d7ae5ea60a4b5549c8a5914361c99147a709d2"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "system-interface"
+version = "0.27.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cc4592f674ce18521c2a81483873a49596655b179f71c5e05d10c1fe66c78745"
+dependencies = [
+ "bitflags",
+ "cap-fs-ext",
+ "cap-std",
+ "fd-lock",
+ "io-lifetimes",
+ "rustix 0.38.44",
+ "windows-sys 0.59.0",
+ "winx",
+]
+
+[[package]]
+name = "target-lexicon"
+version = "0.13.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "adb6935a6f5c20170eeceb1a3835a49e12e19d792f6dd344ccc76a985ca5a6ca"
+
+[[package]]
+name = "tempfile"
+version = "3.27.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "32497e9a4c7b38532efcdebeef879707aa9f794296a4f0244f6f69e9bc8574bd"
+dependencies = [
+ "fastrand",
+ "getrandom 0.4.2",
+ "once_cell",
+ "rustix 1.1.4",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "termcolor"
+version = "1.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "06794f8f6c5c898b3275aebefa6b8a1cb24cd2c6c79397ab15774837a0bc5755"
+dependencies = [
+ "winapi-util",
+]
+
+[[package]]
+name = "thiserror"
+version = "1.0.69"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6aaf5339b578ea85b50e080feb250a3e8ae8cfcdff9a461c9ec2904bc923f52"
+dependencies = [
+ "thiserror-impl 1.0.69",
+]
+
+[[package]]
+name = "thiserror"
+version = "2.0.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4288b5bcbc7920c07a1149a35cf9590a2aa808e0bc1eafaade0b80947865fbc4"
+dependencies = [
+ "thiserror-impl 2.0.18",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "1.0.69"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4fee6c4efc90059e10f81e6d42c60a18f76588c3d74cb83a0b242a2b6c7504c1"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "2.0.18"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ebc4ee7f67670e9b64d05fa4253e753e016c6c95ff35b89b7941d6b856dec1d5"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "tinystr"
+version = "0.8.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "42d3e9c45c09de15d06dd8acf5f4e0e399e85927b7f00711024eb7ae10fa4869"
+dependencies = [
+ "displaydoc",
+ "zerovec",
+]
+
+[[package]]
+name = "tokio"
+version = "1.50.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "27ad5e34374e03cfffefc301becb44e9dc3c17584f414349ebe29ed26661822d"
+dependencies = [
+ "bytes",
+ "libc",
+ "mio",
+ "pin-project-lite",
+ "socket2",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "toml"
+version = "0.9.12+spec-1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf92845e79fc2e2def6a5d828f0801e29a2f8acc037becc5ab08595c7d5e9863"
+dependencies = [
+ "indexmap",
+ "serde_core",
+ "serde_spanned",
+ "toml_datetime",
+ "toml_parser",
+ "toml_writer",
+ "winnow",
+]
+
+[[package]]
+name = "toml_datetime"
+version = "0.7.5+spec-1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "92e1cfed4a3038bc5a127e35a2d360f145e1f4b971b551a2ba5fd7aedf7e1347"
+dependencies = [
+ "serde_core",
+]
+
+[[package]]
+name = "toml_parser"
+version = "1.0.9+spec-1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "702d4415e08923e7e1ef96cd5727c0dfed80b4d2fa25db9647fe5eb6f7c5a4c4"
+dependencies = [
+ "winnow",
+]
+
+[[package]]
+name = "toml_writer"
+version = "1.0.6+spec-1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ab16f14aed21ee8bfd8ec22513f7287cd4a91aa92e44edfe2c17ddd004e92607"
+
+[[package]]
+name = "tracing"
+version = "0.1.44"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "63e71662fa4b2a2c3a26f570f037eb95bb1f85397f3cd8076caed2f026a6d100"
+dependencies = [
+ "pin-project-lite",
+ "tracing-attributes",
+ "tracing-core",
+]
+
+[[package]]
+name = "tracing-attributes"
+version = "0.1.31"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7490cfa5ec963746568740651ac6781f701c9c5ea257c58e057f3ba8cf69e8da"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "tracing-core"
+version = "0.1.36"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "db97caf9d906fbde555dd62fa95ddba9eecfd14cb388e4f491a66d74cd5fb79a"
+dependencies = [
+ "once_cell",
+]
+
+[[package]]
+name = "typenum"
+version = "1.19.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "562d481066bde0658276a35467c4af00bdc6ee726305698a55b86e61d7ad82bb"
+
+[[package]]
+name = "unicode-ident"
+version = "1.0.24"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e6e4313cd5fcd3dad5cafa179702e2b244f760991f45397d14d4ebf38247da75"
+
+[[package]]
+name = "unicode-width"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b4ac048d71ede7ee76d585517add45da530660ef4390e49b098733c6e897f254"
+
+[[package]]
+name = "unicode-xid"
+version = "0.2.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ebc1c04c71510c7f702b52b7c350734c9ff1295c464a03335b00bb84fc54f853"
+
+[[package]]
+name = "unsafe-libyaml"
+version = "0.2.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "673aac59facbab8a9007c7f6108d11f63b603f7cabff99fabf650fea5c32b861"
+
+[[package]]
+name = "url"
+version = "2.5.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ff67a8a4397373c3ef660812acab3268222035010ab8680ec4215f38ba3d0eed"
+dependencies = [
+ "form_urlencoded",
+ "idna",
+ "percent-encoding",
+ "serde",
+]
+
+[[package]]
+name = "utf8_iter"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b6c140620e7ffbb22c2dee59cafe6084a59b5ffc27a8859a5f0d494b5d52b6be"
+
+[[package]]
+name = "uuid"
+version = "1.22.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a68d3c8f01c0cfa54a75291d83601161799e4a89a39e0929f4b0354d88757a37"
+dependencies = [
+ "js-sys",
+ "wasm-bindgen",
+]
+
+[[package]]
+name = "version_check"
+version = "0.9.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b928f33d975fc6ad9f86c8f283853ad26bdd5b10b7f1542aa2fa15e2289105a"
+
+[[package]]
+name = "wasi"
+version = "0.11.1+wasi-snapshot-preview1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ccf3ec651a847eb01de73ccad15eb7d99f80485de043efb2f370cd654f4ea44b"
+
+[[package]]
+name = "wasip2"
+version = "1.0.2+wasi-0.2.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9517f9239f02c069db75e65f174b3da828fe5f5b945c4dd26bd25d89c03ebcf5"
+dependencies = [
+ "wit-bindgen",
+]
+
+[[package]]
+name = "wasip3"
+version = "0.4.0+wasi-0.3.0-rc-2026-01-06"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5428f8bf88ea5ddc08faddef2ac4a67e390b88186c703ce6dbd955e1c145aca5"
+dependencies = [
+ "wit-bindgen",
+]
+
+[[package]]
+name = "wasm-bindgen"
+version = "0.2.114"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6532f9a5c1ece3798cb1c2cfdba640b9b3ba884f5db45973a6f442510a87d38e"
+dependencies = [
+ "cfg-if",
+ "once_cell",
+ "rustversion",
+ "wasm-bindgen-macro",
+ "wasm-bindgen-shared",
+]
+
+[[package]]
+name = "wasm-bindgen-macro"
+version = "0.2.114"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "18a2d50fcf105fb33bb15f00e7a77b772945a2ee45dcf454961fd843e74c18e6"
+dependencies = [
+ "quote",
+ "wasm-bindgen-macro-support",
+]
+
+[[package]]
+name = "wasm-bindgen-macro-support"
+version = "0.2.114"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "03ce4caeaac547cdf713d280eda22a730824dd11e6b8c3ca9e42247b25c631e3"
+dependencies = [
+ "bumpalo",
+ "proc-macro2",
+ "quote",
+ "syn",
+ "wasm-bindgen-shared",
+]
+
+[[package]]
+name = "wasm-bindgen-shared"
+version = "0.2.114"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "75a326b8c223ee17883a4251907455a2431acc2791c98c26279376490c378c16"
+dependencies = [
+ "unicode-ident",
+]
+
+[[package]]
+name = "wasm-compose"
+version = "0.243.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "af801b6f36459023eaec63fdbaedad2fd5a4ab7dc74ecc110a8b5d375c5775e4"
+dependencies = [
+ "anyhow",
+ "heck",
+ "im-rc",
+ "indexmap",
+ "log",
+ "petgraph",
+ "serde",
+ "serde_derive",
+ "serde_yaml",
+ "smallvec",
+ "wasm-encoder 0.243.0",
+ "wasmparser 0.243.0",
+ "wat",
+]
+
+[[package]]
+name = "wasm-encoder"
+version = "0.243.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c55db9c896d70bd9fa535ce83cd4e1f2ec3726b0edd2142079f594fc3be1cb35"
+dependencies = [
+ "leb128fmt",
+ "wasmparser 0.243.0",
+]
+
+[[package]]
+name = "wasm-encoder"
+version = "0.244.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "990065f2fe63003fe337b932cfb5e3b80e0b4d0f5ff650e6985b1048f62c8319"
+dependencies = [
+ "leb128fmt",
+ "wasmparser 0.244.0",
+]
+
+[[package]]
+name = "wasm-encoder"
+version = "0.245.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3f9dca005e69bf015e45577e415b9af8c67e8ee3c0e38b5b0add5aa92581ed5c"
+dependencies = [
+ "leb128fmt",
+ "wasmparser 0.245.1",
+]
+
+[[package]]
+name = "wasm-metadata"
+version = "0.244.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bb0e353e6a2fbdc176932bbaab493762eb1255a7900fe0fea1a2f96c296cc909"
+dependencies = [
+ "anyhow",
+ "indexmap",
+ "wasm-encoder 0.244.0",
+ "wasmparser 0.244.0",
+]
+
+[[package]]
+name = "wasmparser"
+version = "0.243.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f6d8db401b0528ec316dfbe579e6ab4152d61739cfe076706d2009127970159d"
+dependencies = [
+ "bitflags",
+ "hashbrown 0.15.5",
+ "indexmap",
+ "semver",
+ "serde",
+]
+
+[[package]]
+name = "wasmparser"
+version = "0.244.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "47b807c72e1bac69382b3a6fb3dbe8ea4c0ed87ff5629b8685ae6b9a611028fe"
+dependencies = [
+ "bitflags",
+ "hashbrown 0.15.5",
+ "indexmap",
+ "semver",
+]
+
+[[package]]
+name = "wasmparser"
+version = "0.245.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4f08c9adee0428b7bddf3890fc27e015ac4b761cc608c822667102b8bfd6995e"
+dependencies = [
+ "bitflags",
+ "indexmap",
+ "semver",
+]
+
+[[package]]
+name = "wasmprinter"
+version = "0.243.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "eb2b6035559e146114c29a909a3232928ee488d6507a1504d8934e8607b36d7b"
+dependencies = [
+ "anyhow",
+ "termcolor",
+ "wasmparser 0.243.0",
+]
+
+[[package]]
+name = "wasmtime"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e2a83182bf04af87571b4c642300479501684f26bab5597f68f68cded5b098fd"
+dependencies = [
+ "addr2line",
+ "anyhow",
+ "async-trait",
+ "bitflags",
+ "bumpalo",
+ "cc",
+ "cfg-if",
+ "encoding_rs",
+ "futures",
+ "fxprof-processed-profile",
+ "gimli",
+ "hashbrown 0.15.5",
+ "indexmap",
+ "ittapi",
+ "libc",
+ "log",
+ "mach2",
+ "memfd",
+ "object",
+ "once_cell",
+ "postcard",
+ "pulley-interpreter",
+ "rayon",
+ "rustix 1.1.4",
+ "semver",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "smallvec",
+ "target-lexicon",
+ "tempfile",
+ "wasm-compose",
+ "wasm-encoder 0.243.0",
+ "wasmparser 0.243.0",
+ "wasmtime-environ",
+ "wasmtime-internal-cache",
+ "wasmtime-internal-component-macro",
+ "wasmtime-internal-component-util",
+ "wasmtime-internal-cranelift",
+ "wasmtime-internal-fiber",
+ "wasmtime-internal-jit-debug",
+ "wasmtime-internal-jit-icache-coherence",
+ "wasmtime-internal-math",
+ "wasmtime-internal-slab",
+ "wasmtime-internal-unwinder",
+ "wasmtime-internal-versioned-export-macros",
+ "wasmtime-internal-winch",
+ "wat",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "wasmtime-environ"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cb201c41aa23a3642365cfb2e4a183573d85127a3c9d528f56b9997c984541ab"
+dependencies = [
+ "anyhow",
+ "cpp_demangle",
+ "cranelift-bitset",
+ "cranelift-entity",
+ "gimli",
+ "indexmap",
+ "log",
+ "object",
+ "postcard",
+ "rustc-demangle",
+ "semver",
+ "serde",
+ "serde_derive",
+ "smallvec",
+ "target-lexicon",
+ "wasm-encoder 0.243.0",
+ "wasmparser 0.243.0",
+ "wasmprinter",
+ "wasmtime-internal-component-util",
+]
+
+[[package]]
+name = "wasmtime-internal-cache"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fb5b3069d1a67ba5969d0eb1ccd7e141367d4e713f4649aa90356c98e8f19bea"
+dependencies = [
+ "base64",
+ "directories-next",
+ "log",
+ "postcard",
+ "rustix 1.1.4",
+ "serde",
+ "serde_derive",
+ "sha2",
+ "toml",
+ "wasmtime-environ",
+ "windows-sys 0.61.2",
+ "zstd",
+]
+
+[[package]]
+name = "wasmtime-internal-component-macro"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0c924400db7b6ca996fef1b23beb0f41d5c809836b1ec60fc25b4057e2d25d9b"
+dependencies = [
+ "anyhow",
+ "proc-macro2",
+ "quote",
+ "syn",
+ "wasmtime-internal-component-util",
+ "wasmtime-internal-wit-bindgen",
+ "wit-parser 0.243.0",
+]
+
+[[package]]
+name = "wasmtime-internal-component-util"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7d3f65daf4bf3d74ca2fbbe20af0589c42e2b398a073486451425d94fd4afef4"
+
+[[package]]
+name = "wasmtime-internal-cranelift"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "633e889cdae76829738db0114ab3b02fce51ea4a1cd9675a67a65fce92e8b418"
+dependencies = [
+ "cfg-if",
+ "cranelift-codegen",
+ "cranelift-control",
+ "cranelift-entity",
+ "cranelift-frontend",
+ "cranelift-native",
+ "gimli",
+ "itertools",
+ "log",
+ "object",
+ "pulley-interpreter",
+ "smallvec",
+ "target-lexicon",
+ "thiserror 2.0.18",
+ "wasmparser 0.243.0",
+ "wasmtime-environ",
+ "wasmtime-internal-math",
+ "wasmtime-internal-unwinder",
+ "wasmtime-internal-versioned-export-macros",
+]
+
+[[package]]
+name = "wasmtime-internal-fiber"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "deb126adc5d0c72695cfb77260b357f1b81705a0f8fa30b3944e7c2219c17341"
+dependencies = [
+ "cc",
+ "cfg-if",
+ "libc",
+ "rustix 1.1.4",
+ "wasmtime-environ",
+ "wasmtime-internal-versioned-export-macros",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "wasmtime-internal-jit-debug"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e66ff7f90a8002187691ff6237ffd09f954a0ebb9de8b2ff7f5c62632134120"
+dependencies = [
+ "cc",
+ "object",
+ "rustix 1.1.4",
+ "wasmtime-internal-versioned-export-macros",
+]
+
+[[package]]
+name = "wasmtime-internal-jit-icache-coherence"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4b96df23179ae16d54fb3a420f84ffe4383ec9dd06fad3e5bc782f85f66e8e08"
+dependencies = [
+ "anyhow",
+ "cfg-if",
+ "libc",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "wasmtime-internal-math"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "86d1380926682b44c383e9a67f47e7a95e60c6d3fa8c072294dab2c7de6168a0"
+dependencies = [
+ "libm",
+]
+
+[[package]]
+name = "wasmtime-internal-slab"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9b63cbea1c0192c7feb7c0dfb35f47166988a3742f29f46b585ef57246c65764"
+
+[[package]]
+name = "wasmtime-internal-unwinder"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f25c392c7e5fb891a7416e3c34cfbd148849271e8c58744fda875dde4bec4d6a"
+dependencies = [
+ "cfg-if",
+ "cranelift-codegen",
+ "log",
+ "object",
+ "wasmtime-environ",
+]
+
+[[package]]
+name = "wasmtime-internal-versioned-export-macros"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "70f8b9796a3f0451a7b702508b303d654de640271ac80287176de222f187a237"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "wasmtime-internal-winch"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c0063e61f1d0b2c20e9cfc58361a6513d074a23c80b417aac3033724f51648a0"
+dependencies = [
+ "cranelift-codegen",
+ "gimli",
+ "log",
+ "object",
+ "target-lexicon",
+ "wasmparser 0.243.0",
+ "wasmtime-environ",
+ "wasmtime-internal-cranelift",
+ "winch-codegen",
+]
+
+[[package]]
+name = "wasmtime-internal-wit-bindgen"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "587699ca7cae16b4a234ffcc834f37e75675933d533809919b52975f5609e2ef"
+dependencies = [
+ "anyhow",
+ "bitflags",
+ "heck",
+ "indexmap",
+ "wit-parser 0.243.0",
+]
+
+[[package]]
+name = "wasmtime-wasi"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fc2eb9dc95baed3cd86fdfebf9f9f333337eb308bf8bd973e0c7b06d9418c35f"
+dependencies = [
+ "anyhow",
+ "async-trait",
+ "bitflags",
+ "bytes",
+ "cap-fs-ext",
+ "cap-net-ext",
+ "cap-rand",
+ "cap-std",
+ "cap-time-ext",
+ "fs-set-times",
+ "futures",
+ "io-extras",
+ "io-lifetimes",
+ "rustix 1.1.4",
+ "system-interface",
+ "thiserror 2.0.18",
+ "tokio",
+ "tracing",
+ "url",
+ "wasmtime",
+ "wasmtime-wasi-io",
+ "wiggle",
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "wasmtime-wasi-io"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a0b8402f1e04385071fdd96aca97cba995d7376b572e42ce5841d5b6aaf6fa30"
+dependencies = [
+ "anyhow",
+ "async-trait",
+ "bytes",
+ "futures",
+ "wasmtime",
+]
+
+[[package]]
+name = "wast"
+version = "35.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2ef140f1b49946586078353a453a1d28ba90adfc54dde75710bc1931de204d68"
+dependencies = [
+ "leb128",
+]
+
+[[package]]
+name = "wast"
+version = "245.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "28cf1149285569120b8ce39db8b465e8a2b55c34cbb586bd977e43e2bc7300bf"
+dependencies = [
+ "bumpalo",
+ "leb128fmt",
+ "memchr",
+ "unicode-width",
+ "wasm-encoder 0.245.1",
+]
+
+[[package]]
+name = "wat"
+version = "1.245.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cd48d1679b6858988cb96b154dda0ec5bbb09275b71db46057be37332d5477be"
+dependencies = [
+ "wast 245.0.1",
+]
+
+[[package]]
+name = "wiggle"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a69a60bcbe1475c5dc9ec89210ade54823d44f742e283cba64f98f89697c4cec"
+dependencies = [
+ "anyhow",
+ "bitflags",
+ "thiserror 2.0.18",
+ "tracing",
+ "wasmtime",
+ "wiggle-macro",
+]
+
+[[package]]
+name = "wiggle-generate"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "21f3dc0fd4dcfc7736434bb216179a2147835309abc09bf226736a40d484548f"
+dependencies = [
+ "anyhow",
+ "heck",
+ "proc-macro2",
+ "quote",
+ "syn",
+ "witx",
+]
+
+[[package]]
+name = "wiggle-macro"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fea2aea744eded58ae092bf57110c27517dab7d5a300513ff13897325c5c5021"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+ "wiggle-generate",
+]
+
+[[package]]
+name = "winapi"
+version = "0.3.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5c839a674fcd7a98952e593242ea400abe93992746761e38641405d28b00f419"
+dependencies = [
+ "winapi-i686-pc-windows-gnu",
+ "winapi-x86_64-pc-windows-gnu",
+]
+
+[[package]]
+name = "winapi-i686-pc-windows-gnu"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
+
+[[package]]
+name = "winapi-util"
+version = "0.1.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
+dependencies = [
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "winapi-x86_64-pc-windows-gnu"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
+
+[[package]]
+name = "winch-codegen"
+version = "41.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c55de3ac5b8bd71e5f6c87a9e511dd3ceb194bdb58183c6a7bf21cd8c0e46fbc"
+dependencies = [
+ "anyhow",
+ "cranelift-assembler-x64",
+ "cranelift-codegen",
+ "gimli",
+ "regalloc2",
+ "smallvec",
+ "target-lexicon",
+ "thiserror 2.0.18",
+ "wasmparser 0.243.0",
+ "wasmtime-environ",
+ "wasmtime-internal-cranelift",
+ "wasmtime-internal-math",
+]
+
+[[package]]
+name = "windows-core"
+version = "0.62.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b8e83a14d34d0623b51dce9581199302a221863196a1dde71a7663a4c2be9deb"
+dependencies = [
+ "windows-implement",
+ "windows-interface",
+ "windows-link",
+ "windows-result",
+ "windows-strings",
+]
+
+[[package]]
+name = "windows-implement"
+version = "0.60.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "053e2e040ab57b9dc951b72c264860db7eb3b0200ba345b4e4c3b14f67855ddf"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "windows-interface"
+version = "0.59.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3f316c4a2570ba26bbec722032c4099d8c8bc095efccdc15688708623367e358"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "windows-link"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f0805222e57f7521d6a62e36fa9163bc891acd422f971defe97d64e70d0a4fe5"
+
+[[package]]
+name = "windows-result"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7781fa89eaf60850ac3d2da7af8e5242a5ea78d1a11c49bf2910bb5a73853eb5"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "windows-strings"
+version = "0.5.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7837d08f69c77cf6b07689544538e017c1bfcf57e34b4c0ff58e6c2cd3b37091"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "windows-sys"
+version = "0.59.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e38bc4d79ed67fd075bcc251a1c39b32a1776bbe92e5bef1f0bf1f8c531853b"
+dependencies = [
+ "windows-targets",
+]
+
+[[package]]
+name = "windows-sys"
+version = "0.61.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ae137229bcbd6cdf0f7b80a31df61766145077ddf49416a728b02cb3921ff3fc"
+dependencies = [
+ "windows-link",
+]
+
+[[package]]
+name = "windows-targets"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9b724f72796e036ab90c1021d4780d4d3d648aca59e491e6b98e725b84e99973"
+dependencies = [
+ "windows_aarch64_gnullvm",
+ "windows_aarch64_msvc",
+ "windows_i686_gnu",
+ "windows_i686_gnullvm",
+ "windows_i686_msvc",
+ "windows_x86_64_gnu",
+ "windows_x86_64_gnullvm",
+ "windows_x86_64_msvc",
+]
+
+[[package]]
+name = "windows_aarch64_gnullvm"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "32a4622180e7a0ec044bb555404c800bc9fd9ec262ec147edd5989ccd0c02cd3"
+
+[[package]]
+name = "windows_aarch64_msvc"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "09ec2a7bb152e2252b53fa7803150007879548bc709c039df7627cabbd05d469"
+
+[[package]]
+name = "windows_i686_gnu"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e9b5ad5ab802e97eb8e295ac6720e509ee4c243f69d781394014ebfe8bbfa0b"
+
+[[package]]
+name = "windows_i686_gnullvm"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0eee52d38c090b3caa76c563b86c3a4bd71ef1a819287c19d586d7334ae8ed66"
+
+[[package]]
+name = "windows_i686_msvc"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "240948bc05c5e7c6dabba28bf89d89ffce3e303022809e73deaefe4f6ec56c66"
+
+[[package]]
+name = "windows_x86_64_gnu"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "147a5c80aabfbf0c7d901cb5895d1de30ef2907eb21fbbab29ca94c5b08b1a78"
+
+[[package]]
+name = "windows_x86_64_gnullvm"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "24d5b23dc417412679681396f2b49f3de8c1473deb516bd34410872eff51ed0d"
+
+[[package]]
+name = "windows_x86_64_msvc"
+version = "0.52.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "589f6da84c646204747d1270a2a5661ea66ed1cced2631d546fdfb155959f9ec"
+
+[[package]]
+name = "winnow"
+version = "0.7.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "df79d97927682d2fd8adb29682d1140b343be4ac0f08fd68b7765d9c059d3945"
+
+[[package]]
+name = "winx"
+version = "0.36.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3f3fd376f71958b862e7afb20cfe5a22830e1963462f3a17f49d82a6c1d1f42d"
+dependencies = [
+ "bitflags",
+ "windows-sys 0.59.0",
+]
+
+[[package]]
+name = "wit-bindgen"
+version = "0.51.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d7249219f66ced02969388cf2bb044a09756a083d0fab1e566056b04d9fbcaa5"
+dependencies = [
+ "wit-bindgen-rust-macro",
+]
+
+[[package]]
+name = "wit-bindgen-core"
+version = "0.51.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ea61de684c3ea68cb082b7a88508a8b27fcc8b797d738bfc99a82facf1d752dc"
+dependencies = [
+ "anyhow",
+ "heck",
+ "wit-parser 0.244.0",
+]
+
+[[package]]
+name = "wit-bindgen-rust"
+version = "0.51.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b7c566e0f4b284dd6561c786d9cb0142da491f46a9fbed79ea69cdad5db17f21"
+dependencies = [
+ "anyhow",
+ "heck",
+ "indexmap",
+ "prettyplease",
+ "syn",
+ "wasm-metadata",
+ "wit-bindgen-core",
+ "wit-component",
+]
+
+[[package]]
+name = "wit-bindgen-rust-macro"
+version = "0.51.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0c0f9bfd77e6a48eccf51359e3ae77140a7f50b1e2ebfe62422d8afdaffab17a"
+dependencies = [
+ "anyhow",
+ "prettyplease",
+ "proc-macro2",
+ "quote",
+ "syn",
+ "wit-bindgen-core",
+ "wit-bindgen-rust",
+]
+
+[[package]]
+name = "wit-component"
+version = "0.244.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9d66ea20e9553b30172b5e831994e35fbde2d165325bec84fc43dbf6f4eb9cb2"
+dependencies = [
+ "anyhow",
+ "bitflags",
+ "indexmap",
+ "log",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "wasm-encoder 0.244.0",
+ "wasm-metadata",
+ "wasmparser 0.244.0",
+ "wit-parser 0.244.0",
+]
+
+[[package]]
+name = "wit-parser"
+version = "0.243.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "df983a8608e513d8997f435bb74207bf0933d0e49ca97aa9d8a6157164b9b7fc"
+dependencies = [
+ "anyhow",
+ "id-arena",
+ "indexmap",
+ "log",
+ "semver",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "unicode-xid",
+ "wasmparser 0.243.0",
+]
+
+[[package]]
+name = "wit-parser"
+version = "0.244.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ecc8ac4bc1dc3381b7f59c34f00b67e18f910c2c0f50015669dde7def656a736"
+dependencies = [
+ "anyhow",
+ "id-arena",
+ "indexmap",
+ "log",
+ "semver",
+ "serde",
+ "serde_derive",
+ "serde_json",
+ "unicode-xid",
+ "wasmparser 0.244.0",
+]
+
+[[package]]
+name = "witx"
+version = "0.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e366f27a5cabcddb2706a78296a40b8fcc451e1a6aba2fc1d94b4a01bdaaef4b"
+dependencies = [
+ "anyhow",
+ "log",
+ "thiserror 1.0.69",
+ "wast 35.0.2",
+]
+
+[[package]]
+name = "writeable"
+version = "0.6.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9edde0db4769d2dc68579893f2306b26c6ecfbe0ef499b013d731b7b9247e0b9"
+
+[[package]]
+name = "yoke"
+version = "0.8.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "72d6e5c6afb84d73944e5cedb052c4680d5657337201555f9f2a16b7406d4954"
+dependencies = [
+ "stable_deref_trait",
+ "yoke-derive",
+ "zerofrom",
+]
+
+[[package]]
+name = "yoke-derive"
+version = "0.8.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b659052874eb698efe5b9e8cf382204678a0086ebf46982b79d6ca3182927e5d"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+ "synstructure",
+]
+
+[[package]]
+name = "zerocopy"
+version = "0.8.42"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f2578b716f8a7a858b7f02d5bd870c14bf4ddbbcf3a4c05414ba6503640505e3"
+dependencies = [
+ "zerocopy-derive",
+]
+
+[[package]]
+name = "zerocopy-derive"
+version = "0.8.42"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7e6cc098ea4d3bd6246687de65af3f920c430e236bee1e3bf2e441463f08a02f"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "zerofrom"
+version = "0.1.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "50cc42e0333e05660c3587f3bf9d0478688e15d870fab3346451ce7f8c9fbea5"
+dependencies = [
+ "zerofrom-derive",
+]
+
+[[package]]
+name = "zerofrom-derive"
+version = "0.1.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d71e5d6e06ab090c67b5e44993ec16b72dcbaabc526db883a360057678b48502"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+ "synstructure",
+]
+
+[[package]]
+name = "zerotrie"
+version = "0.2.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2a59c17a5562d507e4b54960e8569ebee33bee890c70aa3fe7b97e85a9fd7851"
+dependencies = [
+ "displaydoc",
+ "yoke",
+ "zerofrom",
+]
+
+[[package]]
+name = "zerovec"
+version = "0.11.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6c28719294829477f525be0186d13efa9a3c602f7ec202ca9e353d310fb9a002"
+dependencies = [
+ "yoke",
+ "zerofrom",
+ "zerovec-derive",
+]
+
+[[package]]
+name = "zerovec-derive"
+version = "0.11.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "eadce39539ca5cb3985590102671f2567e659fca9666581ad3411d59207951f3"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
+[[package]]
+name = "zmij"
+version = "1.0.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b8848ee67ecc8aedbaf3e4122217aff892639231befc6a1b58d29fff4c2cabaa"
+
+[[package]]
+name = "zstd"
+version = "0.13.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e91ee311a569c327171651566e07972200e76fcfe2242a4fa446149a3881c08a"
+dependencies = [
+ "zstd-safe",
+]
+
+[[package]]
+name = "zstd-safe"
+version = "7.2.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8f49c4d5f0abb602a93fb8736af2a4f4dd9512e36f7f570d66e65ff867ed3b9d"
+dependencies = [
+ "zstd-sys",
+]
+
+[[package]]
+name = "zstd-sys"
+version = "2.0.16+zstd.1.5.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "91e19ebc2adc8f83e43039e79776e3fda8ca919132d68a1fed6a5faca2683748"
+dependencies = [
+ "cc",
+ "pkg-config",
+]
diff --git a/bench/rust-host/Cargo.toml b/bench/rust-host/Cargo.toml
new file mode 100644
index 0000000..f193627
--- /dev/null
+++ b/bench/rust-host/Cargo.toml
@@ -0,0 +1,9 @@
+[package]
+name = "rust-host"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+wasmtime = { version = "41", features = ["component-model"] }
+wasmtime-wasi = "41"
+anyhow = "1"
diff --git a/bench/rust-host/src/main.rs b/bench/rust-host/src/main.rs
new file mode 100644
index 0000000..b63ddda
--- /dev/null
+++ b/bench/rust-host/src/main.rs
@@ -0,0 +1,79 @@
+use std::time::Instant;
+use wasmtime::component::{Component, Linker};
+use wasmtime::{Config, Engine, Store};
+use wasmtime_wasi::{ResourceTable, WasiCtx, WasiCtxBuilder, WasiCtxView, WasiView};
+
+wasmtime::component::bindgen!({
+    world: "codec",
+    path: "../../codec/wit/codec.wit",
+});
+
+struct State {
+    ctx: WasiCtx,
+    table: ResourceTable,
+}
+
+impl WasiView for State {
+    fn ctx(&mut self) -> WasiCtxView<'_> {
+        WasiCtxView {
+            ctx: &mut self.ctx,
+            table: &mut self.table,
+        }
+    }
+}
+
+fn main() -> anyhow::Result<()> {
+    let mut config = Config::new();
+    config.wasm_component_model(true);
+    let engine = Engine::new(&config)?;
+
+    let wasm_path = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR"))
+        .join("../../codec/identity-c/identity.wasm");
+    let component = Component::from_file(&engine, &wasm_path)?;
+
+    let mut linker: Linker<State> = Linker::new(&engine);
+    wasmtime_wasi::p2::add_to_linker_sync(&mut linker)?;
+
+    let state = State {
+        ctx: WasiCtxBuilder::new().build(),
+        table: ResourceTable::new(),
+    };
+    let mut store = Store::new(&engine, state);
+    let codec = Codec::instantiate(&mut store, &component, &linker)?;
+
+    // Warm up the compilation cache (same as Python test conditions).
+    // NOTE: the exact accessor name depends on wasmtime 41 bindgen output for
+    // a world that re-exports a foreign interface. If this does not compile,
+    // run `cargo build` and read the error — it will name the actual accessor.
+    {
+        let warm = vec![("bytes".to_string(), vec![0u8; 64])];
+        let _ = codec.chonkle_codec_transform().call_decode(&mut store, &warm)?;
+    }
+
+    for (size, label) in [(1usize << 20, "1 MB"), (2usize << 20, "2 MB")] {
+        let inputs = vec![("bytes".to_string(), vec![0u8; size])];
+        for run in 1..=3 {
+            let t0 = Instant::now();
+            let result = codec
+                .chonkle_codec_transform()
+                .call_decode(&mut store, &inputs)?
+                .map_err(|e| anyhow::anyhow!(e))?;
+            let elapsed = t0.elapsed();
+            let out = result.iter().map(|(_, v)| v.len()).sum::<usize>();
+            let total = size + out;
+            let throughput = total as f64 / elapsed.as_secs_f64() / 1_048_576.0;
+            println!(
+                "[TIMING] identity.wasm decode: fn={:.3}s  in={}B out={}B \
+                 abi_total={}B throughput={:.1}MB/s  ({} run {}/3, Rust host)",
+                elapsed.as_secs_f64(),
+                size,
+                out,
+                total,
+                throughput,
+                label,
+                run
+            );
+        }
+    }
+    Ok(())
+}
diff --git a/codec/README.md b/codec/README.md
new file mode 100644
index 0000000..f29e135
--- /dev/null
+++ b/codec/README.md
@@ -0,0 +1,115 @@
+# codec/
+
+Wasm codec implementations for the chonkle pipeline. Codecs fall into two categories:
+
+- **Component Model codecs** implement the `chonkle:codec/transform` WIT interface and are lifted to Component Model `.wasm` binaries. See [docs/reference/codec-contract/component-model.md](../docs/reference/codec-contract/component-model.md).
+- **Core Wasm codecs** are wasm32-wasi reactor modules that export `memory`, `alloc`, `dealloc`, `encode`, and `decode` using the binary port-map wire format. See [docs/reference/codec-contract/core.md](../docs/reference/codec-contract/core.md).
+
+All codec build processes include a post-build step to embed the codec signature into the `.wasm` binary:
+
+```
+chonkle embed-signature <output.wasm> signature.json
+```
+
+Each codec directory contains a `signature.json` build input that is embedded as a `chonkle:signature` custom section.
+
+## Component Model codecs
+
+### identity-c/
+
+Passthrough codec used for benchmarking pipeline overhead. `encode` and
+`decode` are both no-ops: output `bytes` = input `bytes`.
+
+Ports: `bytes` in → `bytes` out.
+
+Toolchain: C + `zig cc` → wasm32-wasi reactor → `wasm-tools` component lift.
+
+```
+cd codec/identity-c
+zig build
+cp zig-out/identity.wasm identity.wasm
+chonkle embed-signature identity.wasm signature.json
+```
+
+### tiff-predictor-2-c/
+
+TIFF horizontal differencing predictor (TIFF Predictor 2). `encode` applies
+differencing; `decode` reconstructs original values. Reversible; encode→decode
+roundtrip is identity.
+
+Ports: `bytes` (required) + `bytes_per_sample` (int, required) + `width`
+(int, required) in → `bytes` out.
+
+Same toolchain as identity-c.
+
+```
+cd codec/tiff-predictor-2-c
+zig build -Doptimize=ReleaseFast
+cp zig-out/tiff-predictor-2.wasm tiff-predictor-2.wasm
+chonkle embed-signature tiff-predictor-2.wasm signature.json
+```
+
+### zlib-rs/
+
+zlib compress/decompress via the `flate2` crate with the `rust_backend`
+feature (pure Rust, no native dependencies).
+
+Ports: `bytes` (required) + `level` (int, encode-only, default 6) in →
+`bytes` out.
+
+Toolchain: `cargo-component` targeting `wasm32-wasip2`.
+
+```
+cd codec/zlib-rs
+cargo component build --target wasm32-wasip2 --release
+cp target/wasm32-wasip1/release/zlib.wasm zlib.wasm
+chonkle embed-signature zlib.wasm signature.json
+```
+
+## Core Wasm codecs
+
+### identity-core-c/
+
+Passthrough codec for benchmarking using the core ABI. `encode` and `decode` are both no-ops: output `bytes` = input `bytes`. Uses the binary port-map wire format instead of the Component Model canonical ABI.
+
+Ports: `bytes` in → `bytes` out.
+
+Exports: `memory`, `alloc`, `dealloc`, `encode`, `decode`.
+
+Toolchain: C + `zig cc` → wasm32-wasi reactor (no Component Model lift step). Same build toolchain as identity-c and tiff-predictor-2-c but without the `wasm-tools component new` step.
+
+```
+cd codec/identity-core-c
+zig build
+cp zig-out/identity-core.wasm identity-core.wasm
+chonkle embed-signature identity-core.wasm signature.json
+```
+
+---
+
+## Shared C infrastructure — shared/
+
+Used by C+Zig codecs:
+
+- **codec.h** / **codec.c** — WIT-generated canonical ABI bindings for Component Model codecs
+- **codec_component_type.o** — component type object produced by wit-bindgen; passed to `zig cc` as a source file alongside the C sources
+- **core_abi.h** / **core_abi.c** — C reference implementation of the core ABI port-map parse/serialize/find operations, used by core wasm codecs like identity-core-c
+
+Regenerate WIT bindings after any WIT interface change:
+
+```
+wit-bindgen c codec/wit/ --world codec --out-dir codec/shared/
+```
+
+## WASI adapter — wasi_snapshot_preview1.reactor.wasm
+
+Lifts a `wasm32-wasi` (WASI preview1) reactor module to a Component Model
+(WASI preview2) component at build time. Used by `wasm-tools component new
+--adapt` in the C+Zig codec builds. Downloaded from the wasmtime v41.0.0
+GitHub release to match the `wasmtime` Python package version.
+
+---
+
+## Adding a codec
+
+See the [codec build guides](../docs/guides/) for step-by-step instructions covering each backend and toolchain.
diff --git a/codec/identity-c/.gitignore b/codec/identity-c/.gitignore
new file mode 100644
index 0000000..c88eb55
--- /dev/null
+++ b/codec/identity-c/.gitignore
@@ -0,0 +1,2 @@
+.zig-cache/
+/zig-out
diff --git a/codec/identity-c/build.zig b/codec/identity-c/build.zig
new file mode 100644
index 0000000..920d1f7
--- /dev/null
+++ b/codec/identity-c/build.zig
@@ -0,0 +1,38 @@
+const std = @import("std");
+
+pub fn build(b: *std.Build) void {
+    // Identity codec is a benchmarking artifact only — always compile at -O2.
+
+    // Step 1: compile C sources to a core wasm32-wasi reactor module.
+    const compile = b.addSystemCommand(&.{
+        b.graph.zig_exe,
+        "cc",
+        "--target=wasm32-wasi",
+        "-mexec-model=reactor",
+        "-fvisibility=hidden",
+        "-lc",
+        "-O2",
+    });
+
+    compile.addArg("-o");
+    const core_wasm = compile.addOutputFileArg("identity-core.wasm");
+    compile.addFileArg(b.path("../shared/codec.c"));
+    compile.addFileArg(b.path("codec_impl.c"));
+    compile.addFileArg(b.path("../shared/codec_component_type.o"));
+
+    // Step 2: lift the core module to a Component Model .wasm via the WASI
+    // preview1 → preview2 adapter bundled in codec/wasi_snapshot_preview1.reactor.wasm.
+    const componentize = b.addSystemCommand(&.{ "wasm-tools", "component", "new" });
+    componentize.addFileArg(core_wasm);
+    componentize.addArg("--adapt");
+    componentize.addPrefixedFileArg(
+        "wasi_snapshot_preview1=",
+        b.path("../wasi_snapshot_preview1.reactor.wasm"),
+    );
+    componentize.addArg("-o");
+    const component = componentize.addOutputFileArg("identity.wasm");
+
+    b.getInstallStep().dependOn(
+        &b.addInstallFile(component, "identity.wasm").step,
+    );
+}
diff --git a/codec/identity-c/codec_impl.c b/codec/identity-c/codec_impl.c
new file mode 100644
index 0000000..a51645c
--- /dev/null
+++ b/codec/identity-c/codec_impl.c
@@ -0,0 +1,98 @@
+/**
+ * WIT entry points for the identity Component Model codec.
+ *
+ * Implements chonkle:codec/transform@0.1.0 — both encode and decode
+ * perform a memcpy, passing bytes through unchanged. This isolates
+ * pure ABI cost from codec computation for benchmarking purposes.
+ *
+ * Expected port-map inputs:
+ *   "bytes"  list<u8>  raw data (required)
+ *
+ * Output port-map:
+ *   "bytes"  list<u8>  same bytes as input
+ */
+
+#include "../shared/codec.h"
+#include <stdlib.h>
+#include <string.h>
+
+/*
+ * Override the weak cabi_realloc from codec.c so that OOM calls
+ * __builtin_trap() instead of abort(), avoiding WASI stderr imports.
+ */
+__attribute__((__export_name__("cabi_realloc")))
+void *cabi_realloc(void *ptr, size_t old_size, size_t align, size_t new_size) {
+    (void)old_size;
+    if (new_size == 0) return (void *)align;
+    void *ret = realloc(ptr, new_size);
+    if (!ret) __builtin_trap();
+    return ret;
+}
+
+/* Find a named port in the input port-map; returns NULL if absent. */
+static codec_list_u8_t *find_port(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    const char *name)
+{
+    size_t name_len = strlen(name);
+    for (size_t i = 0; i < inputs->len; i++) {
+        codec_tuple2_port_name_list_u8_t *e = &inputs->ptr[i];
+        if (e->f0.len == name_len && memcmp(e->f0.ptr, name, name_len) == 0)
+            return &e->f1;
+    }
+    return NULL;
+}
+
+/*
+ * Allocate a port-map with a single "bytes" port pointing to buf/buf_len.
+ * Ownership of buf transfers to the caller's port-map.
+ */
+static bool make_bytes_result(
+    uint8_t *buf, size_t buf_len,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    codec_tuple2_port_name_list_u8_t *entry = malloc(sizeof(*entry));
+    if (!entry) {
+        codec_string_set(err, "out of memory");
+        return false;
+    }
+    codec_string_dup(&entry->f0, "bytes");
+    entry->f1.ptr = buf;
+    entry->f1.len = buf_len;
+    ret->ptr = entry;
+    ret->len = 1;
+    return true;
+}
+
+static bool transform(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    codec_list_u8_t *bytes_port = find_port(inputs, "bytes");
+    if (!bytes_port) { codec_string_set(err, "missing port: bytes"); return false; }
+
+    size_t input_len = bytes_port->len;
+    uint8_t *out = malloc(input_len);
+    if (!out) { codec_string_set(err, "out of memory"); return false; }
+    memcpy(out, bytes_port->ptr, input_len);
+
+    return make_bytes_result(out, input_len, ret, err);
+}
+
+bool exports_chonkle_codec_transform_encode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    return transform(inputs, ret, err);
+}
+
+bool exports_chonkle_codec_transform_decode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    return transform(inputs, ret, err);
+}
diff --git a/codec/identity-c/identity.wasm b/codec/identity-c/identity.wasm
new file mode 100644
index 0000000..942b71d
Binary files /dev/null and b/codec/identity-c/identity.wasm differ
diff --git a/codec/identity-c/signature.json b/codec/identity-c/signature.json
new file mode 100644
index 0000000..29cfb8f
--- /dev/null
+++ b/codec/identity-c/signature.json
@@ -0,0 +1,6 @@
+{
+  "codec_id": "identity",
+  "implementation": "identity-c",
+  "inputs": {"bytes": {"type": "bytes", "required": true}},
+  "outputs": {"bytes": {"type": "bytes"}}
+}
diff --git a/codec/identity-core-c/.gitignore b/codec/identity-core-c/.gitignore
new file mode 100644
index 0000000..936ea61
--- /dev/null
+++ b/codec/identity-core-c/.gitignore
@@ -0,0 +1,3 @@
+*.wasm
+.zig-cache/
+zig-out/
diff --git a/codec/identity-core-c/build.zig b/codec/identity-core-c/build.zig
new file mode 100644
index 0000000..70986ea
--- /dev/null
+++ b/codec/identity-core-c/build.zig
@@ -0,0 +1,24 @@
+const std = @import("std");
+
+pub fn build(b: *std.Build) void {
+    // Identity-core codec: core wasm module (no Component Model lift).
+
+    const compile = b.addSystemCommand(&.{
+        b.graph.zig_exe,
+        "cc",
+        "--target=wasm32-wasi",
+        "-mexec-model=reactor",
+        "-fvisibility=hidden",
+        "-lc",
+        "-O2",
+    });
+
+    compile.addArg("-o");
+    const core_wasm = compile.addOutputFileArg("identity-core.wasm");
+    compile.addFileArg(b.path("../shared/core_abi.c"));
+    compile.addFileArg(b.path("codec_impl.c"));
+
+    b.getInstallStep().dependOn(
+        &b.addInstallFile(core_wasm, "identity-core.wasm").step,
+    );
+}
diff --git a/codec/identity-core-c/codec_impl.c b/codec/identity-core-c/codec_impl.c
new file mode 100644
index 0000000..abe201a
--- /dev/null
+++ b/codec/identity-core-c/codec_impl.c
@@ -0,0 +1,89 @@
+/**
+ * Identity codec using the core ABI (no Component Model).
+ *
+ * Implements the chonkle core ABI (alloc, dealloc, encode, decode)
+ * with the binary port-map wire format. Both encode and decode
+ * perform a memcpy, passing bytes through unchanged.
+ *
+ * Expected port-map inputs:
+ *   "bytes"  raw data (required)
+ *
+ * Output port-map:
+ *   "bytes"  same bytes as input
+ */
+
+#include "../shared/core_abi.h"
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+
+__attribute__((__export_name__("alloc")))
+int32_t codec_alloc(int32_t size) {
+    void *ptr = malloc((size_t)size);
+    return (int32_t)(uintptr_t)ptr;
+}
+
+__attribute__((__export_name__("dealloc")))
+void codec_dealloc(int32_t ptr, int32_t size) {
+    (void)size;
+    free((void *)(uintptr_t)ptr);
+}
+
+static int64_t transform(int32_t pm_ptr, int32_t pm_len) {
+    uint8_t *input_buf = (uint8_t *)(uintptr_t)pm_ptr;
+    core_abi_port_map_t pm = core_abi_parse_port_map(input_buf, (uint32_t)pm_len);
+
+    if (pm.count == 0) {
+        free(input_buf);
+        return CORE_ABI_ERROR;
+    }
+
+    const core_abi_port_t *bytes_port = core_abi_find_port(&pm, "bytes");
+    if (!bytes_port) {
+        core_abi_free_port_map(&pm);
+        free(input_buf);
+        return CORE_ABI_ERROR;
+    }
+
+    /* Copy input bytes before freeing the input buffer. */
+    uint32_t out_data_len = bytes_port->data_len;
+    uint8_t *out_data = malloc(out_data_len);
+    if (!out_data) {
+        core_abi_free_port_map(&pm);
+        free(input_buf);
+        return CORE_ABI_ERROR;
+    }
+    memcpy(out_data, bytes_port->data, out_data_len);
+
+    core_abi_free_port_map(&pm);
+    free(input_buf);
+
+    /* Build output port-map with a single "bytes" port. */
+    core_abi_port_t out_entry;
+    out_entry.name = "bytes";
+    out_entry.name_len = 5;
+    out_entry.data = out_data;
+    out_entry.data_len = out_data_len;
+
+    core_abi_port_map_t out_pm;
+    out_pm.entries = &out_entry;
+    out_pm.count = 1;
+
+    uint32_t ser_len;
+    uint8_t *ser_buf = core_abi_serialize_port_map(&out_pm, &ser_len);
+    free(out_data);
+
+    if (!ser_buf) return CORE_ABI_ERROR;
+
+    return core_abi_pack_result((uint32_t)(uintptr_t)ser_buf, ser_len);
+}
+
+__attribute__((__export_name__("encode")))
+int64_t encode(int32_t pm_ptr, int32_t pm_len) {
+    return transform(pm_ptr, pm_len);
+}
+
+__attribute__((__export_name__("decode")))
+int64_t decode(int32_t pm_ptr, int32_t pm_len) {
+    return transform(pm_ptr, pm_len);
+}
diff --git a/codec/identity-core-c/signature.json b/codec/identity-core-c/signature.json
new file mode 100644
index 0000000..ea9db56
--- /dev/null
+++ b/codec/identity-core-c/signature.json
@@ -0,0 +1,6 @@
+{
+  "codec_id": "identity",
+  "implementation": "identity-core-c",
+  "inputs": {"bytes": {"type": "bytes", "required": true}},
+  "outputs": {"bytes": {"type": "bytes"}}
+}
diff --git a/codec/shared/codec.c b/codec/shared/codec.c
new file mode 100644
index 0000000..bb9949b
--- /dev/null
+++ b/codec/shared/codec.c
@@ -0,0 +1,219 @@
+// Generated by `wit-bindgen` 0.53.1. DO NOT EDIT!
+#include "codec.h"
+#include <stdlib.h>
+#include <string.h>
+
+// Exported Functions from `chonkle:codec/transform@0.1.0`
+
+__attribute__((__weak__, __export_name__("cabi_post_chonkle:codec/transform@0.1.0#encode")))
+void __wasm_export_exports_chonkle_codec_transform_encode_post_return(uint8_t * arg0) {
+  switch ((int32_t) (int32_t) *((uint8_t*) (arg0 + 0))) {
+    case 0: {
+      size_t len0 = *((size_t*) (arg0 + (2*sizeof(void*))));
+      if (len0 > 0) {
+        uint8_t *ptr1 = *((uint8_t **) (arg0 + sizeof(void*)));
+        for (size_t i2 = 0; i2 < len0; i2++) {
+          uint8_t *base = ptr1 + i2 * (4*sizeof(void*));
+          (void) base;
+          if ((*((size_t*) (base + sizeof(void*)))) > 0) {
+            free(*((uint8_t **) (base + 0)));
+          }
+          size_t len = *((size_t*) (base + (3*sizeof(void*))));
+          if (len > 0) {
+            uint8_t *ptr = *((uint8_t **) (base + (2*sizeof(void*))));
+            for (size_t i = 0; i < len; i++) {
+              uint8_t *base = ptr + i * 1;
+              (void) base;
+            }
+            free(ptr);
+          }
+        }
+        free(ptr1);
+      }
+      break;
+    }
+    case 1: {
+      if ((*((size_t*) (arg0 + (2*sizeof(void*))))) > 0) {
+        free(*((uint8_t **) (arg0 + sizeof(void*))));
+      }
+      break;
+    }
+  }
+}
+
+__attribute__((__weak__, __export_name__("cabi_post_chonkle:codec/transform@0.1.0#decode")))
+void __wasm_export_exports_chonkle_codec_transform_decode_post_return(uint8_t * arg0) {
+  switch ((int32_t) (int32_t) *((uint8_t*) (arg0 + 0))) {
+    case 0: {
+      size_t len0 = *((size_t*) (arg0 + (2*sizeof(void*))));
+      if (len0 > 0) {
+        uint8_t *ptr1 = *((uint8_t **) (arg0 + sizeof(void*)));
+        for (size_t i2 = 0; i2 < len0; i2++) {
+          uint8_t *base = ptr1 + i2 * (4*sizeof(void*));
+          (void) base;
+          if ((*((size_t*) (base + sizeof(void*)))) > 0) {
+            free(*((uint8_t **) (base + 0)));
+          }
+          size_t len = *((size_t*) (base + (3*sizeof(void*))));
+          if (len > 0) {
+            uint8_t *ptr = *((uint8_t **) (base + (2*sizeof(void*))));
+            for (size_t i = 0; i < len; i++) {
+              uint8_t *base = ptr + i * 1;
+              (void) base;
+            }
+            free(ptr);
+          }
+        }
+        free(ptr1);
+      }
+      break;
+    }
+    case 1: {
+      if ((*((size_t*) (arg0 + (2*sizeof(void*))))) > 0) {
+        free(*((uint8_t **) (arg0 + sizeof(void*))));
+      }
+      break;
+    }
+  }
+}
+
+// Canonical ABI intrinsics
+
+__attribute__((__weak__, __export_name__("cabi_realloc")))
+void *cabi_realloc(void *ptr, size_t old_size, size_t align, size_t new_size) {
+  (void) old_size;
+  if (new_size == 0) return (void*) align;
+  void *ret = realloc(ptr, new_size);
+  if (!ret) abort();
+  return ret;
+}
+
+__attribute__((__aligned__(sizeof(void*))))
+static uint8_t RET_AREA[(3*sizeof(void*))];
+
+// Helper Functions
+
+void exports_chonkle_codec_transform_port_name_free(exports_chonkle_codec_transform_port_name_t *ptr) {
+  codec_string_free(ptr);
+}
+
+void codec_list_u8_free(codec_list_u8_t *ptr) {
+  size_t list_len = ptr->len;
+  if (list_len > 0) {
+    uint8_t *list_ptr = ptr->ptr;
+    for (size_t i = 0; i < list_len; i++) {
+    }
+    free(list_ptr);
+  }
+}
+
+void codec_tuple2_port_name_list_u8_free(codec_tuple2_port_name_list_u8_t *ptr) {
+  exports_chonkle_codec_transform_port_name_free(&ptr->f0);
+  codec_list_u8_free(&ptr->f1);
+}
+
+void exports_chonkle_codec_transform_port_map_free(exports_chonkle_codec_transform_port_map_t *ptr) {
+  size_t list_len = ptr->len;
+  if (list_len > 0) {
+    codec_tuple2_port_name_list_u8_t *list_ptr = ptr->ptr;
+    for (size_t i = 0; i < list_len; i++) {
+      codec_tuple2_port_name_list_u8_free(&list_ptr[i]);
+    }
+    free(list_ptr);
+  }
+}
+
+void exports_chonkle_codec_transform_result_port_map_string_free(exports_chonkle_codec_transform_result_port_map_string_t *ptr) {
+  if (!ptr->is_err) {
+    exports_chonkle_codec_transform_port_map_free(&ptr->val.ok);
+  } else {
+    codec_string_free(&ptr->val.err);
+  }
+}
+
+void codec_string_set(codec_string_t *ret, const char*s) {
+  ret->ptr = (uint8_t*) s;
+  ret->len = strlen(s);
+}
+
+void codec_string_dup(codec_string_t *ret, const char*s) {
+  ret->len = strlen(s);
+  ret->ptr = (uint8_t*) cabi_realloc(NULL, 0, 1, ret->len * 1);
+  memcpy(ret->ptr, s, ret->len * 1);
+}
+
+void codec_string_dup_n(codec_string_t *ret, const char*s, size_t len) {
+  ret->len = len;
+  ret->ptr = (uint8_t*) cabi_realloc(NULL, 0, 1, ret->len * 1);
+  memcpy(ret->ptr, s, ret->len * 1);
+}
+
+void codec_string_free(codec_string_t *ret) {
+  if (ret->len > 0) {
+    free(ret->ptr);
+  }
+  ret->ptr = NULL;
+  ret->len = 0;
+}
+
+// Component Adapters
+
+__attribute__((__export_name__("chonkle:codec/transform@0.1.0#encode")))
+uint8_t * __wasm_export_exports_chonkle_codec_transform_encode(uint8_t * arg, size_t arg0) {
+  exports_chonkle_codec_transform_port_map_t arg1 = (exports_chonkle_codec_transform_port_map_t) { (codec_tuple2_port_name_list_u8_t*)(arg), (arg0) };
+  exports_chonkle_codec_transform_result_port_map_string_t ret;
+  exports_chonkle_codec_transform_port_map_t ok;
+  codec_string_t err;
+  ret.is_err = !exports_chonkle_codec_transform_encode(&arg1, &ok, &err);
+  if (ret.is_err) {
+    ret.val.err = err;
+  }
+  if (!ret.is_err) {
+    ret.val.ok = ok;
+  }
+  uint8_t *ptr = (uint8_t *) &RET_AREA;
+  if ((ret).is_err) {
+    const codec_string_t *payload2 = &(ret).val.err;*((int8_t*)(ptr + 0)) = 1;
+    *((size_t*)(ptr + (2*sizeof(void*)))) = (*payload2).len;
+    *((uint8_t **)(ptr + sizeof(void*))) = (uint8_t *) (*payload2).ptr;
+  } else {
+    const exports_chonkle_codec_transform_port_map_t *payload = &(ret).val.ok;*((int8_t*)(ptr + 0)) = 0;
+    *((size_t*)(ptr + (2*sizeof(void*)))) = (*payload).len;
+    *((uint8_t **)(ptr + sizeof(void*))) = (uint8_t *) (*payload).ptr;
+  }
+  return ptr;
+}
+
+__attribute__((__export_name__("chonkle:codec/transform@0.1.0#decode")))
+uint8_t * __wasm_export_exports_chonkle_codec_transform_decode(uint8_t * arg, size_t arg0) {
+  exports_chonkle_codec_transform_port_map_t arg1 = (exports_chonkle_codec_transform_port_map_t) { (codec_tuple2_port_name_list_u8_t*)(arg), (arg0) };
+  exports_chonkle_codec_transform_result_port_map_string_t ret;
+  exports_chonkle_codec_transform_port_map_t ok;
+  codec_string_t err;
+  ret.is_err = !exports_chonkle_codec_transform_decode(&arg1, &ok, &err);
+  if (ret.is_err) {
+    ret.val.err = err;
+  }
+  if (!ret.is_err) {
+    ret.val.ok = ok;
+  }
+  uint8_t *ptr = (uint8_t *) &RET_AREA;
+  if ((ret).is_err) {
+    const codec_string_t *payload2 = &(ret).val.err;*((int8_t*)(ptr + 0)) = 1;
+    *((size_t*)(ptr + (2*sizeof(void*)))) = (*payload2).len;
+    *((uint8_t **)(ptr + sizeof(void*))) = (uint8_t *) (*payload2).ptr;
+  } else {
+    const exports_chonkle_codec_transform_port_map_t *payload = &(ret).val.ok;*((int8_t*)(ptr + 0)) = 0;
+    *((size_t*)(ptr + (2*sizeof(void*)))) = (*payload).len;
+    *((uint8_t **)(ptr + sizeof(void*))) = (uint8_t *) (*payload).ptr;
+  }
+  return ptr;
+}
+
+// Ensure that the *_component_type.o object is linked in
+
+extern void __component_type_object_force_link_codec(void);
+__attribute__((used))
+void __component_type_object_force_link_codec_public_use_in_this_compilation_unit(void) {
+  __component_type_object_force_link_codec();
+}
diff --git a/codec/shared/codec.h b/codec/shared/codec.h
new file mode 100644
index 0000000..11c3040
--- /dev/null
+++ b/codec/shared/codec.h
@@ -0,0 +1,78 @@
+// Generated by `wit-bindgen` 0.53.1. DO NOT EDIT!
+#ifndef __BINDINGS_CODEC_H
+#define __BINDINGS_CODEC_H
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdint.h>
+#include <stdbool.h>
+#include <stddef.h>
+
+typedef struct codec_string_t {
+  uint8_t*ptr;
+  size_t len;
+} codec_string_t;
+
+// A named byte buffer — one port in the port map.
+typedef codec_string_t exports_chonkle_codec_transform_port_name_t;
+
+typedef struct {
+  uint8_t *ptr;
+  size_t len;
+} codec_list_u8_t;
+
+typedef struct {
+  exports_chonkle_codec_transform_port_name_t f0;
+  codec_list_u8_t f1;
+} codec_tuple2_port_name_list_u8_t;
+
+typedef struct exports_chonkle_codec_transform_port_map_t {
+  codec_tuple2_port_name_list_u8_t   *ptr;
+  size_t len;
+} exports_chonkle_codec_transform_port_map_t;
+
+typedef struct {
+  bool is_err;
+  union {
+    exports_chonkle_codec_transform_port_map_t ok;
+    codec_string_t err;
+  } val;
+} exports_chonkle_codec_transform_result_port_map_string_t;
+
+// Exported Functions from `chonkle:codec/transform@0.1.0`
+bool exports_chonkle_codec_transform_encode(exports_chonkle_codec_transform_port_map_t *inputs, exports_chonkle_codec_transform_port_map_t *ret, codec_string_t *err);
+bool exports_chonkle_codec_transform_decode(exports_chonkle_codec_transform_port_map_t *inputs, exports_chonkle_codec_transform_port_map_t *ret, codec_string_t *err);
+
+// Helper Functions
+
+void exports_chonkle_codec_transform_port_name_free(exports_chonkle_codec_transform_port_name_t *ptr);
+
+void codec_list_u8_free(codec_list_u8_t *ptr);
+
+void codec_tuple2_port_name_list_u8_free(codec_tuple2_port_name_list_u8_t *ptr);
+
+void exports_chonkle_codec_transform_port_map_free(exports_chonkle_codec_transform_port_map_t *ptr);
+
+void exports_chonkle_codec_transform_result_port_map_string_free(exports_chonkle_codec_transform_result_port_map_string_t *ptr);
+
+// Sets the string `ret` to reference the input string `s` without copying it
+void codec_string_set(codec_string_t *ret, const char*s);
+
+// Creates a copy of the input nul-terminated string `s` and
+// stores it into the component model string `ret`.
+void codec_string_dup(codec_string_t *ret, const char*s);
+
+// Creates a copy of the input string `s` with length `len` and
+// stores it into the component model string `ret`.
+// The length is specified in code units (bytes for UTF-8, 16-bit values for UTF-16).
+void codec_string_dup_n(codec_string_t *ret, const char*s, size_t len);
+
+// Deallocates the string pointed to by `ret`, deallocating
+// the memory behind the string.
+void codec_string_free(codec_string_t *ret);
+
+#ifdef __cplusplus
+}
+#endif
+#endif
diff --git a/codec/shared/codec_component_type.o b/codec/shared/codec_component_type.o
new file mode 100644
index 0000000..98e46e0
Binary files /dev/null and b/codec/shared/codec_component_type.o differ
diff --git a/codec/shared/core_abi.c b/codec/shared/core_abi.c
new file mode 100644
index 0000000..76c4fd7
--- /dev/null
+++ b/codec/shared/core_abi.c
@@ -0,0 +1,111 @@
+/*
+ * Core ABI helpers for chonkle core wasm codecs.
+ *
+ * See core_abi.h for API documentation.
+ */
+#include "core_abi.h"
+
+#include <stdlib.h>
+#include <string.h>
+
+static uint32_t read_u32_le(const uint8_t *p) {
+    return (uint32_t)p[0]
+         | ((uint32_t)p[1] << 8)
+         | ((uint32_t)p[2] << 16)
+         | ((uint32_t)p[3] << 24);
+}
+
+static void write_u32_le(uint8_t *p, uint32_t v) {
+    p[0] = (uint8_t)(v);
+    p[1] = (uint8_t)(v >> 8);
+    p[2] = (uint8_t)(v >> 16);
+    p[3] = (uint8_t)(v >> 24);
+}
+
+core_abi_port_map_t core_abi_parse_port_map(const uint8_t *buf, uint32_t len) {
+    core_abi_port_map_t pm = {NULL, 0};
+    if (len < 4) return pm;
+
+    uint32_t count = read_u32_le(buf);
+    uint32_t offset = 4;
+
+    if (count == 0) { pm.count = 0; return pm; }
+
+    core_abi_port_t *entries = malloc(count * sizeof(core_abi_port_t));
+    if (!entries) return pm;
+
+    for (uint32_t i = 0; i < count; i++) {
+        if (offset + 4 > len) goto fail;
+        uint32_t name_len = read_u32_le(buf + offset);
+        offset += 4;
+
+        if (offset + name_len > len) goto fail;
+        entries[i].name = (const char *)(buf + offset);
+        entries[i].name_len = name_len;
+        offset += name_len;
+
+        if (offset + 4 > len) goto fail;
+        uint32_t data_len = read_u32_le(buf + offset);
+        offset += 4;
+
+        if (offset + data_len > len) goto fail;
+        entries[i].data = buf + offset;
+        entries[i].data_len = data_len;
+        offset += data_len;
+    }
+
+    pm.entries = entries;
+    pm.count = count;
+    return pm;
+
+fail:
+    free(entries);
+    return (core_abi_port_map_t){NULL, 0};
+}
+
+uint8_t *core_abi_serialize_port_map(
+    const core_abi_port_map_t *pm, uint32_t *out_len)
+{
+    uint32_t total = 4;
+    for (uint32_t i = 0; i < pm->count; i++)
+        total += 4 + pm->entries[i].name_len + 4 + pm->entries[i].data_len;
+
+    uint8_t *buf = malloc(total);
+    if (!buf) { *out_len = 0; return NULL; }
+
+    uint32_t offset = 0;
+    write_u32_le(buf + offset, pm->count);
+    offset += 4;
+
+    for (uint32_t i = 0; i < pm->count; i++) {
+        write_u32_le(buf + offset, pm->entries[i].name_len);
+        offset += 4;
+        memcpy(buf + offset, pm->entries[i].name, pm->entries[i].name_len);
+        offset += pm->entries[i].name_len;
+        write_u32_le(buf + offset, pm->entries[i].data_len);
+        offset += 4;
+        memcpy(buf + offset, pm->entries[i].data, pm->entries[i].data_len);
+        offset += pm->entries[i].data_len;
+    }
+
+    *out_len = total;
+    return buf;
+}
+
+const core_abi_port_t *core_abi_find_port(
+    const core_abi_port_map_t *pm, const char *name)
+{
+    uint32_t name_len = (uint32_t)strlen(name);
+    for (uint32_t i = 0; i < pm->count; i++) {
+        if (pm->entries[i].name_len == name_len &&
+            memcmp(pm->entries[i].name, name, name_len) == 0)
+            return &pm->entries[i];
+    }
+    return NULL;
+}
+
+void core_abi_free_port_map(core_abi_port_map_t *pm) {
+    free(pm->entries);
+    pm->entries = NULL;
+    pm->count = 0;
+}
diff --git a/codec/shared/core_abi.h b/codec/shared/core_abi.h
new file mode 100644
index 0000000..8f8769c
--- /dev/null
+++ b/codec/shared/core_abi.h
@@ -0,0 +1,68 @@
+/*
+ * Core ABI helpers for chonkle core wasm codecs.
+ *
+ * Provides port-map serialization/deserialization for the binary wire
+ * format defined in docs/reference/codec-contract/core.md.
+ */
+#ifndef CORE_ABI_H
+#define CORE_ABI_H
+
+#include <stddef.h>
+#include <stdint.h>
+
+/* A single port entry in a port-map. */
+typedef struct {
+    const char *name;
+    uint32_t name_len;
+    const uint8_t *data;
+    uint32_t data_len;
+} core_abi_port_t;
+
+/* A parsed port-map. */
+typedef struct {
+    core_abi_port_t *entries;
+    uint32_t count;
+} core_abi_port_map_t;
+
+/*
+ * Parse a serialized port-map from the wire format.
+ *
+ * The returned port-map's name/data pointers reference the input buffer
+ * directly (zero-copy parse). The caller must not free the input buffer
+ * while the port-map is in use. Free the entries array (not the data)
+ * with core_abi_free_port_map().
+ *
+ * Returns a port-map with count=0 on parse error.
+ */
+core_abi_port_map_t core_abi_parse_port_map(const uint8_t *buf, uint32_t len);
+
+/*
+ * Serialize a port-map to the wire format.
+ *
+ * Returns a malloc'd buffer. Caller takes ownership.
+ * Sets *out_len to the serialized length.
+ * Returns NULL on allocation failure.
+ */
+uint8_t *core_abi_serialize_port_map(
+    const core_abi_port_map_t *pm, uint32_t *out_len);
+
+/*
+ * Find a port by name. Returns NULL if not found.
+ */
+const core_abi_port_t *core_abi_find_port(
+    const core_abi_port_map_t *pm, const char *name);
+
+/*
+ * Free a parsed port-map (frees the entries array, not the referenced data).
+ */
+void core_abi_free_port_map(core_abi_port_map_t *pm);
+
+/* Pack a (ptr, len) pair into a single i64 return value. */
+static inline int64_t core_abi_pack_result(uint32_t ptr, uint32_t len) {
+    return ((int64_t)ptr << 32) | (int64_t)len;
+}
+
+/* Error return value: ptr=0, len=0. */
+#define CORE_ABI_ERROR ((int64_t)0)
+
+#endif /* CORE_ABI_H */
diff --git a/codec/tiff-predictor-2-c/.gitignore b/codec/tiff-predictor-2-c/.gitignore
new file mode 100644
index 0000000..c88eb55
--- /dev/null
+++ b/codec/tiff-predictor-2-c/.gitignore
@@ -0,0 +1,2 @@
+.zig-cache/
+/zig-out
diff --git a/codec/tiff-predictor-2-c/build.zig b/codec/tiff-predictor-2-c/build.zig
new file mode 100644
index 0000000..b0a9fc3
--- /dev/null
+++ b/codec/tiff-predictor-2-c/build.zig
@@ -0,0 +1,44 @@
+const std = @import("std");
+
+pub fn build(b: *std.Build) void {
+    const optimize = b.standardOptimizeOption(.{});
+
+    // Step 1: compile C sources to a core wasm32-wasi reactor module.
+    const compile = b.addSystemCommand(&.{
+        b.graph.zig_exe,
+        "cc",
+        "--target=wasm32-wasi",
+        "-mexec-model=reactor",
+        "-fvisibility=hidden",
+        "-lc",
+    });
+
+    switch (optimize) {
+        .Debug => compile.addArg("-Og"),
+        .ReleaseSafe, .ReleaseFast => compile.addArg("-O2"),
+        .ReleaseSmall => compile.addArg("-Os"),
+    }
+
+    compile.addArg("-o");
+    const core_wasm = compile.addOutputFileArg("tiff-predictor-2-core.wasm");
+    compile.addFileArg(b.path("../shared/codec.c"));
+    compile.addFileArg(b.path("tiff_predictor_2.c"));
+    compile.addFileArg(b.path("codec_impl.c"));
+    compile.addFileArg(b.path("../shared/codec_component_type.o"));
+
+    // Step 2: lift the core module to a Component Model .wasm via the WASI
+    // preview1 → preview2 adapter bundled in codec/wasi_snapshot_preview1.reactor.wasm.
+    const componentize = b.addSystemCommand(&.{ "wasm-tools", "component", "new" });
+    componentize.addFileArg(core_wasm);
+    componentize.addArg("--adapt");
+    componentize.addPrefixedFileArg(
+        "wasi_snapshot_preview1=",
+        b.path("../wasi_snapshot_preview1.reactor.wasm"),
+    );
+    componentize.addArg("-o");
+    const component = componentize.addOutputFileArg("tiff-predictor-2.wasm");
+
+    b.getInstallStep().dependOn(
+        &b.addInstallFile(component, "tiff-predictor-2.wasm").step,
+    );
+}
diff --git a/codec/tiff-predictor-2-c/codec_impl.c b/codec/tiff-predictor-2-c/codec_impl.c
new file mode 100644
index 0000000..db07318
--- /dev/null
+++ b/codec/tiff-predictor-2-c/codec_impl.c
@@ -0,0 +1,137 @@
+/**
+ * WIT entry points for the tiff-predictor-2 Component Model codec.
+ *
+ * Implements chonkle:codec/transform@0.1.0 — encode applies horizontal
+ * differencing, decode undoes it via cumulative sum.
+ *
+ * Expected port-map inputs:
+ *   "bytes"            list<u8>  raw pixel data (required)
+ *   "bytes_per_sample" list<u8>  UTF-8 integer: 1, 2, or 4 (required)
+ *   "width"            list<u8>  UTF-8 integer: samples per row (required)
+ *
+ * Output port-map:
+ *   "bytes"  list<u8>  transformed pixel data (same size as input)
+ */
+
+#include "../shared/codec.h"
+#include <stdlib.h>
+#include <string.h>
+
+/* Forward declarations from tiff_predictor_2.c */
+void diff_rows(uint8_t *buf, int width, int height, int bps);
+void cumsum_rows(uint8_t *buf, int width, int height, int bps);
+
+/*
+ * Override the weak cabi_realloc from codec.c so that OOM calls
+ * __builtin_trap() instead of abort(), avoiding WASI stderr imports.
+ */
+__attribute__((__export_name__("cabi_realloc")))
+void *cabi_realloc(void *ptr, size_t old_size, size_t align, size_t new_size) {
+    (void)old_size;
+    if (new_size == 0) return (void *)align;
+    void *ret = realloc(ptr, new_size);
+    if (!ret) __builtin_trap();
+    return ret;
+}
+
+/* Find a named port in the input port-map; returns NULL if absent. */
+static codec_list_u8_t *find_port(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    const char *name)
+{
+    size_t name_len = strlen(name);
+    for (size_t i = 0; i < inputs->len; i++) {
+        codec_tuple2_port_name_list_u8_t *e = &inputs->ptr[i];
+        if (e->f0.len == name_len && memcmp(e->f0.ptr, name, name_len) == 0)
+            return &e->f1;
+    }
+    return NULL;
+}
+
+/*
+ * Parse an unsigned integer from UTF-8 bytes (e.g. JSON constant "2" or "256").
+ * Ignores non-digit bytes; returns 0 for empty input.
+ */
+static int parse_int(const uint8_t *data, size_t len) {
+    int result = 0;
+    for (size_t i = 0; i < len; i++) {
+        if (data[i] >= '0' && data[i] <= '9')
+            result = result * 10 + (data[i] - '0');
+    }
+    return result;
+}
+
+/*
+ * Allocate a port-map with a single "bytes" port pointing to buf/buf_len.
+ * Ownership of buf transfers to the caller's port-map.
+ */
+static bool make_bytes_result(
+    uint8_t *buf, size_t buf_len,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    codec_tuple2_port_name_list_u8_t *entry = malloc(sizeof(*entry));
+    if (!entry) {
+        codec_string_set(err, "out of memory");
+        return false;
+    }
+    codec_string_dup(&entry->f0, "bytes");
+    entry->f1.ptr = buf;
+    entry->f1.len = buf_len;
+    ret->ptr = entry;
+    ret->len = 1;
+    return true;
+}
+
+static bool transform(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err,
+    int is_encode)
+{
+    codec_list_u8_t *bytes_port = find_port(inputs, "bytes");
+    codec_list_u8_t *bps_port   = find_port(inputs, "bytes_per_sample");
+    codec_list_u8_t *width_port = find_port(inputs, "width");
+
+    if (!bytes_port) { codec_string_set(err, "missing port: bytes");            return false; }
+    if (!bps_port)   { codec_string_set(err, "missing port: bytes_per_sample"); return false; }
+    if (!width_port) { codec_string_set(err, "missing port: width");            return false; }
+
+    int bps   = parse_int(bps_port->ptr,   bps_port->len);
+    int width = parse_int(width_port->ptr, width_port->len);
+
+    if (bps <= 0 || width <= 0) {
+        codec_string_set(err, "bytes_per_sample and width must be positive");
+        return false;
+    }
+
+    size_t input_len = bytes_port->len;
+    int height = (int)(input_len / (size_t)(width * bps));
+
+    uint8_t *out = malloc(input_len);
+    if (!out) { codec_string_set(err, "out of memory"); return false; }
+    memcpy(out, bytes_port->ptr, input_len);
+
+    if (is_encode)
+        diff_rows(out, width, height, bps);
+    else
+        cumsum_rows(out, width, height, bps);
+
+    return make_bytes_result(out, input_len, ret, err);
+}
+
+bool exports_chonkle_codec_transform_encode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    return transform(inputs, ret, err, 1);
+}
+
+bool exports_chonkle_codec_transform_decode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    return transform(inputs, ret, err, 0);
+}
diff --git a/codec/tiff-predictor-2-c/signature.json b/codec/tiff-predictor-2-c/signature.json
new file mode 100644
index 0000000..92b5b12
--- /dev/null
+++ b/codec/tiff-predictor-2-c/signature.json
@@ -0,0 +1,12 @@
+{
+  "codec_id": "tiff-predictor-2",
+  "implementation": "tiff-predictor-2-c",
+  "inputs": {
+    "bytes":            {"type": "bytes", "required": true},
+    "bytes_per_sample": {"type": "int",   "required": true},
+    "width":            {"type": "int",   "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
diff --git a/codec/tiff-predictor-2-c/tiff-predictor-2.wasm b/codec/tiff-predictor-2-c/tiff-predictor-2.wasm
new file mode 100644
index 0000000..34430f0
Binary files /dev/null and b/codec/tiff-predictor-2-c/tiff-predictor-2.wasm differ
diff --git a/codec/tiff-predictor-2-c/tiff_predictor_2.c b/codec/tiff-predictor-2-c/tiff_predictor_2.c
new file mode 100644
index 0000000..9fdafb6
--- /dev/null
+++ b/codec/tiff-predictor-2-c/tiff_predictor_2.c
@@ -0,0 +1,59 @@
+/**
+ * TIFF horizontal differencing predictor (Predictor=2), core algorithm.
+ *
+ * encode: row-wise differencing (diff_rows)
+ * decode: cumulative sum per row (cumsum_rows)
+ *
+ * Supports bytes-per-sample values of 1, 2, and 4.
+ * Both functions operate in-place on a caller-provided buffer.
+ */
+
+#include <stdint.h>
+
+/**
+ * Undo horizontal differencing in-place: cumulative sum per row.
+ * width is in samples; height is derived by the caller.
+ */
+void cumsum_rows(uint8_t *buf, int width, int height, int bps) {
+    for (int row = 0; row < height; row++) {
+        int base = row * width * bps;
+
+        if (bps == 1) {
+            uint8_t *s = buf + base;
+            for (int col = 1; col < width; col++)
+                s[col] += s[col - 1];
+        } else if (bps == 2) {
+            uint16_t *s = (uint16_t *)(buf + base);
+            for (int col = 1; col < width; col++)
+                s[col] += s[col - 1];
+        } else if (bps == 4) {
+            uint32_t *s = (uint32_t *)(buf + base);
+            for (int col = 1; col < width; col++)
+                s[col] += s[col - 1];
+        }
+    }
+}
+
+/**
+ * Apply horizontal differencing in-place: row-wise differences.
+ * Iterates backwards so each subtraction reads the original predecessor.
+ */
+void diff_rows(uint8_t *buf, int width, int height, int bps) {
+    for (int row = 0; row < height; row++) {
+        int base = row * width * bps;
+
+        if (bps == 1) {
+            uint8_t *s = buf + base;
+            for (int col = width - 1; col > 0; col--)
+                s[col] -= s[col - 1];
+        } else if (bps == 2) {
+            uint16_t *s = (uint16_t *)(buf + base);
+            for (int col = width - 1; col > 0; col--)
+                s[col] -= s[col - 1];
+        } else if (bps == 4) {
+            uint32_t *s = (uint32_t *)(buf + base);
+            for (int col = width - 1; col > 0; col--)
+                s[col] -= s[col - 1];
+        }
+    }
+}
diff --git a/codec/wasi_snapshot_preview1.reactor.wasm b/codec/wasi_snapshot_preview1.reactor.wasm
new file mode 100644
index 0000000..39a3270
Binary files /dev/null and b/codec/wasi_snapshot_preview1.reactor.wasm differ
diff --git a/codec/wit/codec.wit b/codec/wit/codec.wit
new file mode 100644
index 0000000..710afbf
--- /dev/null
+++ b/codec/wit/codec.wit
@@ -0,0 +1,14 @@
+package chonkle:codec@0.1.0;
+
+interface transform {
+    /// A named byte buffer — one port in the port map.
+    type port-name = string;
+    type port-map = list<tuple<port-name, list<u8>>>;
+
+    encode: func(inputs: port-map) -> result<port-map, string>;
+    decode: func(inputs: port-map) -> result<port-map, string>;
+}
+
+world codec {
+    export transform;
+}
diff --git a/codec/zlib-rs/.gitignore b/codec/zlib-rs/.gitignore
new file mode 100644
index 0000000..ea8c4bf
--- /dev/null
+++ b/codec/zlib-rs/.gitignore
@@ -0,0 +1 @@
+/target
diff --git a/codec/zlib-rs/Cargo.lock b/codec/zlib-rs/Cargo.lock
new file mode 100644
index 0000000..431d5e3
--- /dev/null
+++ b/codec/zlib-rs/Cargo.lock
@@ -0,0 +1,73 @@
+# This file is automatically @generated by Cargo.
+# It is not intended for manual editing.
+version = 4
+
+[[package]]
+name = "adler2"
+version = "2.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "320119579fcad9c21884f5c4861d16174d0e06250625266f50fe6898340abefa"
+
+[[package]]
+name = "bitflags"
+version = "2.11.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "843867be96c8daad0d758b57df9392b6d8d271134fce549de6ce169ff98a92af"
+
+[[package]]
+name = "cfg-if"
+version = "1.0.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9330f8b2ff13f34540b44e946ef35111825727b38d33286ef986142615121801"
+
+[[package]]
+name = "crc32fast"
+version = "1.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9481c1c90cbf2ac953f07c8d4a58aa3945c425b7185c9154d67a65e4230da511"
+dependencies = [
+ "cfg-if",
+]
+
+[[package]]
+name = "flate2"
+version = "1.1.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "843fba2746e448b37e26a819579957415c8cef339bf08564fe8b7ddbd959573c"
+dependencies = [
+ "crc32fast",
+ "miniz_oxide",
+]
+
+[[package]]
+name = "miniz_oxide"
+version = "0.8.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1fa76a2c86f704bdb222d66965fb3d63269ce38518b83cb0575fca855ebb6316"
+dependencies = [
+ "adler2",
+ "simd-adler32",
+]
+
+[[package]]
+name = "simd-adler32"
+version = "0.3.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e320a6c5ad31d271ad523dcf3ad13e2767ad8b1cb8f047f75a8aeaf8da139da2"
+
+[[package]]
+name = "wit-bindgen-rt"
+version = "0.44.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "653c85dd7aee6fe6f4bded0d242406deadae9819029ce6f7d258c920c384358a"
+dependencies = [
+ "bitflags",
+]
+
+[[package]]
+name = "zlib"
+version = "0.1.0"
+dependencies = [
+ "flate2",
+ "wit-bindgen-rt",
+]
diff --git a/codec/zlib-rs/Cargo.toml b/codec/zlib-rs/Cargo.toml
new file mode 100644
index 0000000..1aecd5b
--- /dev/null
+++ b/codec/zlib-rs/Cargo.toml
@@ -0,0 +1,16 @@
+[package]
+name = "zlib"
+version = "0.1.0"
+edition = "2024"
+
+[lib]
+crate-type = ["cdylib"]
+
+[dependencies]
+wit-bindgen-rt = { version = "0.44.0", features = ["bitflags"] }
+flate2 = { version = "1", default-features = false, features = ["rust_backend"] }
+
+[package.metadata.component]
+package = "chonkle:codec"
+
+[package.metadata.component.dependencies]
diff --git a/codec/zlib-rs/signature.json b/codec/zlib-rs/signature.json
new file mode 100644
index 0000000..ea9a817
--- /dev/null
+++ b/codec/zlib-rs/signature.json
@@ -0,0 +1,11 @@
+{
+  "codec_id": "zlib",
+  "implementation": "zlib-rs",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 6, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
diff --git a/codec/zlib-rs/src/bindings.rs b/codec/zlib-rs/src/bindings.rs
new file mode 100644
index 0000000..cea722e
--- /dev/null
+++ b/codec/zlib-rs/src/bindings.rs
@@ -0,0 +1,450 @@
+// Generated by `wit-bindgen` 0.41.0. DO NOT EDIT!
+// Options used:
+//   * runtime_path: "wit_bindgen_rt"
+#[rustfmt::skip]
+#[allow(dead_code, clippy::all)]
+pub mod exports {
+    pub mod chonkle {
+        pub mod codec {
+            #[allow(dead_code, async_fn_in_trait, unused_imports, clippy::all)]
+            pub mod transform {
+                #[used]
+                #[doc(hidden)]
+                static __FORCE_SECTION_REF: fn() = super::super::super::super::__link_custom_section_describing_imports;
+                use super::super::super::super::_rt;
+                /// A named byte buffer — one port in the port map.
+                pub type PortName = _rt::String;
+                pub type PortMap = _rt::Vec<(PortName, _rt::Vec<u8>)>;
+                #[doc(hidden)]
+                #[allow(non_snake_case)]
+                pub unsafe fn _export_encode_cabi<T: Guest>(
+                    arg0: *mut u8,
+                    arg1: usize,
+                ) -> *mut u8 {
+                    #[cfg(target_arch = "wasm32")] _rt::run_ctors_once();
+                    let base6 = arg0;
+                    let len6 = arg1;
+                    let mut result6 = _rt::Vec::with_capacity(len6);
+                    for i in 0..len6 {
+                        let base = base6
+                            .add(i * (4 * ::core::mem::size_of::<*const u8>()));
+                        let e6 = {
+                            let l0 = *base.add(0).cast::<*mut u8>();
+                            let l1 = *base
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            let len2 = l1;
+                            let bytes2 = _rt::Vec::from_raw_parts(l0.cast(), len2, len2);
+                            let l3 = *base
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>();
+                            let l4 = *base
+                                .add(3 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            let len5 = l4;
+                            (
+                                _rt::string_lift(bytes2),
+                                _rt::Vec::from_raw_parts(l3.cast(), len5, len5),
+                            )
+                        };
+                        result6.push(e6);
+                    }
+                    _rt::cabi_dealloc(
+                        base6,
+                        len6 * (4 * ::core::mem::size_of::<*const u8>()),
+                        ::core::mem::size_of::<*const u8>(),
+                    );
+                    let result7 = T::encode(result6);
+                    let ptr8 = (&raw mut _RET_AREA.0).cast::<u8>();
+                    match result7 {
+                        Ok(e) => {
+                            *ptr8.add(0).cast::<u8>() = (0i32) as u8;
+                            let vec12 = e;
+                            let len12 = vec12.len();
+                            let layout12 = _rt::alloc::Layout::from_size_align_unchecked(
+                                vec12.len() * (4 * ::core::mem::size_of::<*const u8>()),
+                                ::core::mem::size_of::<*const u8>(),
+                            );
+                            let result12 = if layout12.size() != 0 {
+                                let ptr = _rt::alloc::alloc(layout12).cast::<u8>();
+                                if ptr.is_null() {
+                                    _rt::alloc::handle_alloc_error(layout12);
+                                }
+                                ptr
+                            } else {
+                                ::core::ptr::null_mut()
+                            };
+                            for (i, e) in vec12.into_iter().enumerate() {
+                                let base = result12
+                                    .add(i * (4 * ::core::mem::size_of::<*const u8>()));
+                                {
+                                    let (t9_0, t9_1) = e;
+                                    let vec10 = (t9_0.into_bytes()).into_boxed_slice();
+                                    let ptr10 = vec10.as_ptr().cast::<u8>();
+                                    let len10 = vec10.len();
+                                    ::core::mem::forget(vec10);
+                                    *base
+                                        .add(::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>() = len10;
+                                    *base.add(0).cast::<*mut u8>() = ptr10.cast_mut();
+                                    let vec11 = (t9_1).into_boxed_slice();
+                                    let ptr11 = vec11.as_ptr().cast::<u8>();
+                                    let len11 = vec11.len();
+                                    ::core::mem::forget(vec11);
+                                    *base
+                                        .add(3 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>() = len11;
+                                    *base
+                                        .add(2 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<*mut u8>() = ptr11.cast_mut();
+                                }
+                            }
+                            *ptr8
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>() = len12;
+                            *ptr8
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>() = result12;
+                        }
+                        Err(e) => {
+                            *ptr8.add(0).cast::<u8>() = (1i32) as u8;
+                            let vec13 = (e.into_bytes()).into_boxed_slice();
+                            let ptr13 = vec13.as_ptr().cast::<u8>();
+                            let len13 = vec13.len();
+                            ::core::mem::forget(vec13);
+                            *ptr8
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>() = len13;
+                            *ptr8
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>() = ptr13.cast_mut();
+                        }
+                    };
+                    ptr8
+                }
+                #[doc(hidden)]
+                #[allow(non_snake_case)]
+                pub unsafe fn __post_return_encode<T: Guest>(arg0: *mut u8) {
+                    let l0 = i32::from(*arg0.add(0).cast::<u8>());
+                    match l0 {
+                        0 => {
+                            let l1 = *arg0
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>();
+                            let l2 = *arg0
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            let base8 = l1;
+                            let len8 = l2;
+                            for i in 0..len8 {
+                                let base = base8
+                                    .add(i * (4 * ::core::mem::size_of::<*const u8>()));
+                                {
+                                    let l3 = *base.add(0).cast::<*mut u8>();
+                                    let l4 = *base
+                                        .add(::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>();
+                                    _rt::cabi_dealloc(l3, l4, 1);
+                                    let l5 = *base
+                                        .add(2 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<*mut u8>();
+                                    let l6 = *base
+                                        .add(3 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>();
+                                    let base7 = l5;
+                                    let len7 = l6;
+                                    _rt::cabi_dealloc(base7, len7 * 1, 1);
+                                }
+                            }
+                            _rt::cabi_dealloc(
+                                base8,
+                                len8 * (4 * ::core::mem::size_of::<*const u8>()),
+                                ::core::mem::size_of::<*const u8>(),
+                            );
+                        }
+                        _ => {
+                            let l9 = *arg0
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>();
+                            let l10 = *arg0
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            _rt::cabi_dealloc(l9, l10, 1);
+                        }
+                    }
+                }
+                #[doc(hidden)]
+                #[allow(non_snake_case)]
+                pub unsafe fn _export_decode_cabi<T: Guest>(
+                    arg0: *mut u8,
+                    arg1: usize,
+                ) -> *mut u8 {
+                    #[cfg(target_arch = "wasm32")] _rt::run_ctors_once();
+                    let base6 = arg0;
+                    let len6 = arg1;
+                    let mut result6 = _rt::Vec::with_capacity(len6);
+                    for i in 0..len6 {
+                        let base = base6
+                            .add(i * (4 * ::core::mem::size_of::<*const u8>()));
+                        let e6 = {
+                            let l0 = *base.add(0).cast::<*mut u8>();
+                            let l1 = *base
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            let len2 = l1;
+                            let bytes2 = _rt::Vec::from_raw_parts(l0.cast(), len2, len2);
+                            let l3 = *base
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>();
+                            let l4 = *base
+                                .add(3 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            let len5 = l4;
+                            (
+                                _rt::string_lift(bytes2),
+                                _rt::Vec::from_raw_parts(l3.cast(), len5, len5),
+                            )
+                        };
+                        result6.push(e6);
+                    }
+                    _rt::cabi_dealloc(
+                        base6,
+                        len6 * (4 * ::core::mem::size_of::<*const u8>()),
+                        ::core::mem::size_of::<*const u8>(),
+                    );
+                    let result7 = T::decode(result6);
+                    let ptr8 = (&raw mut _RET_AREA.0).cast::<u8>();
+                    match result7 {
+                        Ok(e) => {
+                            *ptr8.add(0).cast::<u8>() = (0i32) as u8;
+                            let vec12 = e;
+                            let len12 = vec12.len();
+                            let layout12 = _rt::alloc::Layout::from_size_align_unchecked(
+                                vec12.len() * (4 * ::core::mem::size_of::<*const u8>()),
+                                ::core::mem::size_of::<*const u8>(),
+                            );
+                            let result12 = if layout12.size() != 0 {
+                                let ptr = _rt::alloc::alloc(layout12).cast::<u8>();
+                                if ptr.is_null() {
+                                    _rt::alloc::handle_alloc_error(layout12);
+                                }
+                                ptr
+                            } else {
+                                ::core::ptr::null_mut()
+                            };
+                            for (i, e) in vec12.into_iter().enumerate() {
+                                let base = result12
+                                    .add(i * (4 * ::core::mem::size_of::<*const u8>()));
+                                {
+                                    let (t9_0, t9_1) = e;
+                                    let vec10 = (t9_0.into_bytes()).into_boxed_slice();
+                                    let ptr10 = vec10.as_ptr().cast::<u8>();
+                                    let len10 = vec10.len();
+                                    ::core::mem::forget(vec10);
+                                    *base
+                                        .add(::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>() = len10;
+                                    *base.add(0).cast::<*mut u8>() = ptr10.cast_mut();
+                                    let vec11 = (t9_1).into_boxed_slice();
+                                    let ptr11 = vec11.as_ptr().cast::<u8>();
+                                    let len11 = vec11.len();
+                                    ::core::mem::forget(vec11);
+                                    *base
+                                        .add(3 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>() = len11;
+                                    *base
+                                        .add(2 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<*mut u8>() = ptr11.cast_mut();
+                                }
+                            }
+                            *ptr8
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>() = len12;
+                            *ptr8
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>() = result12;
+                        }
+                        Err(e) => {
+                            *ptr8.add(0).cast::<u8>() = (1i32) as u8;
+                            let vec13 = (e.into_bytes()).into_boxed_slice();
+                            let ptr13 = vec13.as_ptr().cast::<u8>();
+                            let len13 = vec13.len();
+                            ::core::mem::forget(vec13);
+                            *ptr8
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>() = len13;
+                            *ptr8
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>() = ptr13.cast_mut();
+                        }
+                    };
+                    ptr8
+                }
+                #[doc(hidden)]
+                #[allow(non_snake_case)]
+                pub unsafe fn __post_return_decode<T: Guest>(arg0: *mut u8) {
+                    let l0 = i32::from(*arg0.add(0).cast::<u8>());
+                    match l0 {
+                        0 => {
+                            let l1 = *arg0
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>();
+                            let l2 = *arg0
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            let base8 = l1;
+                            let len8 = l2;
+                            for i in 0..len8 {
+                                let base = base8
+                                    .add(i * (4 * ::core::mem::size_of::<*const u8>()));
+                                {
+                                    let l3 = *base.add(0).cast::<*mut u8>();
+                                    let l4 = *base
+                                        .add(::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>();
+                                    _rt::cabi_dealloc(l3, l4, 1);
+                                    let l5 = *base
+                                        .add(2 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<*mut u8>();
+                                    let l6 = *base
+                                        .add(3 * ::core::mem::size_of::<*const u8>())
+                                        .cast::<usize>();
+                                    let base7 = l5;
+                                    let len7 = l6;
+                                    _rt::cabi_dealloc(base7, len7 * 1, 1);
+                                }
+                            }
+                            _rt::cabi_dealloc(
+                                base8,
+                                len8 * (4 * ::core::mem::size_of::<*const u8>()),
+                                ::core::mem::size_of::<*const u8>(),
+                            );
+                        }
+                        _ => {
+                            let l9 = *arg0
+                                .add(::core::mem::size_of::<*const u8>())
+                                .cast::<*mut u8>();
+                            let l10 = *arg0
+                                .add(2 * ::core::mem::size_of::<*const u8>())
+                                .cast::<usize>();
+                            _rt::cabi_dealloc(l9, l10, 1);
+                        }
+                    }
+                }
+                pub trait Guest {
+                    fn encode(inputs: PortMap) -> Result<PortMap, _rt::String>;
+                    fn decode(inputs: PortMap) -> Result<PortMap, _rt::String>;
+                }
+                #[doc(hidden)]
+                macro_rules! __export_chonkle_codec_transform_0_1_0_cabi {
+                    ($ty:ident with_types_in $($path_to_types:tt)*) => {
+                        const _ : () = { #[unsafe (export_name =
+                        "chonkle:codec/transform@0.1.0#encode")] unsafe extern "C" fn
+                        export_encode(arg0 : * mut u8, arg1 : usize,) -> * mut u8 {
+                        unsafe { $($path_to_types)*:: _export_encode_cabi::<$ty > (arg0,
+                        arg1) } } #[unsafe (export_name =
+                        "cabi_post_chonkle:codec/transform@0.1.0#encode")] unsafe extern
+                        "C" fn _post_return_encode(arg0 : * mut u8,) { unsafe {
+                        $($path_to_types)*:: __post_return_encode::<$ty > (arg0) } }
+                        #[unsafe (export_name = "chonkle:codec/transform@0.1.0#decode")]
+                        unsafe extern "C" fn export_decode(arg0 : * mut u8, arg1 :
+                        usize,) -> * mut u8 { unsafe { $($path_to_types)*::
+                        _export_decode_cabi::<$ty > (arg0, arg1) } } #[unsafe
+                        (export_name = "cabi_post_chonkle:codec/transform@0.1.0#decode")]
+                        unsafe extern "C" fn _post_return_decode(arg0 : * mut u8,) {
+                        unsafe { $($path_to_types)*:: __post_return_decode::<$ty > (arg0)
+                        } } };
+                    };
+                }
+                #[doc(hidden)]
+                pub(crate) use __export_chonkle_codec_transform_0_1_0_cabi;
+                #[cfg_attr(target_pointer_width = "64", repr(align(8)))]
+                #[cfg_attr(target_pointer_width = "32", repr(align(4)))]
+                struct _RetArea(
+                    [::core::mem::MaybeUninit<
+                        u8,
+                    >; 3 * ::core::mem::size_of::<*const u8>()],
+                );
+                static mut _RET_AREA: _RetArea = _RetArea(
+                    [::core::mem::MaybeUninit::uninit(); 3
+                        * ::core::mem::size_of::<*const u8>()],
+                );
+            }
+        }
+    }
+}
+#[rustfmt::skip]
+mod _rt {
+    #![allow(dead_code, clippy::all)]
+    pub use alloc_crate::string::String;
+    pub use alloc_crate::vec::Vec;
+    #[cfg(target_arch = "wasm32")]
+    pub fn run_ctors_once() {
+        wit_bindgen_rt::run_ctors_once();
+    }
+    pub unsafe fn string_lift(bytes: Vec<u8>) -> String {
+        if cfg!(debug_assertions) {
+            String::from_utf8(bytes).unwrap()
+        } else {
+            String::from_utf8_unchecked(bytes)
+        }
+    }
+    pub unsafe fn cabi_dealloc(ptr: *mut u8, size: usize, align: usize) {
+        if size == 0 {
+            return;
+        }
+        let layout = alloc::Layout::from_size_align_unchecked(size, align);
+        alloc::dealloc(ptr, layout);
+    }
+    pub use alloc_crate::alloc;
+    extern crate alloc as alloc_crate;
+}
+/// Generates `#[unsafe(no_mangle)]` functions to export the specified type as
+/// the root implementation of all generated traits.
+///
+/// For more information see the documentation of `wit_bindgen::generate!`.
+///
+/// ```rust
+/// # macro_rules! export{ ($($t:tt)*) => (); }
+/// # trait Guest {}
+/// struct MyType;
+///
+/// impl Guest for MyType {
+///     // ...
+/// }
+///
+/// export!(MyType);
+/// ```
+#[allow(unused_macros)]
+#[doc(hidden)]
+macro_rules! __export_codec_impl {
+    ($ty:ident) => {
+        self::export!($ty with_types_in self);
+    };
+    ($ty:ident with_types_in $($path_to_types_root:tt)*) => {
+        $($path_to_types_root)*::
+        exports::chonkle::codec::transform::__export_chonkle_codec_transform_0_1_0_cabi!($ty
+        with_types_in $($path_to_types_root)*:: exports::chonkle::codec::transform);
+    };
+}
+#[doc(inline)]
+pub(crate) use __export_codec_impl as export;
+#[cfg(target_arch = "wasm32")]
+#[unsafe(
+    link_section = "component-type:wit-bindgen:0.41.0:chonkle:codec@0.1.0:codec:encoded world"
+)]
+#[doc(hidden)]
+#[allow(clippy::octal_escapes)]
+pub static __WIT_BINDGEN_COMPONENT_TYPE: [u8; 280] = *b"\
+\0asm\x0d\0\x01\0\0\x19\x16wit-component-encoding\x04\0\x07\x9c\x01\x01A\x02\x01\
+A\x02\x01B\x0a\x01s\x04\0\x09port-name\x03\0\0\x01p}\x01o\x02\x01\x02\x01p\x03\x04\
+\0\x08port-map\x03\0\x04\x01j\x01\x05\x01s\x01@\x01\x06inputs\x05\0\x06\x04\0\x06\
+encode\x01\x07\x04\0\x06decode\x01\x07\x04\0\x1dchonkle:codec/transform@0.1.0\x05\
+\0\x04\0\x19chonkle:codec/codec@0.1.0\x04\0\x0b\x0b\x01\0\x05codec\x03\0\0\0G\x09\
+producers\x01\x0cprocessed-by\x02\x0dwit-component\x070.227.1\x10wit-bindgen-rus\
+t\x060.41.0";
+#[inline(never)]
+#[doc(hidden)]
+pub fn __link_custom_section_describing_imports() {
+    wit_bindgen_rt::maybe_link_cabi_realloc();
+}
diff --git a/codec/zlib-rs/src/lib.rs b/codec/zlib-rs/src/lib.rs
new file mode 100644
index 0000000..3fc7533
--- /dev/null
+++ b/codec/zlib-rs/src/lib.rs
@@ -0,0 +1,46 @@
+#[allow(warnings)]
+mod bindings;
+
+use bindings::exports::chonkle::codec::transform::{Guest, PortMap};
+use flate2::read::ZlibDecoder;
+use flate2::write::ZlibEncoder;
+use flate2::Compression;
+use std::io::{Read, Write};
+
+struct Component;
+
+fn find_port<'a>(inputs: &'a PortMap, name: &str) -> Option<&'a Vec<u8>> {
+    inputs.iter().find(|(k, _)| k == name).map(|(_, v)| v)
+}
+
+impl Guest for Component {
+    fn encode(inputs: PortMap) -> Result<PortMap, String> {
+        let bytes = find_port(&inputs, "bytes")
+            .ok_or_else(|| "missing port: bytes".to_string())?;
+
+        let level = find_port(&inputs, "level")
+            .and_then(|v| std::str::from_utf8(v).ok())
+            .and_then(|s| s.trim().parse::<u32>().ok())
+            .unwrap_or(6)
+            .min(9);
+
+        let mut enc = ZlibEncoder::new(Vec::new(), Compression::new(level));
+        enc.write_all(bytes).map_err(|e| e.to_string())?;
+        let compressed = enc.finish().map_err(|e| e.to_string())?;
+
+        Ok(vec![("bytes".to_string(), compressed)])
+    }
+
+    fn decode(inputs: PortMap) -> Result<PortMap, String> {
+        let bytes = find_port(&inputs, "bytes")
+            .ok_or_else(|| "missing port: bytes".to_string())?;
+
+        let mut dec = ZlibDecoder::new(bytes.as_slice());
+        let mut out = Vec::new();
+        dec.read_to_end(&mut out).map_err(|e| e.to_string())?;
+
+        Ok(vec![("bytes".to_string(), out)])
+    }
+}
+
+bindings::export!(Component with_types_in bindings);
diff --git a/codec/zlib-rs/wit/world.wit b/codec/zlib-rs/wit/world.wit
new file mode 100644
index 0000000..710afbf
--- /dev/null
+++ b/codec/zlib-rs/wit/world.wit
@@ -0,0 +1,14 @@
+package chonkle:codec@0.1.0;
+
+interface transform {
+    /// A named byte buffer — one port in the port map.
+    type port-name = string;
+    type port-map = list<tuple<port-name, list<u8>>>;
+
+    encode: func(inputs: port-map) -> result<port-map, string>;
+    decode: func(inputs: port-map) -> result<port-map, string>;
+}
+
+world codec {
+    export transform;
+}
diff --git a/codec/zlib-rs/zlib.wasm b/codec/zlib-rs/zlib.wasm
new file mode 100644
index 0000000..288665f
Binary files /dev/null and b/codec/zlib-rs/zlib.wasm differ
diff --git a/demo/chonkle-pipeline.ipynb b/demo/chonkle-pipeline.ipynb
index ce10f98..72abb8b 100644
--- a/demo/chonkle-pipeline.ipynb
+++ b/demo/chonkle-pipeline.ipynb
@@ -4,16 +4,11 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Chonkle: Codec Pipelines with WebAssembly\n",
+    "# Chonkle: Codec Pipelines\n",
     "\n",
-    "**chonkle** is a codec pipeline library for encoding and decoding chunked array data. It is capable of mixing standard Python codecs (via numcodecs) with custom WebAssembly codecs.\n",
+    "**chonkle** is a codec pipeline library for encoding and decoding chunked data. Pipelines are directed acyclic graphs (DAGs) of codec steps. Each step can use a different backend: native Python (via numcodecs), core Wasm, or Component Model Wasm.\n",
     "\n",
-    "This notebook first demonstrates a round-trip decode/encode of a COG tile (chunk) using Python codecs, then replaces one of the Python codecs with one written in C and compiled to Wasm. Why Wasm?\n",
-    "\n",
-    "1. **Performant** — codecs can be written in languages like C or Rust and run at near-native speed\n",
-    "2. **Portable** — distribute a single .wasm file for any platform\n",
-    "3. **Safe** - arbitrary codec code runs sandboxed and cannot corrupt the host process\n",
-    "4. **Distributable** — a plain file, hostable on any file server or artifact registry"
+    "This notebook demonstrates a round-trip decode/encode of a COG tile using a two-step pipeline where the zlib step runs as a native Python codec and the TIFF predictor step runs as a Wasm codec. The same codec can ship as either a Core Wasm module or a Component Model Wasm component — chonkle handles both transparently."
    ]
   },
   {
@@ -29,7 +24,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -39,7 +34,9 @@
     "import numpy as np\n",
     "from chunk_utils import cog_metadata, describe_bytes, download_cog, extract_tile\n",
     "\n",
-    "from chonkle import decode, encode"
+    "from chonkle.executor import run\n",
+    "from chonkle.pipeline import prepare\n",
+    "from chonkle.resolver import Resolver"
    ]
   },
   {
@@ -53,14 +50,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 30,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Downloaded data/B04.tif\n"
+      "Already downloaded: data/B04.tif\n"
      ]
     }
    ],
@@ -88,7 +85,7 @@
    "source": [
     "### Inspect the COG metadata\n",
     "\n",
-    "We can inspect the downloaded image metadata for information we'll need to build a JSON structure that defines the codec pipeline. Key tags to note:\n",
+    "We can inspect the downloaded image metadata for information we'll need to build the pipeline definition. Key tags to note:\n",
     "\n",
     "- **Compression**: 8 = DEFLATE\n",
     "- **Predictor**: 2 = horizontal differencing\n",
@@ -98,7 +95,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 31,
+   "execution_count": 3,
    "metadata": {},
    "outputs": [
     {
@@ -147,12 +144,12 @@
    "source": [
     "### Extract a tile (chunk)\n",
     "\n",
-    "Next, let's extract the raw bytes of tile 0 (the first 1024×1024 pixel block) to use as our test chunk. We'll use the SHA-256 hash as our reference for verifying round-trip (decode then encode) fidelity."
+    "Next, let's extract the raw bytes of tile 0 (the first 1024×1024 pixel block) to use as our test chunk. We'll use the SHA-256 hash as our reference for verifying round-trip fidelity."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 32,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [
     {
@@ -172,238 +169,172 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Python codecs\n",
+    "## Pipeline definition\n",
     "\n",
-    "Now that we have our test chunk, we can can decode and re-encode the chunk using a pipeline built entirely from Python codecs either native to — or registered with — numcodecs. This establishes is our baseline pipeline that we will eventually modify to use a Wasm codec."
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "### Pipeline definition\n",
+    "A chonkle pipeline is a JSON document that defines a DAG of codec steps. Each step has a `codec_id` and a set of named input ports wired to pipeline inputs, constants, or outputs from other steps. The wiring reference format is:\n",
     "\n",
-    "A chonkle codec pipeline is a modified form of [Zarr v3's codec metadata](https://zarr-specs.readthedocs.io/en/latest/v3/core/index.html#metadata) — a JSON list where each entry has a `name` and `configuration`. chonkle adds a `type` field to distinguish between `\"numcodecs\"` (Python codecs) and `\"wasm\"` (WebAssembly codecs), and an optional `uri` field for locating Wasm modules.\n",
+    "- `input.<name>` — a pipeline-level input\n",
+    "- `constant.<name>` — a named constant (JSON-encoded as bytes at runtime)\n",
+    "- `<step_name>.<port>` — an output port from another step\n",
     "\n",
-    "The list describes the **encoding** order; decoding applies the codecs in reverse.\n",
+    "The pipeline declares a `direction` (\"encode\" or \"decode\") indicating which direction the DAG was authored in. The same pipeline supports both directions: running in the authored direction executes steps in topological order, while running in the opposite direction inverts the DAG (reversed order, opposite codec function).\n",
     "\n",
-    "For this COG tile the encoding pipeline is:\n",
+    "For this COG tile, the decode pipeline is:\n",
     "\n",
-    "1. `tiff_predictor_2` — horizontal differencing (numpy → numpy)\n",
-    "2. `bytes` — serialize array to raw bytes (numpy → bytes)\n",
-    "3. `zlib` — DEFLATE compression at level 9 (bytes → bytes)\n",
+    "1. **zlib** — DEFLATE decompression (bytes → bytes)\n",
+    "2. **predictor2** — undo TIFF horizontal differencing (bytes → bytes)\n",
     "\n",
-    "`zlib` is a built-in numcodecs codec. `tiff_predictor_2` and `bytes` are custom `Codec` subclasses that chonkle registers with numcodecs at import time."
+    "Codec parameters (`level`, `bytes_per_sample`, `width`) are declared as pipeline constants and wired to codec input ports."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 33,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
-    "codecs_numcodecs = [\n",
-    "    {\n",
-    "        \"name\": \"tiff_predictor_2\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {},\n",
+    "wasm_path = Path(\"tiff-predictor-2-c.wasm\").resolve()\n",
+    "\n",
+    "pipeline = {\n",
+    "    \"codec_id\": \"cog-zlib-predictor2\",\n",
+    "    \"direction\": \"decode\",\n",
+    "    \"inputs\": {\"bytes\": {\"type\": \"bytes\"}},\n",
+    "    \"constants\": {\n",
+    "        \"bytes_per_sample\": {\"type\": \"int\", \"value\": 2},\n",
+    "        \"width\": {\"type\": \"int\", \"value\": 1024},\n",
+    "        \"level\": {\"type\": \"int\", \"value\": 9},\n",
     "    },\n",
-    "    {\n",
-    "        \"name\": \"bytes\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\n",
-    "            \"endian\": \"little\",\n",
-    "            \"data_type\": \"uint16\",\n",
-    "            \"shape\": [1024, 1024],\n",
+    "    \"sources\": {\"tiff-predictor-2\": wasm_path.as_uri()},\n",
+    "    \"outputs\": {\"bytes\": \"predictor2.bytes\"},\n",
+    "    \"steps\": {\n",
+    "        \"zlib\": {\n",
+    "            \"codec_id\": \"zlib\",\n",
+    "            \"inputs\": {\n",
+    "                \"bytes\": \"input.bytes\",\n",
+    "                \"level\": \"constant.level\",\n",
+    "            },\n",
+    "        },\n",
+    "        \"predictor2\": {\n",
+    "            \"codec_id\": \"tiff-predictor-2\",\n",
+    "            \"inputs\": {\n",
+    "                \"bytes\": \"zlib.bytes\",\n",
+    "                \"bytes_per_sample\": \"constant.bytes_per_sample\",\n",
+    "                \"width\": \"constant.width\",\n",
+    "            },\n",
     "        },\n",
     "    },\n",
-    "    {\n",
-    "        \"name\": \"zlib\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\"level\": 9},\n",
-    "    },\n",
-    "]"
+    "}"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "### Decode\n",
+    "### Embedded codec signatures\n",
     "\n",
-    "Decoding applies the pipeline in reverse: decompress, deserialize, then undo the predictor. This reverse order is what is needed when we want to decode a chunk loaded from disk or fetched over the network. In our case, the chunk's raw bytes are loaded into the variable `tile_bytes`."
+    "Every chonkle `.wasm` binary carries a `chonkle:signature` custom section with the codec's metadata: its `codec_id`, `implementation` name, and typed input/output port descriptors. The `detect_codec_type` function reads the binary header to distinguish Core Wasm modules from Component Model components."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 34,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "shape: (1024, 1024)  dtype: uint16  min: 0  max: 10552\n"
+      "Backend: component\n",
+      "\n",
+      "{\n",
+      "  \"codec_id\": \"tiff-predictor-2\",\n",
+      "  \"implementation\": \"tiff-predictor-2-c\",\n",
+      "  \"inputs\": {\n",
+      "    \"bytes\": {\n",
+      "      \"type\": \"bytes\",\n",
+      "      \"required\": true\n",
+      "    },\n",
+      "    \"bytes_per_sample\": {\n",
+      "      \"type\": \"int\",\n",
+      "      \"required\": true\n",
+      "    },\n",
+      "    \"width\": {\n",
+      "      \"type\": \"int\",\n",
+      "      \"required\": true\n",
+      "    }\n",
+      "  },\n",
+      "  \"outputs\": {\n",
+      "    \"bytes\": {\n",
+      "      \"type\": \"bytes\"\n",
+      "    }\n",
+      "  }\n",
+      "}\n"
      ]
     }
    ],
    "source": [
-    "arr = decode(chunk_bytes, codecs_numcodecs)\n",
-    "print(f\"shape: {arr.shape}  dtype: {arr.dtype}  min: {arr.min()}  max: {arr.max()}\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "To make the result more tangible, let's plot the numpy array (the decoded chunk) as a 2D image using matplotlib."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 35,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "image/png": "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",
-      "text/plain": [
-       "<Figure size 600x600 with 2 Axes>"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    }
-   ],
-   "source": [
-    "fig, ax = plt.subplots(figsize=(6, 6))\n",
-    "vmin, vmax = np.percentile(arr[arr > 0], [2, 98])\n",
-    "im = ax.imshow(arr, cmap=\"viridis\", vmin=vmin, vmax=vmax)\n",
-    "ax.set_title(\"Decoded tile (numcodecs)\")\n",
-    "fig.colorbar(im, ax=ax, shrink=0.8)\n",
-    "plt.show()"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "### Encode and verify round trip\n",
+    "import json\n",
     "\n",
-    "Encoding the decoded array back through the same pipeline should produce bytes\n",
-    "identical to the original tile."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 36,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "size: 1.381 MiB | sha256: 2c02e7e60074d6767ccb4c44de2da249d331fd82e107431e41cfe4069bae0d62\n",
-      "Round trip: OK\n"
-     ]
-    }
-   ],
-   "source": [
-    "encoded = encode(arr, codecs_numcodecs)\n",
-    "describe_bytes(encoded)\n",
-    "assert encoded == chunk_bytes, \"Round-trip mismatch!\"\n",
-    "print(\"Round trip: OK\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## Wasm codecs\n",
-    "\n",
-    "One of the goals of chonkle is to demonstrate running codecs compiled to WebAssembly, which we will do here by replacing the numcodecs `tiff_predictor_2` codec in the pipeline with a Wasm codec written in C. Since the Wasm codec operates on **raw bytes** rather than numpy arrays, the `bytes` codec must appear **before** the Wasm `tiff_predictor_2` in the pipeline (in encoding order). The Wasm codec also needs explicit configuration for `bytes_per_sample` and `width`.\n",
+    "from chonkle.wasm_signature import detect_codec_type, read_signature\n",
     "\n",
-    "An additional `uri` field tells chonkle where to find the `.wasm` module. Supported schemes are `file://`, `https://`, and `oci://`. We'll start with a local module, then demonstrate loading the same module from an HTTPS URL and an OCI registry."
+    "print(f\"Backend: {detect_codec_type(wasm_path).value}\\n\")\n",
+    "print(json.dumps(read_signature(wasm_path), indent=2))"
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": 37,
+   "cell_type": "markdown",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "wasm_path = Path(\"tiff-predictor-2-c.wasm\").resolve()\n",
+    "### Codec resolution\n",
     "\n",
-    "codecs_wasm_local = [\n",
-    "    {\n",
-    "        \"name\": \"bytes\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\n",
-    "            \"endian\": \"little\",\n",
-    "            \"data_type\": \"uint16\",\n",
-    "            \"shape\": [1024, 1024],\n",
-    "        },\n",
-    "    },\n",
-    "    {\n",
-    "        \"name\": \"tiff_predictor_2\",\n",
-    "        \"type\": \"wasm\",\n",
-    "        \"uri\": wasm_path.as_uri(),\n",
-    "        \"configuration\": {\"bytes_per_sample\": 2, \"width\": 1024},\n",
-    "    },\n",
-    "    {\n",
-    "        \"name\": \"zlib\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\"level\": 9},\n",
-    "    },\n",
-    "]"
+    "The `Resolver` maps each step's `codec_id` to an implementation. It checks a local codec store and a set of bundled native (numcodecs) codecs, selecting among available backends using a configurable preference order (default: native > core Wasm > Component Model Wasm). The pipeline's `sources` field provides download URIs for codecs not already available locally."
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "### Local (`file://`)\n",
+    "### Decode\n",
     "\n",
-    "We'll decode the same chunk using the Wasm codec loaded from a local file and verify that the result is identical to the array produced by the Python codec pipeline."
+    "To decode, we prepare the pipeline for the \"decode\" direction, then run it with the raw tile bytes as input. Since the pipeline was authored for \"decode\", this runs the steps in their declared order. The result is `dict[str, bytes]` — we convert to a numpy array afterward."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 38,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "shape: (1024, 1024)  dtype: uint16  min: 0  max: 10552\n",
-      "Decoded arrays match: OK\n"
+      "shape: (1024, 1024)  dtype: uint16  min: 0  max: 10552\n"
      ]
     }
    ],
    "source": [
-    "arr_wasm = decode(chunk_bytes, codecs_wasm_local)\n",
-    "print(f\"shape: {arr_wasm.shape}  dtype: {arr_wasm.dtype}  min: {arr_wasm.min()}  max: {arr_wasm.max()}\")\n",
+    "prepared = prepare(pipeline, \"decode\")\n",
+    "result = run(prepared, {\"bytes\": chunk_bytes})\n",
     "\n",
-    "np.testing.assert_array_equal(arr_wasm, arr)\n",
-    "print(\"Decoded arrays match: OK\")"
+    "arr = np.frombuffer(result[\"bytes\"], dtype=np.uint16).reshape(1024, 1024)\n",
+    "print(f\"shape: {arr.shape}  dtype: {arr.dtype}  min: {arr.min()}  max: {arr.max()}\")"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "As before, we'll plot the decoded array to provide a tangible check that the decoded data is equivalent to what we saw before."
+    "Let's plot the decoded chunk as a 2D image."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 39,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [
     {
      "data": {
-      "image/png": "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",
+      "image/png": "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",
       "text/plain": [
        "<Figure size 600x600 with 2 Axes>"
       ]
@@ -414,9 +345,9 @@
    ],
    "source": [
     "fig, ax = plt.subplots(figsize=(6, 6))\n",
-    "vmin, vmax = np.percentile(arr_wasm[arr_wasm > 0], [2, 98])\n",
-    "im = ax.imshow(arr_wasm, cmap=\"viridis\", vmin=vmin, vmax=vmax)\n",
-    "ax.set_title(\"Decoded tile (Wasm file://)\")\n",
+    "vmin, vmax = np.percentile(arr[arr > 0], [2, 98])\n",
+    "im = ax.imshow(arr, cmap=\"viridis\", vmin=vmin, vmax=vmax)\n",
+    "ax.set_title(\"Decoded tile\")\n",
     "fig.colorbar(im, ax=ax, shrink=0.8)\n",
     "plt.show()"
    ]
@@ -425,12 +356,14 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Finally, we'll re-encode the array using the Wasm codec and verify that the resulting bytes are identical to the original chunk we extracted from the COG tile."
+    "### Encode and verify round trip\n",
+    "\n",
+    "Encoding uses the same pipeline definition but prepared for the \"encode\" direction. Since the pipeline was authored for \"decode\", this inverts the execution: steps run in reversed order and each codec's `encode` function is called instead of `decode`. The result should be identical to the original tile bytes."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 40,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [
     {
@@ -443,9 +376,11 @@
     }
    ],
    "source": [
-    "encoded_wasm = encode(arr_wasm, codecs_wasm_local)\n",
-    "describe_bytes(encoded_wasm)\n",
-    "assert encoded_wasm == chunk_bytes, \"Round-trip mismatch!\"\n",
+    "enc_prepared = prepare(pipeline, \"encode\")\n",
+    "encoded = run(enc_prepared, {\"bytes\": result[\"bytes\"]})\n",
+    "\n",
+    "describe_bytes(encoded[\"bytes\"])\n",
+    "assert encoded[\"bytes\"] == chunk_bytes, \"Round-trip mismatch!\"\n",
     "print(\"Round trip: OK\")"
    ]
   },
@@ -453,58 +388,46 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "### HTTPS\n",
+    "## Remote Wasm sources\n",
     "\n",
-    "The same module can be fetched from an HTTPS URL — just swap the `uri` in the pipeline definition. chonkle downloads and caches it automatically."
+    "Wasm codecs can be fetched from HTTPS URLs or OCI registries. Here we download the same `tiff-predictor-2` codec from each and verify that the decoded output matches."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### HTTPS"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 41,
+   "execution_count": 10,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Decoded arrays match: OK\n",
-      "size: 1.381 MiB | sha256: 2c02e7e60074d6767ccb4c44de2da249d331fd82e107431e41cfe4069bae0d62\n",
-      "Round trip: OK\n"
+      "Decoded bytes match: OK\n"
      ]
     }
    ],
    "source": [
-    "codecs_wasm_https = [\n",
-    "    {\n",
-    "        \"name\": \"bytes\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\n",
-    "            \"endian\": \"little\",\n",
-    "            \"data_type\": \"uint16\",\n",
-    "            \"shape\": [1024, 1024],\n",
-    "        },\n",
-    "    },\n",
-    "    {\n",
-    "        \"name\": \"tiff_predictor_2\",\n",
-    "        \"type\": \"wasm\",\n",
-    "        \"uri\": \"https://github.com/cylf-dev/tiff-predictor-2-c/releases/download/v0.1.0/tiff-predictor-2.wasm\",\n",
-    "        \"configuration\": {\"bytes_per_sample\": 2, \"width\": 1024},\n",
-    "    },\n",
-    "    {\n",
-    "        \"name\": \"zlib\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\"level\": 9},\n",
+    "resolver_https = Resolver(\n",
+    "    force_sources={\n",
+    "        \"tiff-predictor-2\": (\n",
+    "            \"https://github.com/cylf-dev/tiff-predictor-2-c/\"\n",
+    "            \"releases/download/v0.3.0/tiff-predictor-2-c.wasm\"\n",
+    "        ),\n",
     "    },\n",
-    "]\n",
+    ")\n",
     "\n",
-    "arr_https = decode(chunk_bytes, codecs_wasm_https)\n",
-    "np.testing.assert_array_equal(arr_https, arr)\n",
-    "print(\"Decoded arrays match: OK\")\n",
+    "prepared_https = prepare(pipeline, \"decode\", resolver=resolver_https)\n",
+    "result_https = run(prepared_https, {\"bytes\": chunk_bytes})\n",
     "\n",
-    "encoded_https = encode(arr_https, codecs_wasm_https)\n",
-    "describe_bytes(encoded_https)\n",
-    "assert encoded_https == chunk_bytes, \"Round-trip mismatch!\"\n",
-    "print(\"Round trip: OK\")"
+    "assert result_https[\"bytes\"] == result[\"bytes\"]\n",
+    "print(\"Decoded bytes match: OK\")"
    ]
   },
   {
@@ -513,56 +436,34 @@
    "source": [
     "### OCI\n",
     "\n",
-    "The module can also be pulled from an OCI registry by updating the `uri` accordingly."
+    "The same codec can be pulled from an OCI registry by changing the URI scheme."
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 42,
+   "execution_count": 11,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Decoded arrays match: OK\n",
-      "size: 1.381 MiB | sha256: 2c02e7e60074d6767ccb4c44de2da249d331fd82e107431e41cfe4069bae0d62\n",
-      "Round trip: OK\n"
+      "Decoded bytes match: OK\n"
      ]
     }
    ],
    "source": [
-    "codecs_wasm_oci = [\n",
-    "    {\n",
-    "        \"name\": \"bytes\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\n",
-    "            \"endian\": \"little\",\n",
-    "            \"data_type\": \"uint16\",\n",
-    "            \"shape\": [1024, 1024],\n",
-    "        },\n",
-    "    },\n",
-    "    {\n",
-    "        \"name\": \"tiff_predictor_2\",\n",
-    "        \"type\": \"wasm\",\n",
-    "        \"uri\": \"oci://ghcr.io/cylf-dev/tiff-predictor-2-c:v0.1.0\",\n",
-    "        \"configuration\": {\"bytes_per_sample\": 2, \"width\": 1024},\n",
+    "resolver_oci = Resolver(\n",
+    "    force_sources={\n",
+    "        \"tiff-predictor-2\": \"oci://ghcr.io/cylf-dev/tiff-predictor-2-c:v0.3.0\",\n",
     "    },\n",
-    "    {\n",
-    "        \"name\": \"zlib\",\n",
-    "        \"type\": \"numcodecs\",\n",
-    "        \"configuration\": {\"level\": 9},\n",
-    "    },\n",
-    "]\n",
+    ")\n",
     "\n",
-    "arr_oci = decode(chunk_bytes, codecs_wasm_oci)\n",
-    "np.testing.assert_array_equal(arr_oci, arr)\n",
-    "print(\"Decoded arrays match: OK\")\n",
+    "prepared_oci = prepare(pipeline, \"decode\", resolver=resolver_oci)\n",
+    "result_oci = run(prepared_oci, {\"bytes\": chunk_bytes})\n",
     "\n",
-    "encoded_oci = encode(arr_oci, codecs_wasm_oci)\n",
-    "describe_bytes(encoded_oci)\n",
-    "assert encoded_oci == chunk_bytes, \"Round-trip mismatch!\"\n",
-    "print(\"Round trip: OK\")"
+    "assert result_oci[\"bytes\"] == result[\"bytes\"]\n",
+    "print(\"Decoded bytes match: OK\")"
    ]
   },
   {
@@ -571,7 +472,7 @@
    "source": [
     "## Summary\n",
     "\n",
-    "chonkle codec pipelines can freely mix Python and WebAssembly codecs. A codec written in C or Rust can be compiled to Wasm and published to a file server, HTTPS endpoint, or OCI registry. Consumers can then drop that codec into their pipeline — it will run safely and at near-native speed, with no platform-specific binaries or manual compilation required."
+    "chonkle codec pipelines are DAGs of codec steps defined in JSON, with typed ports wired together. A single pipeline definition supports both encode and decode via direction inversion. Codec resolution is automatic: the resolver picks the best available backend (native Python, core Wasm, or Component Model Wasm) based on a configurable preference order. Wasm codecs can be loaded from local files, HTTPS URLs, or OCI registries — and pipelines can embed their own source URIs to be fully self-describing."
    ]
   }
  ],
diff --git a/demo/tiff-predictor-2-c.wasm b/demo/tiff-predictor-2-c.wasm
index 333efd9..34430f0 100755
Binary files a/demo/tiff-predictor-2-c.wasm and b/demo/tiff-predictor-2-c.wasm differ
diff --git a/docs/CACHING.md b/docs/CACHING.md
deleted file mode 100644
index ab5e53e..0000000
--- a/docs/CACHING.md
+++ /dev/null
@@ -1,34 +0,0 @@
-# Caching
-
-This document describes how caching currently works in chonkle and what we've considered but haven't implemented yet.
-
-## Wasm file download cache
-
-Downloaded `.wasm` files are cached on disk to avoid redundant network fetches. See `wasm_download.py`.
-
-- **HTTPS**: keyed by `sha256(url)`, stored at `<cache_dir>/https/<hash>/<filename>`
-- **OCI**: keyed by registry reference, stored at `<cache_dir>/oci/<ref_path>/`
-- **Default location**: `$TMPDIR/chonkle/wasm/` (overridable via `CHONKLE_CACHE_DIR`)
-- **Force re-download**: `CHONKLE_FORCE_DOWNLOAD=1` or `force=True` parameter
-- **Atomic writes**: HTTPS downloads use tempfile + rename to avoid partial files
-
-There is no eviction, TTL, or size management. However, Wasm modules are small (typically KBs), the default `$TMPDIR` location is cleaned by the OS periodically, and the force-download provides an escape hatch for stale cache scenarios. A more robust cache can be implemented in the future if needed.
-
-## wasmtime compiled-code disk cache
-
-Every call to `_wasm_call` in `wasm_runner.py` creates a `wasmtime.Engine` and compiles the `.wasm` file to native machine code — via `Module.from_file()` on the Core path or `Component.from_file()` on the Component path. This compilation is the most expensive per-call overhead in the Wasm path.
-
-To avoid this repeated compilation, the Engine is configured with wasmtime's built-in compiled-code cache (`config.cache = True`). wasmtime transparently stores compiled native code on disk. When `Module.from_file()` or `Component.from_file()` is called for a `.wasm` file that has already been compiled with the same wasmtime version, the cached native code is loaded instead of recompiling.
-
-This is separate from the download cache above: the download cache stores `.wasm` files to avoid network fetches, while the compilation cache stores native machine code to avoid recompilation.
-
-wasmtime manages the cache location and eviction automatically (`~/.cache` on Linux, `~/Library/Caches` on macOS). The cache invalidates automatically when the wasmtime version changes.
-
-## Warm-container deployments
-
-The compiled-code disk cache eliminates recompilation, but each new process invocation still reads the cached file from disk. In warm-container environments (e.g. AWS Lambda), the same process handles multiple invocations, so caching the `Engine` and compiled `Module` objects in Python module-level variables would skip that disk read on subsequent invocations within the same container.
-
-- `Engine`, `Module`, and `Component` are immutable — safe to reuse across calls with no risk of state leaking between invocations
-- The `Engine` only needs a single instance (singleton) since the compilation settings don't vary
-- A module-level `dict` keyed by resolved file path would cache compiled `Module` or `Component` objects (one per `.wasm` file, type depending on the path taken)
-- `Store` and `Instance` hold mutable state (the Wasm linear memory) — these must be fresh per-call
diff --git a/docs/CODEC_CONTRACT.md b/docs/CODEC_CONTRACT.md
deleted file mode 100644
index c3f2361..0000000
--- a/docs/CODEC_CONTRACT.md
+++ /dev/null
@@ -1,81 +0,0 @@
-# Codec Contract
-
-This document specifies the interface contract between chonkle (the host) and a `.wasm` codec module. Two formats are supported — Core modules and Components — each with a different contract. See [Core modules and Components](WASM.md#core-modules-and-components) for guidance on choosing between them.
-
-## Core module ABI
-
-An ABI (Application Binary Interface) defines how two separately compiled pieces of code talk to each other at the binary level — what functions exist, what types their arguments and return values have, and how memory is laid out.
-
-Every Core Wasm codec module must export a `memory` and three functions. The `memory` export is the module's linear memory — a resizable byte array that the host reads from and writes to in order to pass data across the Wasm boundary. It is automatically provided by the compiler when you target `wasm32`; codec authors do not need to define it.
-
-The functions that codec authors must implement:
-
-| Export | Signature | Purpose |
-| --- | --- | --- |
-| `alloc` | `(size: i32) -> i32` | Allocate `size` bytes, return a pointer |
-| `dealloc` | `(ptr: i32, size: i32)` | Free a previously allocated buffer |
-| `decode` | `(input_ptr: i32, input_len: i32, config_ptr: i32, config_len: i32) -> i64` | Decode data; return packed result |
-| `encode` | `(input_ptr: i32, input_len: i32, config_ptr: i32, config_len: i32) -> i64` | Encode data; return packed result |
-
-A module may export `decode`, `encode`, or both. The host calls whichever function matches the pipeline direction.
-
-### Calling convention
-
-The host (`wasm_runner.py`) performs the following sequence for each codec call (`decode` or `encode`):
-
-1. **Allocate and write input** — Call `alloc(len(data))`, then write the raw input bytes into linear memory at the returned pointer.
-2. **Allocate and write config** — Serialize the configuration dict as a JSON string, call `alloc(len(json_bytes))`, then write the JSON bytes into linear memory.
-3. **Call the codec function** (`decode` or `encode`) — Pass the four arguments: `input_ptr`, `input_len`, `config_ptr`, `config_len`.
-4. **Unpack the result** — The return value is a single `i64` encoding both the output pointer and length: `out_ptr = (result >> 32) & 0xFFFFFFFF`, `out_len = result & 0xFFFFFFFF`.
-5. **Read output** — Copy `out_len` bytes from linear memory starting at `out_ptr`.
-6. **Deallocate** — Call `dealloc` three times to free the input, config, and output buffers.
-
-### Error signaling
-
-If the codec function returns `out_ptr = 0` and `out_len = 0`, the host treats this as a failure and raises a `RuntimeError`.
-
-## Component interface
-
-Component codecs use the WIT/Canonical ABI rather than a manually managed memory interface. The toolchain generates all memory management glue; codec authors work only with high-level types.
-
-A component must export `encode` and/or `decode` with this WIT signature:
-
-```wit
-encode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-decode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-```
-
-### Export location
-
-The function may be exported directly at the world level or nested inside any named interface. `wasm_runner.py` discovers it by introspecting the component's type at runtime, so the interface name and WIT package name do not need to be known in advance.
-
-World-level export:
-
-```wit
-world my-codec {
-    export encode;
-    export decode;
-}
-```
-
-Interface-nested export (either style is valid):
-
-```wit
-interface codec {
-    encode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-    decode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-}
-
-world my-codec {
-    export codec;
-}
-```
-
-### Runtime requirements
-
-- Components must be **WASIp2-compatible**. The host calls `linker.add_wasip2()` unconditionally; components that do not target WASIp2 will fail to instantiate.
-- `config` is passed as a **JSON string** — the same convention as Core modules. Deserialize it inside the component to access codec parameters.
-
-### Error signaling
-
-Return the `Err` variant of `result<list<u8>, string>` to signal failure. The host raises `RuntimeError` with the error string as the message. Return `Ok(list<u8>)` on success.
diff --git a/docs/DISTRIBUTION.md b/docs/DISTRIBUTION.md
deleted file mode 100644
index 65846ab..0000000
--- a/docs/DISTRIBUTION.md
+++ /dev/null
@@ -1,141 +0,0 @@
-# Remote Storage Options for .wasm Files
-
-This document compares approaches for making `.wasm` build artifacts publicly available for download.
-
-## GitHub Releases
-
-Releases let you attach binary assets to tagged versions of a repository.
-
-### How it works
-
-1. Tag a version in the repo (e.g., `v0.1.0`)
-2. CI builds the `.wasm` file
-3. CI creates a release and attaches the `.wasm` as an asset using `gh release create` or the `softprops/action-gh-release` action
-
-Download URL follows a predictable pattern:
-
-```text
-https://github.com/{owner}/{repo}/releases/download/{tag}/file.wasm
-```
-
-The latest release can be queried via the GitHub API.
-
-### Pros
-
-- Simple HTTP GET to download — no special tooling needed on the consumer side
-- Trivial CI setup
-- No authentication required for public repos
-- Versioned naturally via git tags
-- 2 GB per file limit
-
-### Cons
-
-- Not a registry — no standardized discovery, search, or metadata beyond what GitHub provides
-- No concept of namespaces, media types, or multi-artifact manifests
-- Consumers must know the exact repo and tag to fetch from
-
-## GitHub Packages (language-specific registries)
-
-GitHub Packages is an umbrella that hosts several package registries: npm, Maven, NuGet, and a Docker/OCI registry. The Docker/OCI registry is **GitHub Container Registry (GHCR)**, which is covered separately below because it works fundamentally differently — it supports arbitrary binary artifacts and anonymous public access. This section covers the language-specific registries only.
-
-### How it works
-
-The language-specific registries have no generic binary support. To use them for `.wasm` files, you would need to wrap the artifact in a supported package format — for example, publishing an npm package that contains the `.wasm` file, which consumers would install via npm/yarn.
-
-### Pros
-
-- Integrates with existing package manager workflows if consumers already use npm, Maven, etc.
-- Scoped to the repository that published it
-- Free for public repos
-
-### Cons
-
-- No native support for generic binary artifacts — you must shoehorn `.wasm` into a package format
-- npm packages require `npm install` to consume, adding a Node.js dependency for non-JS consumers
-- Most GitHub Packages registries require authentication even for public packages (Docker/GHCR is the exception)
-- Adds unnecessary packaging overhead and complexity for a single `.wasm` file
-
-## GitHub Container Registry (GHCR) with OCI Artifacts
-
-GHCR (`ghcr.io`) is the OCI-compliant registry within GitHub Packages. While technically part of the same product, it behaves very differently from the language-specific registries above: it supports arbitrary OCI artifacts (not just Docker images), allows anonymous pulls for public packages, and is scoped to orgs/users rather than individual repos.
-
-This is the approach the broader Wasm ecosystem is converging on.
-
-### How it works
-
-1. CI builds the `.wasm` file
-2. Push the `.wasm` as an OCI artifact using a tool like [`oras`](https://oras.land/):
-
-```bash
-oras push ghcr.io/{owner}/{repo}:v0.1.0 codec.wasm:application/wasm
-```
-
-1. Consumers pull it with:
-
-```bash
-oras pull ghcr.io/{owner}/{repo}:v0.1.0
-```
-
-Or use an OCI client library in their language of choice.
-
-### Pros
-
-- OCI registries are becoming the standard distribution mechanism for Wasm modules (used by wasmCloud, Spin, Bytecode Alliance tooling, etc.)
-- Supports rich metadata via OCI manifests (media types, annotations, multi-platform artifacts)
-- Public packages on GHCR do not require authentication to pull
-- Free for public repos
-- Standardized protocol enables interoperability — artifacts could move to any OCI-compliant registry
-
-### Cons
-
-- More complex CI setup than Releases
-- Consumers need an OCI client rather than a simple HTTP GET
-- Python OCI client library ecosystem is less mature than HTTP libraries
-- Less familiar to contributors who haven't worked with OCI outside of Docker
-
-## Wasm Component Registry (warg / wa.dev)
-
-The Bytecode Alliance has developed [warg](https://warg.io/), a registry protocol for Wasm packages, with [wa.dev](https://wa.dev/) as the public registry instance. The underlying transport is OCI, and the active tooling lives in the [`wasm-pkg-tools`](https://github.com/bytecodealliance/wasm-pkg-tools) project (the `wkg` CLI).
-
-### Component Model orientation
-
-warg and wa.dev are built around the [WebAssembly Component Model](https://component-model.bytecodealliance.org/), a higher-level abstraction on top of core WebAssembly. The registry expects artifacts to be Component Model components — a binary format distinct from plain wasm modules. Components declare typed interfaces using WIT (WebAssembly Interface Types), and the registry can index that metadata for discovery. Packages are addressed as `namespace:name@version` (a warg protocol convention), and even WIT interface definitions are compiled to wasm and published as packages.
-
-chonkle supports both **Core Wasm modules** and **Component Model components**. For Core modules, which don't use WIT or the Component Model, the lower-level `wkg oci` commands could technically be used, but the registry's discovery, metadata, and tooling wouldn't apply — we'd be working against the grain of the ecosystem. For Component codecs, warg/wa.dev is a better fit, though the tooling is still early-stage.
-
-### Pros
-
-- Purpose-built for Wasm distribution — the most "correct" long-term home for Wasm artifacts
-- Federated and decentralized by design — multiple registries can be linked
-- Rich metadata, namespacing, and type-safe interoperability for components
-- Built on OCI, so the underlying transport is standardized
-
-### Cons
-
-- Oriented toward Component Model artifacts; Core Wasm modules don't benefit from its metadata or discovery
-- Still maturing — the original warg registry implementation is no longer actively developed, with work continuing in `wasm-pkg-tools`
-- Tooling and ecosystem are early-stage
-
-## Comparison
-
-| | Releases | Packages | GHCR (OCI) | warg / wa.dev |
-| --- | --- | --- | --- | --- |
-| Setup complexity | Low | Medium | Medium | High |
-| Consumer complexity | Low (HTTP GET) | Medium (package manager) | Medium (OCI client) | Medium (wkg CLI) |
-| Auth required (public) | No | Yes (most registries) | No | No |
-| Generic binary support | Yes | No (must wrap in package format) | Yes (OCI artifacts) | No (Component Model) |
-| Discovery / metadata | Minimal | Package manager conventions | OCI manifests, annotations | Rich (Component Model interfaces, namespaces) |
-| Ecosystem alignment | Generic | Language-specific | Wasm-native direction | Wasm-native (Component Model) |
-| Versioning | Git tags | Package versions | OCI tags / digests | Semantic (namespace:name@ver) |
-| Maturity | Stable | Stable | Stable | Early-stage |
-
-## Current approach
-
-Codec repositories use a single GitHub Actions workflow that publishes `.wasm` files to both **GitHub Releases** (as release assets) and **GHCR** (as OCI artifacts). This provides two consumption paths:
-
-- **GitHub Releases** — simple HTTP GET, no tooling required on the consumer side
-- **GHCR** — OCI-based distribution, aligned with where the Wasm ecosystem is heading
-
-**warg / wa.dev** is a natural fit for distributing Component codecs (e.g. `componentize-py`-authored modules), but its tooling is still early-stage. For Core Wasm modules, which don't align with the Component Model registry model, it provides little benefit and adds complexity.
-
-**GitHub Packages** (npm, Maven, etc.) is not a good fit for this use case and can be ruled out.
diff --git a/docs/OVERVIEW.md b/docs/OVERVIEW.md
new file mode 100644
index 0000000..3a1670b
--- /dev/null
+++ b/docs/OVERVIEW.md
@@ -0,0 +1,37 @@
+# Overview
+
+This document covers why chonkle uses Wasm, why the host is written in Python, and how the three codec backends fit together. For general Wasm background, see the [wasm/](wasm/) knowledge base. For specs and reference material, see [reference/](reference/).
+
+## Why Wasm for custom codecs?
+
+Custom codecs should be fast, portable, and safe. Native code (C, C++, Rust compiled to shared libraries) is fast but requires platform-specific compilation, and it runs with the same privileges as the host process — it can read and write arbitrary memory, access the filesystem, and make network calls. There is no built-in boundary between the host and a native codec.
+
+Wasm gives us all three properties:
+
+- **Fast:** Compiled C/Rust code runs at near-native speed.
+- **Portable:** The same `.wasm` binary runs on any OS and architecture — the runtime compiles it to native code, so codec authors distribute one file instead of per-platform builds.
+- **Safe:** The runtime's sandbox ensures a codec module can only access its own memory — it cannot touch the host process, the filesystem, or the network unless explicitly permitted.
+
+## Why Python as the host?
+
+Python is widely used in the geospatial and scientific data communities (our target audience), which means more people can read, understand, and contribute to the host code than if it were written in a compiled language like Rust or C++.
+
+The tradeoff is performance for Component Model codecs: the wasmtime Python bindings implement canonical ABI lifting and lowering through Python object allocation rather than direct memory operations, running at approximately 1.7 MB/s. Core Wasm codecs bypass this bottleneck entirely by using `Memory.read()`/`Memory.write()`, which are single C-level calls achieving approximately 10 GB/s. A native-language host (Rust, C++) would eliminate the Python overhead for all codec types but limit accessibility. See [design/CANONICAL_ABI_PERF.md](design/CANONICAL_ABI_PERF.md) for measurements.
+
+## Codec backends
+
+chonkle supports three codec backends. All three implement the `Codec` ABC with `call(direction, port_map)` and `signature()` methods. See the [codec contract](reference/codec-contract/) for full interface specifications.
+
+**Component Model Wasm** — `.wasm` components implementing the `chonkle:codec/transform@0.1.0` WIT interface. Uses the canonical ABI for data transfer. Throughput is limited by the Python canonical ABI binding (approximately 1.7 MB/s measured). See [codec-contract/COMPONENT_MODEL.md](reference/codec-contract/COMPONENT_MODEL.md).
+
+**Core Wasm** — wasm32-wasi reactor modules using a binary port-map wire format via `Memory.read()`/`Memory.write()`. Achieves approximately 10 GB/s throughput. `CoreWasmCodec` keeps module instances alive for single-copy transfer between sequential core wasm steps via `ctypes.memmove`. See [codec-contract/CORE.md](reference/codec-contract/CORE.md).
+
+**Native (numcodecs)** — Python codecs from the numcodecs library. Zero Wasm overhead. `numcodecs` and `numpy` are optional dependencies, imported lazily. Supports `"bytes"` and `"ndarray"` calling conventions. See [codec-contract/NATIVE.md](reference/codec-contract/NATIVE.md).
+
+## Execution model
+
+`prepare(pipeline, direction)` parses the pipeline JSON, resolves codec_ids, validates wiring against codec signatures, and returns a `PreparedPipeline`. `run(prepared, inputs)` executes the DAG by calling `codec.call(direction, port_map)` for each step in topological order.
+
+Data flows between steps through `value_store: dict[str, bytes | CoreWasmRef]`. Port-map builders materialize `CoreWasmRef` for non-core downstream codecs and pass them through for core downstream codecs (enabling single-copy transfer). Final pipeline outputs are always materialized to bytes.
+
+See [design/DATA_COPIES.md](design/DATA_COPIES.md) for copy-count accounting across all backend combinations.
diff --git a/docs/README.md b/docs/README.md
new file mode 100644
index 0000000..61bc253
--- /dev/null
+++ b/docs/README.md
@@ -0,0 +1,46 @@
+# chonkle
+
+chonkle is a Python host for running data codec pipelines. Pipelines are DAGs of codec steps backed by Component Model Wasm, Core Wasm, or native Python (numcodecs) implementations.
+
+## Overview
+
+- [OVERVIEW.md](OVERVIEW.md) — Why Wasm, why Python, codec backends summary, and execution model
+
+## Guides
+
+Step-by-step guides for building codecs.
+
+- [guides/](guides/) — How to build each codec type: [Component Model (Rust)](guides/COMPONENT_MODEL_RUST.md), [Component Model (C)](guides/COMPONENT_MODEL_C.md), [Core Wasm](guides/CORE.md), [Native](guides/NATIVE.md)
+
+## Reference
+
+Specs and reference material for codec and pipeline authors.
+
+- [reference/codec-contract/](reference/codec-contract/) — Codec contract: common requirements, plus per-backend specs ([COMPONENT_MODEL.md](reference/codec-contract/COMPONENT_MODEL.md), [CORE.md](reference/codec-contract/CORE.md), [NATIVE.md](reference/codec-contract/NATIVE.md))
+- [reference/CODEC_RESOLUTION.md](reference/CODEC_RESOLUTION.md) — Codec resolution chain, backend preference, binary detection
+- [reference/PIPELINE_SCHEMA.md](reference/PIPELINE_SCHEMA.md) — Pipeline JSON schema: DAG structure, wiring references, step fields
+- [reference/protospec/](reference/protospec/) — Codec inventory: named codecs with typed signatures
+
+## Design
+
+Design rationale, analysis, and comparisons.
+
+- [design/DATA_COPIES.md](design/DATA_COPIES.md) — Copy counts per inter-step edge across all backend combinations, and approaches to reduce them
+- [design/CANONICAL_ABI_PERF.md](design/CANONICAL_ABI_PERF.md) — Performance investigation: where the time goes, Python vs. native host throughput, mitigations
+- [design/CACHING.md](design/CACHING.md) — Local codec store, download cache, and wasmtime compilation cache
+- [design/CODEC_DISTRIBUTION.md](design/CODEC_DISTRIBUTION.md) — Options for distributing `.wasm` codec artifacts (GitHub Releases, GHCR/OCI, warg)
+- [design/PIPELINE_TRADEOFFS.md](design/PIPELINE_TRADEOFFS.md) — Codec pipeline architecture trade-off analysis
+- [design/F3_COMPARISON.md](design/F3_COMPARISON.md) — Comparison with F3
+
+## Wasm background
+
+Foundational Wasm concepts referenced throughout these docs.
+
+- [wasm/OVERVIEW.md](wasm/OVERVIEW.md) — What is WebAssembly? Virtual CPU, runtimes, portability
+- [wasm/WASI.md](wasm/WASI.md) — System interface, `wasm32-wasi` target, command vs reactor
+- [wasm/MEMORY.md](wasm/MEMORY.md) — Linear memory, memory growth, `dlmalloc`, buffer strategies
+- [wasm/COMPONENT_MODEL.md](wasm/COMPONENT_MODEL.md) — Components vs Core modules, composition, toolchain
+- [wasm/WIT.md](wasm/WIT.md) — WIT interface definition language
+- [wasm/WIT_RESOURCES.md](wasm/WIT_RESOURCES.md) — WIT resource types: handles, ownership, handle tables
+- [wasm/EXECUTION_MODEL.md](wasm/EXECUTION_MODEL.md) — Compilation, instantiation, and invocation lifecycle
+- [wasm/MULTI_MEMORY.md](wasm/MULTI_MEMORY.md) — Multi-memory feature, toolchain limitations
diff --git a/docs/WASM.md b/docs/WASM.md
deleted file mode 100644
index 3771cc0..0000000
--- a/docs/WASM.md
+++ /dev/null
@@ -1,153 +0,0 @@
-# Running Custom Codecs with WebAssembly
-
-This document covers how chonkle uses WebAssembly for custom codecs. For general Wasm knowledge, see the knowledge base:
-
-- [What is WebAssembly?](wasm/OVERVIEW.md) — virtual CPU, runtimes, portability
-- [WASI](wasm/WASI.md) — system interface, `wasm32-wasi` target, command vs reactor
-- [Linear memory](wasm/MEMORY.md) — memory growth, `dlmalloc`, buffer strategies
-- [Distribution](DISTRIBUTION.md) — remote storage options for `.wasm` files
-- [Codec contract](CODEC_CONTRACT.md) — the interface contract between host and `.wasm` codec modules
-- [WIT](wasm/WIT.md) — WIT interface definition language and the Component Model
-
-## Why Wasm and Python?
-
-### Why Wasm for custom codecs?
-
-Custom codecs should be fast, portable, and safe. The traditional options for fast codecs — C extensions or Cython — require platform-specific compilation and complex build tooling. Furthermore, native code loaded into your process (via C extensions, shared libraries, etc.) runs with the same privileges as your program: it can read and write any of your process's memory, access the filesystem, and make network calls. There is no built-in boundary between your code and the native extension.
-
-Wasm gives us all three properties:
-
-- **Fast:** Compiled C/Rust code runs at near-native speed.
-- **Portable:** The same `.wasm` binary runs on any OS and architecture — the runtime compiles it to native code, so codec authors distribute one file instead of per-platform builds.
-- **Safe:** The runtime's sandbox ensures a codec module can only access its own memory — it cannot touch the host process, the filesystem, or the network unless explicitly permitted.
-
-### Why Python as the host?
-
-Python is widely used in the geospatial community (our initial target audience), which means more people can read, understand, and contribute to the host code than if it were written in a compiled language like Rust or C++. It also gives us direct access to Zarr's numcodecs library, so standard codecs work out of the box alongside custom Wasm codecs.
-
-## How this project uses Wasm
-
-Wasm codecs are loaded and executed from Python using Wasmtime. Both Core modules and Components are supported; see [Core modules and Components](#core-modules-and-components). Custom codec source code and build tooling live in separate repositories; this project only contains the host-side runner.
-
-### Codec pipelines
-
-The codec pipeline (`pipeline.py`) applies a sequence of codec steps to encode or decode a chunk. Each step can be a **Python codec** (via numcodecs) or a **Wasm codec**. They can be freely mixed in any order. For example, a COG tile pipeline might look like:
-
-```text
-numpy array  ──[bytes (Python)]──►  raw bytes  ──[tiff_predictor_2 (Wasm)]──►  differenced bytes  ──[zlib (Python)]──►  compressed bytes
-```
-
-**Encoding** applies codecs in forward order (as shown above). **Decoding** applies them in reverse order, unwinding the encoding. Each step receives the output of the previous step.
-
-### Pipeline metadata format
-
-Each chunk has a sidecar JSON file that specifies its codec pipeline. Wasm codec steps use `"type": "wasm"` and include a URI pointing to the `.wasm` file:
-
-```json
-{
-  "codecs": [
-    {
-      "name": "bytes",
-      "type": "numcodecs",
-      "configuration": {
-        "endian": "little",
-        "data_type": "uint16",
-        "shape": [1024, 1024]
-      }
-    },
-    {
-      "name": "tiff_predictor_2",
-      "type": "wasm",
-      "uri": "file:///path/to/tiff-predictor-2-c.wasm",
-      "configuration": {
-        "bytes_per_sample": 2,
-        "width": 1024
-      }
-    },
-    {
-      "name": "zlib",
-      "type": "numcodecs",
-      "configuration": {}
-    }
-  ]
-}
-```
-
-Supported URI schemes: `file://`, `https://`, and `oci://`.
-
-### The codec contract
-
-See [CODEC_CONTRACT.md](CODEC_CONTRACT.md) for the full specification — the Core module ABI and the Component WIT interface.
-
-## Core modules and Components
-
-Chonkle supports both Wasm formats. Detection is automatic: `wasm_runner.py` reads the first 8 bytes of each `.wasm` file — Core modules have version bytes `01 00 00 00`, Components have `0d 00 01 00`. The appropriate call path is selected without any configuration required.
-
-Wasm has two approaches to host-module communication. The **Core** approach passes only numbers (`i32`, `i64`, `f32`, `f64`) across the boundary — to exchange bytes or strings, the host manually writes data into linear memory and passes pointers. The **Component Model** uses WIT (Wasm Interface Type) definitions and a Canonical ABI to automatically marshal rich types like `list<u8>` and `string`.
-
-### Core module path
-
-Core modules must export the custom ABI described in [CODEC_CONTRACT.md](CODEC_CONTRACT.md#core-module-abi): `alloc`, `dealloc`, `memory`, and `encode`/`decode`. The host handles all memory management explicitly. Both freestanding and WASI reactor modules are supported.
-
-### Component Model path
-
-Components must export an `encode` and/or `decode` function with signature:
-
-```wit
-func(data: list<u8>, config: string) -> result<list<u8>, string>
-```
-
-The function may be exported directly at the world level or within an interface. `wasm_runner.py` discovers it at runtime by introspecting the component's type, so the WIT package name does not need to be known in advance. Components built with `componentize-py` against WASIp2 are supported.
-
-Error handling: if the component function returns the `Err` variant of the result, chonkle raises `RuntimeError` with the error string.
-
-### Choosing between Core modules and Components
-
-The right choice depends on the codec author's language and toolchain.
-
-**Core modules** have a minimal author-side contract: implement four functions (`alloc`, `dealloc`, `encode`, `decode`). For C, this maps naturally to the language — memory management is explicit anyway, and there are no extra build steps or WIT files to maintain. The ABI is self-contained and stable; it does not depend on `wit-bindgen`, `wasm-tools`, or the Canonical ABI spec.
-
-**Components** are well-suited for high-level languages like Python. `componentize-py` generates a Component naturally from a Python implementation, and the WIT/Canonical ABI handles all type marshalling automatically. For C or Rust, adopting the Component Model adds build steps (a `.wit` file, generated glue code) that the Core path avoids.
-
-## Memory and the copy cost
-
-A natural question when passing potentially large chunks through Wasm codecs is: *can we avoid copying data into Wasm linear memory?*
-
-The answer appears to be **no** — at least not today. Wasm's sandbox model means a module can only read and write its own linear memory — it cannot dereference a host pointer or access the Python heap. The host *can* read and write the module's linear memory, but not the reverse. So data needs to be copied in before processing and copied out after. Each call to `wasm_decode()` or `wasm_encode()` performs two copies:
-
-```text
-   Python bytes
-         │
-     memcpy in           (copy 1: input)
-         │
-         ▼
-Wasm linear memory
-         │
-    decode runs
-         │
-         ▼
-Wasm linear memory
-         │
-     memcpy out          (copy 2: output)
-         │
-         ▼
-   Python bytes
-```
-
-It is worth noting that in a mixed pipeline, native-to-native (from one numcodec codec to another) steps may also involve allocation overhead (e.g. returning a new `bytes` object), though Wasm codecs have the additional cost of copying input into linear memory. Also, the cost of the copies could be small relative to decode computation for certain codecs (decompression, perhaps), and is likely very small relative to the network I/O required to fetch a custom Wasm codec. This would need to be verified through profiling, but it may not be worth optimizing further until we have evidence that the copies are a bottleneck.
-
-### Approaches explored
-
-Several approaches to eliminating copies were explored and are noted here in case they are worth following up on in the future.
-
-- **Shared memory (threads proposal)** — Wasm's `SharedMemory` allows multiple Wasm *instances* to share memory, but it appears to be designed for multi-threaded Wasm execution, not host-guest data sharing. The host would still copy data in.
-- **Sharing memory between pipeline steps** — Adjacent Wasm codecs run as separate instances with separate linear memories. Linking them to a shared memory would require `--import-memory` and tightly couple independently authored modules. It would also break when a native codec sits between two Wasm codecs.
-- **Wasm-first architecture** — Allocating in Wasm memory and having native codecs operate on it in place doesn't seem to work because numcodecs implementations generally return new `bytes` or `ndarray` objects rather than writing into a caller-supplied buffer.
-- **Custom memory backing** — Wasmtime's Rust API has a `MemoryCreator` trait for custom allocation, but it does not appear to be exposed in wasmtime-py.
-
-### Future possibilities
-
-Links to relevant zero-copy discussions:
-
-- Caller-supplied buffers in the [WASI roadmap](https://wasi.dev/roadmap).
-- An [open Component Model proposal](https://github.com/WebAssembly/component-model/issues/398) exploring zero-copy shared views.
diff --git a/docs/design/CACHING.md b/docs/design/CACHING.md
new file mode 100644
index 0000000..fcfae7d
--- /dev/null
+++ b/docs/design/CACHING.md
@@ -0,0 +1,52 @@
+# Caching
+
+Chonkle uses three layers of caching to avoid redundant downloads, recompilation, and re-instantiation. Each layer answers a different question:
+
+| Layer             | What it caches                      | Where                | Managed by            |
+| ----------------- | ----------------------------------- | -------------------- | --------------------- |
+| Codec store       | Downloaded `.wasm` binaries         | `~/.chonkle/codecs/` | chonkle               |
+| Compilation cache | Compiled native machine code        | OS cache dir         | wasmtime              |
+| In-process        | Compiled modules and live instances | Memory               | codec wrapper objects |
+
+A cold start (first run with a new codec) hits all three layers. A warm start (same codec, same wasmtime version) skips the download and recompilation entirely.
+
+## Codec store
+
+The codec store is the only durable cache for downloaded Wasm binaries. It lives at `~/.chonkle/codecs/` (overridable via `CHONKLE_CODEC_STORE`) and is organized by codec_id:
+
+```text
+~/.chonkle/codecs/
+  zlib/
+    zlib-rs.wasm
+  tiff-predictor-2/
+    tiff-predictor-2-c.wasm
+```
+
+The filename comes from the `implementation` field in the binary's embedded signature.
+
+**Population:** When a pipeline declares a `sources` URI for a codec that is not already in the store, the binary is downloaded to a temporary directory, then copied into the store. Subsequent resolutions for the same codec_id skip the download.
+
+**Invalidation:** Manual. Delete the directory or set `CHONKLE_FORCE_INSTALL=1` to overwrite an existing entry on next download.
+
+**Integrity:** HTTPS downloads write to a tempfile and rename on completion, avoiding partial files.
+
+## Compilation cache
+
+Wasmtime maintains a separate disk cache of compiled native machine code (`config.cache = True`). When a `.wasm` file is loaded, wasmtime checks this cache before compiling. On a hit, precompiled native code is loaded directly.
+
+This is orthogonal to the codec store: the codec store holds portable `.wasm` binaries; the compilation cache holds platform-specific compiled output derived from them.
+
+**Location:** Managed by wasmtime (`~/.cache` on Linux, `~/Library/Caches` on macOS).
+
+**Invalidation:** Automatic when the wasmtime version changes.
+
+**Cost:** With a warm compilation cache, loading a component costs ~0.003s (see [CANONICAL_ABI_PERF.md](CANONICAL_ABI_PERF.md)).
+
+## In-process lifetime
+
+Codec wrapper objects compile and instantiate Wasm modules once during pipeline preparation, not on every codec call:
+
+- **Component codecs** load and compile the component at init time. Each call creates only a new store and instance (mutable Wasm state that cannot be reused across calls).
+- **Core codecs** load, compile, and instantiate the module at init time, keeping the instance and its linear memory alive for the codec's lifetime. The persistent instance enables single-copy data transfer between core codecs via direct memory access.
+
+All codec instantiations share a single wasmtime engine, which in turn shares the compilation cache.
diff --git a/docs/design/CANONICAL_ABI_PERF.md b/docs/design/CANONICAL_ABI_PERF.md
new file mode 100644
index 0000000..b530caa
--- /dev/null
+++ b/docs/design/CANONICAL_ABI_PERF.md
@@ -0,0 +1,205 @@
+# Component Model Canonical ABI Performance Investigation
+
+## Background
+
+Integration tests that run the full executor against real compiled Component Model
+codecs take 4–9 seconds each. To find where the time is going, `time.perf_counter()`
+splits were added around each operation in `ComponentCodec.call()`:
+
+```
+from_file   store+linker   instantiate   fn      post_return   normalize
+0.003s      0.000s         0.000s        2.239s  0.000s        0.000s
+```
+
+Everything except `fn(store, port_map)` is essentially free once wasmtime's
+compiled-code disk cache is warm.
+
+## ABI Traffic Is the Bottleneck
+
+`fn()` time scales linearly with **total bytes through the ABI** (input + output)
+at approximately **1.4–1.9 MB/s**:
+
+| Codec      | in       | out    | abi total | fn()   | throughput |
+|------------|----------|--------|-----------|--------|------------|
+| zlib       | 4 KB     | 1 MB   | 1.00 MB   | 0.529s | 1.90 MB/s  |
+| identity   | 1 MB     | 1 MB   | 2.00 MB   | 1.320s | 1.52 MB/s  |
+| predictor2 | 1 MB     | 1 MB   | 2.00 MB   | 1.295s | 1.54 MB/s  |
+| zlib (COG) | 1.4 MB   | 2 MB   | 3.38 MB   | 2.053s | 1.65 MB/s  |
+| identity   | 2 MB     | 2 MB   | 4.00 MB   | 2.771s | 1.44 MB/s  |
+| predictor2 | 2 MB     | 2 MB   | 4.00 MB   | 2.507s | 1.60 MB/s  |
+
+## Codec Computation Is Negligible
+
+A minimal identity codec (`codec/identity-c/`) performs only a `malloc` +
+`memcpy` inside WASM with no other computation. At equal ABI traffic, its
+`fn()` times are nearly identical to predictor2:
+
+| Codec      | abi total | fn()   | throughput |
+|------------|-----------|--------|------------|
+| identity   | 2.00 MB   | 1.320s | 1.52 MB/s  |
+| predictor2 | 2.00 MB   | 1.295s | 1.54 MB/s  |
+| identity   | 4.00 MB   | 2.771s | 1.44 MB/s  |
+| predictor2 | 4.00 MB   | 2.507s | 1.60 MB/s  |
+
+The `fn()` cost is ABI lifting/lowering, not codec computation.
+
+## Lowering vs Lifting
+
+To determine whether input (lowering, Python→WASM) or output (lifting,
+WASM→Python) dominates, two zlib calls were constructed with the same total
+ABI traffic (~2 MB) but opposite distributions:
+
+| Case              | in      | out     | abi total | fn()   | throughput |
+|-------------------|---------|---------|-----------|--------|------------|
+| encode (lowering) | 2.00 MB | 2 KB    | 2.00 MB   | 1.197s | 1.67 MB/s  |
+| decode (lifting)  | 2 KB    | 2.00 MB | 2.00 MB   | 1.024s | 1.96 MB/s  |
+
+The direction of data flow does not materially affect throughput.
+
+## Python Binding Is the Root Cause
+
+To determine whether the 1.4–1.9 MB/s rate is inherent to the canonical ABI or
+specific to the Python binding layer, two minimal CLIs call the same identity
+codec with the same 2 MB buffer:
+
+- `bench/rust-host/` — Rust binary using `wasmtime-rs 41` typed bindings
+- `bench/python-host/time_abi_raw.py` — Python script using `wasmtime-py 41`
+  raw function call, bypassing the chonkle executor entirely
+
+Both warm up the compilation cache before timing, and run 3 iterations at 2 MB.
+
+| Host                    | Size | abi total | fn() run 1 | fn() run 2 | fn() run 3 | throughput |
+|-------------------------|------|-----------|------------|------------|------------|------------|
+| Python (wasmtime-py 41) | 2 MB | 4 MB      | 2.287s     | 2.284s     | 2.349s     | 1.7 MB/s   |
+| Rust (wasmtime-rs 41)   | 2 MB | 4 MB      | 0.001s     | 0.001s     | 0.000s     | 9,113 MB/s |
+
+Rust throughput is ~5,000× faster at the same WASM binary and buffer size.
+The bottleneck is the Python binding layer, not the canonical ABI itself.
+
+Critically, **the Component Model hides its linear memory from the host**:
+
+```python
+mem_idx = instance.get_export_index(store, "memory")
+# → None. Memory is not accessible from a component instance.
+```
+
+There is no way to bypass the canonical ABI and use `Memory.write()` from a
+component instance in Python.
+
+## Architectural Implications
+
+Every step in the current Python executor requires two passes through Python:
+
+```
+Python bytes ── step N input
+  ↓ lower  (copy 1, ~1.7 MB/s)
+WASM linear memory ── step N runs
+  ↓ lift   (copy 2, ~1.7 MB/s)
+Python bytes ── step N output
+```
+
+Each copy is charged at the rate for its own data size (input of step N for
+copy 1, output of step N for copy 2). For a 2-step Component Model codec pipeline with 2 MB tiles,
+total ABI traffic is ~7.5 MB across 4 copies, taking ~4–5 s.
+
+The result is **2 copies per step edge**. Both copies pass through Python's
+heap. In a native host (Rust/Go), the same 2 copies occur but at C memcpy
+speed — the Python binding is the amplifier, not the architecture itself.
+
+For the full edge-type accounting, see [DATA_COPIES.md](DATA_COPIES.md).
+
+## Mitigation for the Python Orchestrator
+
+The correct long-term fix is a native orchestrator. While the Python executor remains in
+use, there is one potential approach to reduce the impact of the Python-speed
+bottleneck.
+
+### Native Python extension for canonical ABI
+
+#### The exact mechanism causing 1.7 MB/s
+
+wasmtime-py's canonical ABI type adaptation runs in Python. The source
+(`wasmtime/_component/_types.py`, `ListType.convert_to_c()`) iterates every
+element individually:
+
+```python
+for e in val:                                         # O(N) Python loop
+    element.convert_to_c(store, e, pointer(raw.data[i]))
+
+# U8.convert_to_c() — called once per byte:
+if not isinstance(val, int):                          # isinstance per byte
+    raise TypeError(...)
+ptr.contents.kind = ffi.WASMTIME_COMPONENT_U8
+ptr.contents.of.u8 = val                              # ctypes field set per byte
+```
+
+A 2 MB buffer triggers ~2M Python loop iterations, ~2M `isinstance()` checks,
+and ~2M ctypes field writes per copy direction. There is no bulk memcpy path.
+The C FFI layer (`wasmtime_component_func_call`) receives a fully pre-marshalled
+`wasmtime_component_val_t` array built entirely by Python.
+
+wasmtime-rs's `Lower` trait for `list<u8>`, by contrast, writes the byte slice
+into Wasm linear memory via a single `memory.write()` — one memcpy, no
+per-element dispatch. The bench results demonstrate the gap: 1.7 MB/s (Python)
+vs 9,113 MB/s (Rust) at the same Wasm binary and buffer size.
+
+#### What the extension does
+
+A Rust crate (`chonkle-wasmtime`) using [PyO3](https://pyo3.rs/) and
+[maturin](https://github.com/PyO3/maturin) wraps wasmtime-rs's component model
+APIs directly. Since `codec/wit/codec.wit` is fixed, `wasmtime::component::bindgen!`
+generates fully typed Rust bindings at compile time:
+
+```rust
+wasmtime::component::bindgen!({ path: "codec/wit/codec.wit", world: "codec" });
+```
+
+The generated `Lower`/`Lift` impls handle `port-map` with one `memory.write()`
+per buffer, not per byte. PyO3 exposes `Engine`, `Component`, and `Store` as
+`#[pyclass]` types. The Python→Rust boundary cost for `bytes` → `Vec<u8>` is a
+single C-level memcpy (~44 ns total PyO3 call overhead), not an O(N) Python
+operation.
+
+`ComponentCodec.call()` replaces its wasmtime-py function call with the extension.
+No other executor logic changes. No codec changes.
+
+#### Expected outcome
+
+All 2N copies that currently run at ~1.7 MB/s run at memcpy speed (~10 GB/s).
+For a 2-step Component Model codec pipeline at 2 MB tiles, per-step call time drops
+from ~2.5 s to the range shown in `bench/rust-host/` (~1 ms). The 4–9 s test times become
+sub-10 ms.
+
+#### Engineering cost
+
+- Build infrastructure: maturin required; the chonkle package gains a Rust
+  compilation step and per-platform wheels (macOS arm64/x86_64, Linux
+  aarch64/x86_64).
+- wasmtime-py and the extension cannot share internal state; the extension owns
+  its own `Engine`/`Store`/`Linker` lifecycle, replacing wasmtime-py for
+  component calls.
+
+#### Upstream path
+
+wasmtime-py maintainers intend to eventually move canonical ABI adaptation into
+the C API layer, which would fix this without a standalone extension
+([wasmtime-py #309](https://github.com/bytecodealliance/wasmtime-py/issues/309)).
+As of wasmtime-py 41 that has not happened; no timeline is set.
+
+#### Migration
+
+When the orchestrator moves to Rust, the extension is superseded: the same
+wasmtime-rs APIs it uses become direct calls in the orchestrator binary. No
+codec changes at any point.
+
+## Conclusion
+
+The root cause is entirely in the Python binding layer, not in the Component
+Model architecture. The correct fix is a native (Rust) orchestrator. No WIT
+changes are required. All existing codecs, including those authored in Python via
+`componentize-py`, continue to work through the same Component Model interface.
+The Python executor should be treated as a development and test host, not a
+production runtime.
+
+See [DATA_COPIES.md](DATA_COPIES.md) for the copy-count accounting and why copy
+elimination within the current Component Model spec is not achievable at this time.
diff --git a/docs/design/CODEC_DISTRIBUTION.md b/docs/design/CODEC_DISTRIBUTION.md
new file mode 100644
index 0000000..74f0e47
--- /dev/null
+++ b/docs/design/CODEC_DISTRIBUTION.md
@@ -0,0 +1,55 @@
+# Codec Distribution Strategy
+
+## Context
+
+Chonkle produces three kinds of codec backends:
+
+- **Component Model Wasm** — compiled to a Component Model component binary with a typed WIT interface
+- **Core Wasm** — compiled to a plain wasm module using a binary port-map wire format with raw `alloc`/`dealloc`/`encode`/`decode` exports; ABI is incompatible with the Component Model
+- **Native** — Python numcodecs wrappers; distributed as Python packages through standard Python packaging, not as Wasm artifacts
+
+This document covers distribution of the two Wasm backends only.
+
+## Decision
+
+Publish Wasm codec artifacts to **GitHub Releases** (as release assets) and **GHCR** (as OCI artifacts) in parallel. Defer distribution via warg / wa.dev until the tooling matures.
+
+## Options considered
+
+### GitHub Releases
+
+Attach `.wasm` files as assets to tagged releases via CI (`gh release create` or `softprops/action-gh-release`). Download URL follows a predictable pattern; no authentication required for public repos. Consumers fetch with a plain HTTP GET.
+
+**Key trade-off:** zero consumer tooling requirement, but no registry semantics — no namespacing, media types, or manifest metadata. Not where the Wasm ecosystem is heading.
+
+### GitHub Container Registry (GHCR) with OCI artifacts
+
+Push `.wasm` files as OCI artifacts to `ghcr.io` using [`oras`](https://oras.land/). Anonymous pulls for public packages; supports rich metadata via OCI manifests. OCI is the distribution substrate the broader Wasm ecosystem is converging on — wasmCloud, Spin, and Bytecode Alliance tooling all use it.
+
+**Key trade-off:** consumers need an OCI client rather than a plain HTTP GET, and the Python OCI client ecosystem is less mature. The complexity is real but bounded.
+
+### warg / wa.dev
+
+The Bytecode Alliance's [warg](https://warg.io/) registry protocol, with [wa.dev](https://wa.dev/) as the public instance. Built around the Component Model: artifacts are expected to be component binaries that declare typed WIT interfaces; packages are addressed as `namespace:name@version`. Built on OCI transport; higher-level registries like wa.dev build the Component Model protocol on top of it. Active tooling lives in [`wasm-pkg-tools`](https://github.com/bytecodealliance/wasm-pkg-tools) (`wkg` CLI).
+
+**Key trade-off:** the most semantically correct home for Component Model codecs, but the original warg registry implementation is no longer actively developed and the tooling is early-stage. Core Wasm codecs are incompatible with the Component Model ABI and cannot be published here regardless of tooling maturity.
+
+### GitHub Packages (language-specific registries)
+
+npm, Maven, NuGet, etc. No native support for generic binary artifacts — distribution would require wrapping `.wasm` files in a language-specific package format. Not applicable.
+
+## Rationale
+
+The dual Releases + GHCR approach covers both consumption patterns that matter today:
+
+- Releases provides a zero-friction path for consumers who just need to download a file
+- GHCR aligns with the ecosystem trajectory and supports richer metadata
+
+warg / wa.dev is the right long-term home for Component Model codecs specifically, but migrating to it requires both tooling maturity and a decision about how to handle Core Wasm codecs (which are ineligible). Until that is resolved, a format-agnostic mechanism is necessary anyway.
+
+## Consequences
+
+- CI must push to both Releases and GHCR on each tagged release
+- Consumers using OCI need `oras` or an OCI client library
+- When warg tooling matures, Component Model codecs can be additionally published to wa.dev; Core Wasm codecs will continue to use Releases or GHCR
+- Native codecs are out of scope here — they follow standard Python packaging
diff --git a/docs/design/DATA_COPIES.md b/docs/design/DATA_COPIES.md
new file mode 100644
index 0000000..8b7f204
--- /dev/null
+++ b/docs/design/DATA_COPIES.md
@@ -0,0 +1,85 @@
+# Data Copy Accounting
+
+This document covers the number of data copies at each inter-step edge in a chonkle pipeline, broken down by source and destination codec backend. For canonical ABI throughput measurements, see [CANONICAL_ABI_PERF.md](CANONICAL_ABI_PERF.md).
+
+## Codec Backends
+
+chonkle has three codec backends:
+
+- **Core Wasm** (`CoreWasmCodec`) — core wasm32-wasi reactor modules. The host accesses linear memory directly via `Memory.data_ptr()` and `Memory.read()`/`Memory.write()`. Outputs are returned as `CoreWasmRef` (a deferred reference to data in the module's linear memory).
+- **Component Model** (`ComponentCodec`) — Component Model components. Linear memory is hidden from the host; all data crosses the boundary through the canonical ABI.
+- **Native** (`NativeCodec`) — numcodecs Python codecs. Data lives on the Python heap as `bytes` objects.
+
+## Fundamental Constraints
+
+Each backend holds data in a different memory space. Wasm modules (core and component) each have their own isolated linear memory. Native codecs hold Python `bytes` on the Python heap. No backend can access another's memory directly, so moving data between steps requires at least one copy — except native-to-native, where Python passes the `bytes` object by reference.
+
+The cost of a copy depends on the mechanism. Core module transfers use `Memory.read()`/`Memory.write()` and `ctypes.memmove`, which are single C-level calls running at ~10 GB/s. Component Model transfers go through the canonical ABI, which allocates Python objects at ~1.7 MB/s in the Python host. In a native (Rust/Go) orchestrator, the canonical ABI bottleneck disappears and all Wasm copies run at memcpy speed.
+
+## Inter-Step Copy Counts
+
+The table below shows the number of copies per inter-step edge in the Python executor. These are inter-step copies only, not per-invocation totals.
+
+| Source / Dest | Core Wasm | Component Model | Native |
+| --- | --- | --- | --- |
+| Core Wasm | 1 | 2 | 1 |
+| Component Model | 2 | 2 | 1 |
+| Native | 1 | 1 | 0 |
+
+### Edge-by-edge breakdown
+
+**Core Wasm to Core Wasm (1 copy):** The source codec returns a `CoreWasmRef` pointing into its linear memory. The port-map builder detects a core-to-core edge, allocates in the destination module via `alloc()`, obtains raw pointers to both memories via `Memory.data_ptr()`, and performs `ctypes.memmove`. The source pointer is obtained before destination allocation so the source memory is not invalidated. One copy at ~10 GB/s.
+
+**Core Wasm to Component Model (2 copies):** The `CoreWasmRef` is materialized to Python `bytes` via `Memory.read()` (~10 GB/s). The bytes are then lowered into the component's linear memory through the canonical ABI (~1.7 MB/s). Two copies; the second is the bottleneck.
+
+**Core Wasm to Native (1 copy):** The `CoreWasmRef` is materialized to Python `bytes` via `Memory.read()` (~10 GB/s). The native codec reads from the `bytes` object directly. One copy.
+
+**Component Model to Core Wasm (2 copies):** The component's output is lifted to Python `bytes` through the canonical ABI (~1.7 MB/s). The bytes are then written into the core module's linear memory via `Memory.write()` (~10 GB/s). Two copies; the first is the bottleneck.
+
+**Component Model to Component Model (2 copies):** Output is lifted to Python `bytes` via the canonical ABI, then lowered into the next component's memory via the canonical ABI. Two copies, both at ~1.7 MB/s.
+
+**Component Model to Native (1 copy):** Output is lifted to Python `bytes` via the canonical ABI (~1.7 MB/s). The native codec reads from the `bytes` object directly. One copy.
+
+**Native to Core Wasm (1 copy):** The native codec returns Python `bytes`. The executor writes these into the core module's linear memory via `Memory.write()` (~10 GB/s). One copy.
+
+**Native to Component Model (1 copy):** The native codec returns Python `bytes`. These are lowered into the component's memory through the canonical ABI (~1.7 MB/s). One copy.
+
+**Native to Native (0 copies):** Python passes the `bytes` object by reference. No copy.
+
+### Cost summary
+
+Copies involving the canonical ABI are the bottleneck. At ~1.7 MB/s, a single canonical ABI copy of 2 MB takes ~1.2 s. All other copies (materialize, `Memory.read`/`Memory.write`, `ctypes.memmove`) run at ~10 GB/s and are negligible in comparison.
+
+## Copy Reduction
+
+### Single-copy transfer (implemented)
+
+When both source and destination are core wasm modules, the executor uses `_single_copy_transfer`: allocate in the destination module, get raw pointers to both memories, `ctypes.memmove`. This avoids materializing to Python `bytes` and achieves one copy at ~10 GB/s instead of two copies through `Memory.read()` then `Memory.write()`.
+
+### Shared memory between core modules
+
+Each core wasm module has its own isolated linear memory. An earlier design used a single shared `wasmtime.Memory` for all modules, enabling zero-copy transfers. This was abandoned because every module's data segments target the same low addresses — when `linker.instantiate()` writes a module's data segments, it overwrites whatever a previous module left there. This prevents multiple module instances from coexisting simultaneously, which is required for single-copy transfer and eventual parallel execution of independent DAG branches.
+
+The Wasm multi-memory proposal (part of Wasm 3.0) could in theory solve this: each module keeps a private memory for data segments and heap, and all modules share a second memory for the data plane. In practice, C and Rust compilers (LLVM) can only emit load/store instructions targeting a module's default memory — there is no way to compile code that reads from or writes to a second memory. Library-wrapping codecs (zlib, zstd, etc.) would need to copy data between shared and private memory on every call, which is worse than the current design. See [../wasm/MULTI_MEMORY.md](../wasm/MULTI_MEMORY.md) for the full toolchain analysis.
+
+### Pipeline composition
+
+`wasm-tools compose` can merge multiple Component Model components into a single component. Intra-component calls share linear memory, so inter-step copies disappear. This requires a static pipeline known at build time, which is incompatible with the dynamic DAG model. Even with a static DAG, composition links function calls rather than data flow: in a fan-out, each downstream component independently calls the upstream component, causing it to run once per consumer instead of once for the graph.
+
+### Component Model resources
+
+Component Model resources are handle-based types: a component holds an integer handle to an object owned by another component, and the underlying data does not move when the handle is passed. See [../wasm/WIT_RESOURCES.md](../wasm/WIT_RESOURCES.md) for the general concepts.
+
+There are two patterns for sharing a resource across component boundaries. Pattern 1 (direct composition) requires B's WIT to name A's package at compile time — `use a:comp/blobs@0.1.0.{blob}`. This achieves 1 copy per edge but requires static coupling. chonkle pipelines are assembled at runtime from JSON; no codec knows its neighbors when it is compiled, so Pattern 1 is incompatible with the dynamic DAG model.
+
+Pattern 2 (shared buffer-store component) lets both A and B import from a shared third component. The orchestrator only exchanges handles, but data still crosses two component boundaries: A copies into the buffer-store via `blob.constructor()`, and B copies out via `blob.as-bytes()`. This gives 2 copies per edge — the same as `list<u8>` value passing.
+
+Additionally, wasmtime-py 41 does not support Component Model resource types. The Python API does not expose handles, resource tables, or destructors. wasmtime-rs supports resources fully, so they become available as a design option when the orchestrator moves to Rust — but the analysis above shows they offer no copy-count advantage for an orchestrated pipeline with dynamically assembled steps.
+
+### Caller-supplied buffers (not yet available)
+
+The [WASI roadmap](https://wasi.dev/roadmap) includes caller-supplied buffers, which would allow a host to pass a pre-allocated buffer for the module to write into, eliminating the copy-out step. This is not yet part of the spec.
+
+### Native Python extension for canonical ABI
+
+A Rust extension (PyO3 + wasmtime-rs) could replace the Python canonical ABI path with typed bindings that use `memory.write()` per buffer instead of per-byte Python iteration. This would bring Component Model copies to memcpy speed (~10 GB/s), making all Wasm-involving edges equivalent in cost to core module edges. This does not reduce the number of copies — Component Model edges would still be 2 copies — but it eliminates the Python overhead that makes those copies slow. See [CANONICAL_ABI_PERF.md](CANONICAL_ABI_PERF.md) for details.
diff --git a/docs/design/F3_COMPARISON.md b/docs/design/F3_COMPARISON.md
new file mode 100644
index 0000000..b0ceba1
--- /dev/null
+++ b/docs/design/F3_COMPARISON.md
@@ -0,0 +1,71 @@
+# F3 vs. chonkle: Comparison and Tradeoff Analysis
+
+Last reviewed: 2026-03-28
+
+References:
+- [F3: The Open-Source Data File Format for the Future (CMU SIGMOD 2025)](https://db.cs.cmu.edu/papers/2025/zeng-sigmod2025.pdf)
+- [github.com/future-file-format/F3](https://github.com/future-file-format/F3)
+
+---
+
+## What F3 proposes
+
+F3 is a replacement for Parquet/ORC — a self-describing columnar file format for analytics workloads.
+
+- **IOUnit/EncUnit hierarchy**: IOUnit is the I/O parallelism unit (multiple pages); EncUnit is an atomic byte buffer encoded by a single algorithm. Stacking EncUnits within a column composes encodings hierarchically.
+- **Wasm embedded in the data file**: When a new encoding is used, its Wasm decoder is embedded inside the file itself so any reader can decode it without pre-installed libraries. The codec travels with the data.
+- **Vortex encodings**: 15+ built-in columnar encodings (bit-packing, ALP, dictionary, etc.) composable as linear chains per column.
+- **FlatBuffers** for zero-copy metadata deserialization, replacing Parquet's Thrift.
+- No explicit WIT or Component Model usage — a "general-purpose API" for Wasm decoders without a formal interface type system.
+
+---
+
+## Key differences
+
+| Dimension | F3 | chonkle |
+|---|---|---|
+| **Domain** | Structured tabular analytics (Parquet replacement) | Chunked array data — columnar formats (Parquet, ORC) and raster imagery (COG, satellite) |
+| **Data model** | Typed columnar schema, FlatBuffer metadata | Opaque named byte buffers (`port-map`), schema-agnostic |
+| **Codec composition** | Linear stack within a column (EncUnit chain) | Arbitrary DAG with named ports, fan-out/fan-in |
+| **Where Wasm lives** | Embedded inside the data file as a portable fallback decoder | Fetched from external registry (`file://`, `https://`, `oci://`) at pipeline execution time |
+| **Interface contract** | Informal "general-purpose API" — no WIT types | Three backends: Component Model WIT (`chonkle:codec/transform`), Core Wasm binary ABI, and native Python (numcodecs) |
+| **Bidirectionality** | Separate write (encode) and read (decode) paths | Single pipeline definition, invertible at runtime via DAG reversal |
+| **Self-description** | Data files are self-describing (decoder embedded) | Pipelines are separate from data; data is not self-describing |
+| **Metadata format** | FlatBuffers for column schema and statistics | `chonkle:signature` custom section embedded in the `.wasm` binary (port descriptors, direction constraints) |
+| **Codec distribution** | Embedded in file; central repo proposed | `file://`, `https://`, and `oci://` URIs; remote codecs cached in a local store |
+
+---
+
+## F3's advantages over chonkle's approach
+
+**Self-describing data is a strong archival guarantee.** If you encode data today with a chonkle pipeline, the pipeline JSON is the only thing that knows how to decode it. Lose the pipeline, lose the data. F3's "decoder travels with the data" property is directly relevant for NASA data that must be readable in 20 years without external dependencies.
+
+**Schema awareness enables query pushdown.** F3 knows column types, shapes, and statistics. It can skip I/O for irrelevant columns, push down predicates, and support partial decoding without reading the whole column. Chonkle treats everything as opaque bytes — any schema understanding lives outside the pipeline.
+
+**Random access within EncUnits.** Because F3 tracks structure, it can decode a row range without touching the rest of the column. Chonkle pipelines are all-or-nothing: a step receives a full port-map and emits a full port-map.
+
+**Ecosystem integration target.** F3 aims to plug into Spark, DuckDB, and Pandas by replacing Parquet. Chonkle has no query engine integration story.
+
+---
+
+## Chonkle's advantages over F3's approach
+
+**Arbitrary DAG composition.** F3's EncUnit chain is linear per column. Chonkle can express fan-out (one output routed to multiple inputs), fan-in (multiple outputs merged), and cross-step data routing via named wiring refs. This is the right model for both columnar formats (e.g., splitting a Parquet page into rep levels, def levels, and values for independent processing) and multi-channel imagery where byte planes are split, processed, and recombined — something a linear stacking model cannot express without significant contortion.
+
+**Formal interface contract.** Component Model codecs use WIT (`chonkle:codec/transform`) for a verifiable, language-agnostic contract. Wasm backends (Component Model and Core Wasm) carry port descriptors and direction constraints as a `chonkle:signature` custom section embedded in the binary; native (numcodecs) backends use bundled JSON signature files. All backends go through the same runtime port-level validation. F3's "general-purpose API" is informal by comparison — nothing prevents a codec from violating its contract until runtime.
+
+**Bidirectional pipelines.** F3 has no concept of inverting a pipeline. For imagery processing, encode and decode are two sides of the same transform. A single chonkle DAG runs in both directions, which eliminates a class of encode/decode asymmetry bugs and removes the need to author a separate decode path.
+
+**Codec sandboxing.** Each chonkle codec is an isolated Wasm component. A buggy or malicious codec cannot corrupt the host process. F3 embeds Wasm from potentially untrusted data files; chonkle fetches codecs from a controlled registry with signature verification. F3 acknowledges a 10–30% performance penalty for the sandboxed path.
+
+**Concrete distribution model.** Chonkle resolves codecs from `file://`, `https://`, and `oci://` URIs, with remote artifacts cached in a local store. F3 notes "central repository verification proposed" without specifying it. For a system pulling codecs on demand, a concrete distribution story matters.
+
+---
+
+## Where the problems diverge
+
+F3 is optimizing for **reading existing data efficiently and portably** — a file format problem. Chonkle is optimizing for **composing transforms flexibly** — a pipeline problem. These are different problem shapes, and the approaches are not competing so much as operating at different levels of the stack.
+
+**The gap chonkle does not address:** if you run a chonkle pipeline to encode data, how does a reader know which pipeline to use to decode it? Currently the answer is "the caller knows," which is appropriate for a tightly controlled system but fragile at archival or interoperability scale.
+
+One synthesis worth considering: a content-addressed pipeline URI could be written as a sidecar alongside encoded data, giving chonkle's compositional power with something closer to F3's self-description guarantee. That is structurally what F3 does with embedded Wasm, but at pipeline granularity rather than individual codec granularity. The protospec's `codec_id` field already points in this direction — a stable identifier that could resolve to a pipeline definition in a registry.
diff --git a/docs/design/PIPELINE_TRADEOFFS.md b/docs/design/PIPELINE_TRADEOFFS.md
new file mode 100644
index 0000000..e0aa657
--- /dev/null
+++ b/docs/design/PIPELINE_TRADEOFFS.md
@@ -0,0 +1,605 @@
+# Codec Pipeline Architecture: Trade-off Analysis
+
+This document analyzes the trade-offs across two related design axes for codec
+pipelines that support WebAssembly:
+
+1. **Mixed native + Wasm codecs vs. all-Wasm codecs** — whether to allow
+   non-Wasm codecs (Python C extensions, shared libraries, etc.) alongside Wasm
+   codecs in the same execution graph.
+
+2. **Core module ABI vs. Component Model vs. both** — which Wasm interface style
+   to use for the Wasm codecs: raw integer pointer/length conventions (Core), the
+   typed WIT/Canonical ABI layer (Component Model), or a mix of the two.
+
+These axes interact. The combination chosen affects performance, portability,
+security, testability, and operational complexity. Both axes are analyzed
+separately, then the interactions are summarized.
+
+---
+
+## Background: Data Movement at Step Boundaries
+
+Understanding copy counts is foundational to both analyses. Wasm's sandbox model
+means a module can read and write only its own linear memory — it cannot
+dereference a host pointer. The host can read and write the module's linear
+memory but not the reverse. This isolation is why data copies at step boundaries
+are currently unavoidable.
+
+For a Component Model codec invocation, two copies occur at each step boundary
+regardless of host language — one to lower data into the component's linear
+memory, one to lift output back to the host:
+
+```
+     host bytes
+         │
+      copy in   (lower: host → component linear memory)
+         │
+         ▼
+component linear memory
+         │
+    codec runs
+         │
+         ▼
+component linear memory
+         │
+      copy out  (lift: component linear memory → host)
+         │
+         ▼
+     host bytes
+```
+
+Core Wasm codecs differ: the output stays in linear memory as a `CoreWasmRef`
+deferred reference. When the next step is also a Core Wasm codec, the executor
+does one `ctypes.memmove` between linear memories — one copy total for the
+edge, not two. When the next step is Native, the output is read out of linear memory into
+Python bytes — one copy. When the next step is Component Model, that read is
+followed by a second copy to lower the bytes into the component's linear memory
+via the canonical ABI.
+
+The number of copies per edge, and their cost, depend on both source and
+destination backend:
+
+| Step A → Step B                               | Copies | Transfer mechanism            | Speed (Python host)  |
+|-----------------------------------------------|--------|-------------------------------|----------------------|
+| Native → Native                               | 0      | Python object reference       | —                    |
+| Native → Core module                          | 1      | `Memory.write()` (ctypes)     | ~10 GB/s             |
+| Core module → Native                          | 1      | `Memory.read()` (ctypes)      | ~10 GB/s             |
+| Core module → Core module (Python host)       | 1      | `ctypes.memmove` (CoreWasmRef)| ~10 GB/s             |
+| Component → Component (Python host)           | 2      | Canonical ABI lift + lower    | 1.4–1.9 MB/s †       |
+| Component → Component (Rust/Go host)          | 2      | Single memcpy per direction   | ~10 GB/s             |
+
+† Measured on wasmtime-py 41 against the `identity` and `predictor2` codecs with
+1–2 MB buffers. The Python binding (`wasmtime._component._types.ListType`) iterates
+every byte individually in Python: one `isinstance()` check and one ctypes field
+write per byte. At 2 MB, this is ~2 million Python loop iterations per copy
+direction. The Rust binding for the same WASM binary does a single `memory.write()`
+call (one memcpy). Measured throughput: 1.7 MB/s (Python raw bench) vs. 9,113 MB/s
+(Rust bench) at identical buffer sizes. This bottleneck is in the Python binding
+layer; it is not a property of the Component Model architecture. The correct fix is
+a Rust or Go orchestrator.
+
+---
+
+## Part 1: Mixed Native + Wasm vs. All-Wasm Pipelines
+
+### Definitions
+
+A **native codec** runs directly in the host process: Python C extensions
+(numcodecs, Cython, ctypes, cffi), shared libraries, or pure Python functions.
+No sandbox, no runtime overhead, full access to host memory and hardware.
+
+A **Wasm codec** is compiled to a `.wasm` binary and executed inside a Wasm
+runtime (e.g., Wasmtime). The runtime's sandbox enforces memory isolation.
+
+> Note: `.wasm` files are not directly executable. The runtime compiles them to
+> native machine code (`.cwasm`) on first load and can cache the result on disk.
+> The `.wasm` artifact is portable; the compiled `.cwasm` is platform-specific.
+
+---
+
+### Mixed Native and Wasm Pipelines
+
+**Pros**
+
+- **Incremental adoption.** Existing native codecs can be used immediately
+  without rewriting. Teams can introduce Wasm codecs alongside existing Python
+  packages without a full migration.
+
+- **Zero-copy at native-to-native boundaries.** When two adjacent pipeline steps
+  are both native (e.g., two numcodecs transforms), data passes as a Python object
+  reference. No serialization or memory copy occurs at that edge.
+
+- **Full hardware access.** Native codecs have unrestricted access to GPUs,
+  hardware crypto accelerators, architecture-specific SIMD (AVX, NEON), and any
+  OS-level facility. Wasm does not support GPU offload and relies on WASI for
+  system interfaces, which does not yet cover all hardware access patterns.
+
+- **Mature toolchain.** Native codecs use standard package managers (pip, conda)
+  and standard build tools. No Wasm-specific toolchain is required for the native
+  steps.
+
+- **Native debugging.** Stack traces from native codecs are standard Python/C
+  frames. Debuggers and profilers (pdb, py-spy, gdb) attach directly without
+  Wasm frame indirection.
+
+**Cons**
+
+- **Non-uniform interface contract.** Native and Wasm codecs have different
+  calling conventions. The orchestrator must implement at least two code paths.
+  Load-time interface verification applies only to the Wasm subset; native codecs
+  have no equivalent mechanism.
+
+- **Inconsistent security boundary.** Wasm codecs are sandboxed; native codecs
+  run in the host process without isolation. A bug or supply-chain compromise in
+  a native codec has full access to the host process, its memory, and its
+  filesystem. The overall pipeline's security posture is determined by the
+  least-sandboxed codec.
+
+- **Heterogeneous distribution.** Native codecs are Python packages with
+  platform-specific builds (`.so`, `.pyd`). Wasm codecs are `.wasm` binaries with
+  different registries and verification schemes. Versioning, pinning, auditing,
+  and caching use two distinct mechanisms.
+
+- **Portability limited by native codecs.** A pipeline is only as portable as its
+  least-portable native codec. A native codec that depends on a platform-specific
+  shared library cannot run in a containerized or cross-platform environment
+  without platform-specific builds or layers.
+
+- **No static composition.** `wasm-tools compose` and `wac` (the WebAssembly
+  composition tools) operate only on Component Model components. A pipeline that
+  includes native steps cannot be statically merged into a single artifact.
+
+- **Test strategy fragmentation.** Unit-testing a native codec requires different
+  mocking and isolation strategies than testing a Wasm codec. Test infrastructure
+  must account for both codec types.
+
+---
+
+### All-Wasm Pipelines
+
+**Pros**
+
+- **Single orchestrator code path.** The orchestrator deals only with Wasm
+  codecs; there is no native dispatch branch to maintain in parallel. All
+  steps share the same artifact format, distribution infrastructure, and
+  signature-reading machinery.
+
+- **Portable artifacts.** The same `.wasm` binary runs on any OS and architecture
+  that the runtime supports (Linux x86_64, macOS arm64, cloud serverless
+  environments) without changes to the artifact. Platform-specific work is handled
+  by the runtime's compilation step.
+
+- **Uniform security boundary.** Every codec is sandboxed. A buggy or compromised
+  codec cannot access host memory, the filesystem, or the network beyond what the
+  runtime explicitly grants via WASI. The trust model is consistent across all
+  pipeline steps.
+
+- **Uniform distribution.** All codecs are `.wasm` artifacts carrying embedded
+  `chonkle:signature` custom sections. They use the same registry and caching
+  infrastructure.
+
+- **Language-agnostic authoring.** Any language with a Wasm toolchain — Rust
+  (`cargo-component`), C/C++ (`wit-bindgen-c`), Python (`componentize-py`), Go
+  (`wit-bindgen-go`) — can implement the same codec interface. The host does not
+  care what language the codec was written in.
+
+- **Static composition (Component Model, fixed topology).** `wasm-tools compose`
+  can merge a fixed pipeline topology into a single component. Intra-component
+  calls share a single linear memory, so inter-step copies are eliminated. A
+  2-step composed pipeline pays 2 copies total (one into the composed component,
+  one out) rather than 4. This requires the pipeline topology to be known at build
+  time; it is not applicable to dynamic DAG pipelines.
+
+**Cons**
+
+- **At least 1 copy per step boundary.** Wasm's sandbox requires data to cross
+  the host–codec boundary at each step edge. Component Model codecs incur 2
+  copies per edge (lower + lift through the Canonical ABI). Core Wasm codecs
+  incur 1 copy per edge (a `memmove` between linear memories for adjacent Core
+  steps, or a single read-out for non-Core downstream codecs). Neither matches
+  the 0-copy path available when two native codecs share process memory.
+  (Static composition reduces Component Model pipeline cost to 2 total, but
+  only for fixed topologies.)
+
+- **Canonical ABI overhead under Python orchestrators (Component Model only).**
+  With wasmtime-py, each copy direction through the Component Model Canonical
+  ABI runs at 1.4–1.9 MB/s for `list<u8>` data. A 2-step pipeline processing
+  2 MB tiles takes approximately 4–5 s. Core Wasm codecs do not use the
+  Canonical ABI; their copy path runs at `memmove` speed. A native (Rust/Go)
+  orchestrator eliminates the Canonical ABI overhead entirely: both copy
+  directions run at memcpy speed (~10 GB/s). The Wasm architecture is not at
+  fault; the bottleneck is the Python binding implementation.
+
+- **Limited hardware access.** GPU compute, architecture-specific SIMD beyond
+  what the Wasm SIMD proposal covers, and OS-level I/O not exposed through WASI
+  are not available inside a codec without explicit runtime host function
+  bindings. Codecs that need hardware accelerators are constrained by what the
+  runtime exposes.
+
+- **Toolchain requirements for codec authors.** Codec authors need a
+  Wasm-capable toolchain and must maintain `.wasm` build artifacts alongside
+  source. Component Model toolchains (cargo-component, componentize-py) are
+  newer than general Wasm toolchains and vary in maturity by language.
+
+- **Runtime scope.** Not every execution environment supports the same Wasm
+  features. Component Model requires a runtime that implements it (Wasmtime,
+  Wasmer 4+). Browser-native Component Model support requires polyfills as of
+  2025. IoT-targeted runtimes (wasm3, WAMR) have limited or no Component Model
+  support.
+
+- **Debugging requires runtime-aware tools.** Useful stack traces from Wasm
+  frames require DWARF debug info embedded in the binary (optional and larger)
+  and runtime support for source-mapped frames (e.g., Wasmtime's `--debug`
+  mode). Standard Python debuggers and profilers do not see inside Wasm frames.
+
+---
+
+### Summary: Mixed vs. All-Wasm
+
+| Property                    | Mixed Native + Wasm                  | All-Wasm                              |
+|-----------------------------|--------------------------------------|---------------------------------------|
+| Interface uniformity        | No — dual calling convention         | ABI-dependent: WIT or binary port-map |
+| Security isolation          | Inconsistent (weakest codec wins)    | Uniform sandbox per codec             |
+| Portability                 | Limited by native codecs             | Platform-agnostic `.wasm` artifact    |
+| Step boundary copies        | 0–2; varies by step backend pair     | 1 (Core→Core); 2 (Component Model)    |
+| Static composition          | Not applicable                       | Available (Component Model, fixed topology) |
+| Distribution uniformity     | No — packages + `.wasm` files        | Yes — `.wasm` artifacts only          |
+| Migration barrier           | Low — existing codecs reusable       | Higher — requires Wasm toolchain      |
+| Hardware access             | Full (native steps)                  | Limited to WASI + host functions      |
+| Load-time verification      | Wasm steps only                      | Component Model steps only            |
+| Debugging                   | Native steps: standard tools; Wasm steps: runtime-aware | Runtime-aware throughout |
+| Orchestrator code paths     | Multiple (by codec type)             | One                                   |
+
+---
+
+## Part 2: Core Module ABI vs. Component Model
+
+This section covers the interface style used for the Wasm codecs — how data
+crosses the host–codec boundary. This choice is orthogonal to Part 1 but
+interacts with it (discussed in Part 3).
+
+### Definitions
+
+**Core module ABI** (hereafter "Core ABI"): The codec is compiled as a Core
+WebAssembly module. Its exported functions use only Wasm's numeric primitives
+(`i32`, `i64`, `f32`, `f64`). Data exchange uses manually agreed
+pointer/length conventions; the host reads and writes the module's linear
+memory directly using `Memory.write()` and `Memory.read()`.
+
+**Component Model**: The codec is compiled as a Component Model component, a
+distinct binary format (different magic bytes: `0d 00 01 00` vs. `01 00 00 00`
+for Core modules). Its interface is declared in WIT; the Canonical ABI handles
+all type marshalling automatically. The component's linear memory is hidden
+from the host — there is no `Memory.write()` path available.
+
+---
+
+### Core Module ABI
+
+**Pros**
+
+- **Fast host↔codec data transfer in all host languages, including Python.**
+  The host exchanges data with the module via `Memory.write()` and
+  `Memory.read()`, which are single C-level calls (one memcpy each) regardless
+  of host language. In Python, this is the same speed as numpy memory access.
+  There is no per-byte Python loop. This is the primary performance advantage
+  over the Component Model's Python binding.
+
+- **Broadest runtime compatibility.** Core WebAssembly is the foundational spec.
+  Every Wasm runtime — Wasmtime, Wasmer, wasm3, browser WebAssembly, WAMR —
+  supports Core modules. Component Model support is a superset of capabilities
+  that not all runtimes implement.
+
+- **Simpler binary format.** Core modules have less format overhead than
+  Component Model wrappers. Inspection with tools like `wasm2wat` or
+  `wasm-objdump` is more direct, which simplifies debugging and auditing.
+
+- **Minimal toolchain requirements.** Any language that can target
+  `wasm32-wasi` or `wasm32-unknown-unknown` produces a Core module without
+  needing Component Model toolchain support. Component Model toolchains are
+  newer and not uniformly available across all languages.
+
+- **No ABI evolution risk from the WIT/Canonical ABI layer.** The calling
+  convention is agreed between the codec author and the host; it does not change
+  unless one side explicitly changes it. There is no shared specification layer
+  that can introduce a breaking change in a new runtime version.
+
+**Cons**
+
+- **No machine-readable interface.** There is no formal description of what the
+  module expects or produces. The host and codec author agree on a calling
+  convention out-of-band (documentation, header files). Violations — wrong
+  pointer size, off-by-one in a length, mismatched buffer layout — surface as
+  silent data corruption or incorrect output, not as a load-time error.
+
+- **No automated code generation.** Every host language must independently
+  implement the same pointer/length marshalling protocol. There is no toolchain
+  equivalent of `wit-bindgen` to generate correct glue from a shared definition.
+
+- **No load-time interface verification.** The runtime validates that exported
+  function type signatures use the right numeric primitive types. It cannot
+  verify that the codec interprets those integers as a port-map in any
+  particular way.
+
+- **No ecosystem composition.** `wasm-tools compose` and `wac` operate on
+  Component Model components only. Core modules cannot be statically composed
+  into a single-binary pipeline.
+
+- **Manual memory management.** The host must call the module's exported
+  allocator to reserve memory, write data in, call the codec function, and then
+  free the allocation. Incorrect management leaks memory within the module's
+  linear memory or causes use-after-free during a codec invocation.
+
+- **No rich type system at the boundary.** All values must be encoded as
+  integers or floats. Strings, lists, records, and variant types require a
+  manually defined serialization convention, which must be independently
+  implemented on both sides and is not self-describing.
+
+---
+
+### Component Model
+
+**Pros**
+
+- **Interface verification at load time.** The runtime checks that the
+  component's WIT exports match the expected interface before any code runs.
+  Type and name mismatches are caught immediately as a hard error — not as
+  silent runtime data corruption.
+
+- **Automated code generation.** WIT toolchains generate all marshalling glue
+  from the `.wit` file: `wit-bindgen` for Rust and C, `cargo-component` for
+  Rust, `componentize-py` for Python, `wit-bindgen-go` for Go. Codec authors
+  work with native types in their language; the generated code handles memory
+  management and type adaptation.
+
+- **Language-agnostic interoperability.** Any language with a Component Model
+  toolchain can implement the same WIT interface. A Rust host can call a Python
+  codec and a C codec through one uniform code path. No manual coordination
+  between codec authors and host developers is needed beyond the shared `.wit`
+  file.
+
+- **Rich type system.** WIT supports strings, `list<T>`, `tuple<T, U>`,
+  `option<T>`, `result<T, E>`, `record` (named struct), `variant` (tagged
+  union), and `resource` (opaque handle). These types are expressed directly in
+  the interface and require no manual serialization convention.
+
+- **Memory isolation by design.** The component's linear memory is not
+  accessible to the host. The host cannot read or write it outside the WIT
+  interface. This is both a security property (host code cannot corrupt component
+  memory) and a correctness property (no accidental aliasing).
+
+- **Self-describing binary.** Components embed their WIT interface declarations
+  in the binary. Tooling can inspect a component's interface without consulting
+  out-of-band documentation.
+
+- **Static composition.** `wasm-tools compose` and `wac` can merge a fixed
+  pipeline topology into one component. Intra-component calls share a single
+  linear memory, eliminating the Canonical ABI crossing (and therefore the copy)
+  at each inter-step boundary. For an N-step pipeline, this reduces total copies
+  from 2N to 2 regardless of step count.
+
+- **Ecosystem alignment.** The Wasm ecosystem — WASI P2, warg/wa.dev component
+  registry, Bytecode Alliance tooling, wasmCloud, Spin — is converging on the
+  Component Model as the standard interface for Wasm artifacts. Components are
+  first-class citizens in this ecosystem; Core modules are a lower-level
+  substrate.
+
+**Cons**
+
+- **Canonical ABI cost under Python hosts.** wasmtime-py's Canonical ABI
+  implementation for `list<u8>` iterates each byte individually in Python:
+  one `isinstance()` check and one ctypes field write per byte, plus Python
+  loop overhead. At 2 MB, this is approximately 2 million Python iterations per
+  copy direction. Measured throughput: ~1.7 MB/s. The Rust binding for the
+  same Wasm binary: ~9,100 MB/s. A Rust or Go orchestrator eliminates this
+  discrepancy entirely; both directions run at memcpy speed.
+
+- **Linear memory hidden from host.** Unlike Core modules, there is no
+  `Memory.write()` path available from a component instance in Python:
+
+  ```python
+  mem_idx = instance.get_export_index(store, "memory")
+  # → None. Memory is not accessible from a component instance.
+  ```
+
+  There is no performance shortcut available from a Python host. The Canonical
+  ABI is the only data exchange path.
+
+- **Narrower runtime support.** Full Component Model support requires a runtime
+  that implements it — Wasmtime, Wasmer 4+. Browser-native Component Model
+  support requires polyfills as of 2025. IoT-focused runtimes (wasm3, WAMR)
+  have limited or no Component Model support.
+
+- **WIT versioning coordination.** WIT interfaces are versioned (e.g.,
+  `chonkle:codec@0.1.0`). A breaking WIT change requires updating every codec
+  and every consumer. This is good discipline but adds coordination overhead
+  compared to Core modules where calling convention changes are unilateral and
+  unversioned.
+
+- **Toolchain maturity varies by language.** Rust (`cargo-component`) and Python
+  (`componentize-py`) have relatively mature Component Model support. Go and Zig
+  have partial or evolving support as of 2025. Some languages have no Component
+  Model toolchain at all.
+
+- **Binary size overhead.** Component wrappers add format overhead (WIT metadata,
+  component section, adapter modules). The absolute difference is typically small
+  (kilobytes to tens of kilobytes) but not zero.
+
+---
+
+### Hybrid: Both Core ABI and Component Model in One Pipeline
+
+Using both Core modules and Component Model components in the same pipeline —
+typically Core ABI for steps where Python-host data transfer speed matters,
+Component Model for steps that need interface verification or are authored in
+multiple languages.
+
+**Pros**
+
+- **Per-codec optimization.** High-throughput, large-buffer codecs that must
+  remain in a Python-orchestrated pipeline can use Core ABI for fast
+  `Memory.write()`/`Memory.read()` transfers. Metadata-handling or
+  complex-interface codecs can use Component Model for interface verification and
+  code generation.
+
+- **Migration path.** Existing Core module codecs can coexist with newer
+  Component Model codecs during a transition. The pipeline continues to work
+  while codecs are migrated incrementally.
+
+- **Broader runtime reach for Core steps.** Core module steps run on any Wasm
+  runtime. Only the Component Model steps require a runtime with Component Model
+  support.
+
+**Cons**
+
+- **Dual calling convention in the orchestrator.** The executor must implement
+  two distinct codec invocation code paths: manual pointer/length exchange for
+  Core modules, Canonical ABI dispatch for Component Model components. These are
+  not composable and increase orchestrator complexity.
+
+- **No uniform interface contract.** Core module steps have no load-time
+  verification. The overall pipeline's type safety is bounded by the unverified
+  subset.
+
+- **No static composition across the whole pipeline.** `wasm-tools compose`
+  cannot link Core modules to Component Model components at a single boundary.
+  Each Core↔Component boundary is a separate copy that cannot be eliminated by
+  composition tooling.
+
+- **Security asymmetry.** Core module linear memory is accessible to the host;
+  component memory is not. The trust model differs per codec, which complicates
+  auditing.
+
+- **Distribution incompatibility.** Component Model registries (warg/wa.dev)
+  treat components as first-class artifacts. Core modules require separate
+  distribution infrastructure or must be treated as opaque blobs.
+
+- **Increased orchestrator complexity without a clear resolution point.** The
+  dual calling convention adds code paths that cannot be unified. Unlike a
+  temporary migration, this complexity persists as long as both backend types
+  are in use.
+
+---
+
+### Summary: Core ABI vs. Component Model vs. Hybrid
+
+| Property                        | Core Module ABI                       | Component Model                    | Hybrid                             |
+|---------------------------------|---------------------------------------|------------------------------------|-------------------------------------|
+| Interface verification at load  | No                                    | Yes (WIT)                          | Partial (Component steps only)     |
+| Automated code generation       | No                                    | Yes (wit-bindgen et al.)           | Partial                             |
+| Data transfer speed (Python)    | ~ctypes speed (~10 GB/s)              | 1.4–1.9 MB/s (Python binding)      | Mixed                               |
+| Data transfer speed (Rust/Go)   | ~10 GB/s                              | ~10 GB/s                           | ~10 GB/s                            |
+| Linear memory accessible        | Yes (host uses Memory.read/write)     | No (hidden by design)              | Mixed                               |
+| Static composition              | No                                    | Yes (fixed topology only)          | No (cannot span both types)         |
+| Runtime support breadth         | All runtimes                          | Wasmtime, Wasmer 4+                | All runtimes (Core steps only)      |
+| Rich types at boundary          | No (integers/floats only)             | Yes (WIT type system)              | Partial                             |
+| Self-describing binary          | No                                    | Yes                                | Partial                             |
+| Memory management               | Manual (host manages alloc/free)      | Automated (Canonical ABI)          | Mixed                               |
+| Security model                  | Sandboxed; memory readable by host    | Full isolation                     | Inconsistent per codec              |
+| Ecosystem alignment             | General Wasm                          | Bytecode Alliance / WASI P2        | Mixed                               |
+| Orchestrator code paths         | One                                   | One                                | Two                                 |
+
+---
+
+## Part 3: Interaction of the Two Axes
+
+The two axes are independent choices but they interact in practice.
+
+**A mixed pipeline still requires a Core-or-Component decision for its Wasm
+subset.** Allowing native codecs does not resolve the Core ABI vs. Component
+Model question for the Wasm steps in the same pipeline. Both choices must be
+made.
+
+**The Python canonical ABI performance problem applies only to Component
+Model + Python host.** Using Core ABI with a Python orchestrator avoids the
+~1.7 MB/s bottleneck entirely, because `Memory.write()` and `Memory.read()` are
+single C-level calls in Python. If the orchestrator moves to Rust or Go, the
+bottleneck disappears and both Core ABI and Component Model achieve memcpy speed.
+The correct fix for a Python orchestrator that uses the Component Model is a
+native extension or a rewritten orchestrator — not a switch to Core ABI.
+
+**Static composition is only available for all-Wasm + Component Model +
+fixed topology.** It requires all pipeline steps to be Component Model components
+and the topology to be known at build time. Mixed pipelines, Core ABI pipelines,
+and dynamic DAG pipelines cannot use it.
+
+**The long-term ecosystem trajectory favors Component Model.** WASI P2, warg/wa.dev,
+and Bytecode Alliance tooling are converging on the Component Model as the
+standard Wasm interface layer. Core ABI remains the foundational substrate but
+is not the direction of new development.
+
+**Static composition and native orchestration have different cost profiles.**
+Static composition eliminates inter-step copies by merging the pipeline at build
+time. A Rust or Go orchestrator eliminates the Python binding cost but keeps the 2N
+copies. For large tiles in a many-step pipeline, static composition reduces
+memory traffic; a Rust or Go orchestrator reduces per-copy overhead. Both improvements
+are independent and composable.
+
+---
+
+## Part 4: chonkle's Architecture and Rationale
+
+chonkle currently supports all three options: Component Model components, Core
+Wasm modules, and native Python codecs can coexist in the same pipeline. This
+section describes the choices made in this proof-of-concept and the
+acknowledged costs.
+
+### What was chosen and why
+
+**Native backend** — the Python numcodecs ecosystem provides a large catalog of
+existing codecs with hardware-accelerated implementations (SIMD, GPU, hardware
+crypto). Reimplementing them as Wasm would impose a high barrier to incremental
+adoption. Native edges also avoid copies entirely for native-to-native steps
+and have unrestricted hardware access.
+
+**Core Wasm backend** — under a Python orchestrator, `Memory.write()` and
+`Memory.read()` transfers run at ~10 GB/s. The Component Model Canonical ABI
+under wasmtime-py runs at ~1.7 MB/s for `list<u8>`. For data-intensive codecs
+in the current Python executor, Core ABI offers substantially better throughput.
+Core-to-core edges further reduce copies to one via `CoreWasmRef` deferred
+references and `ctypes.memmove`.
+
+**Component Model backend** — provides load-time interface verification,
+automated code generation via `wit-bindgen`, and language-agnostic authoring.
+WIT defines the codec contract for Component Model codecs. When the throughput
+constraint is less critical or when the orchestrator is a compiled language,
+Component Model is worth considering for new codecs.
+
+### Backend consolidation
+
+The current design does not commit to a single backend long-term. If the
+orchestrator moves to Rust, the Canonical ABI bottleneck disappears and both
+Wasm backends achieve memcpy throughput; at that point the backend distinction
+would become a toolchain and portability question rather than a performance one.
+How the three-backend design evolves from here is an open question.
+
+### Acknowledged costs
+
+- Three calling conventions in the orchestrator (`ComponentCodec`,
+  `CoreWasmCodec`, `NativeCodec`), each with its own code path.
+- No uniform interface contract: Core ABI and native codecs have no load-time
+  verification. Interface checking applies only to Component Model steps.
+- Inconsistent security boundary: native codecs run in the host process without
+  isolation; Core module linear memory is readable by the host; only Component
+  Model components have full isolation.
+- No static composition across the full pipeline. `wasm-tools compose` requires
+  all-Component-Model steps with a fixed topology.
+
+---
+
+## Appendix: Benchmark Data
+
+Performance figures cited in this document are from measurements taken in this
+repository using wasmtime-py 41 and wasmtime-rs 41 on the same `.wasm` binary
+and same buffer sizes.
+
+| Measurement                                      | Value          | Source                    |
+|--------------------------------------------------|----------------|---------------------------|
+| wasmtime-py Canonical ABI throughput (`list<u8>`)| 1.4–1.9 MB/s   | CANONICAL_ABI_PERF.md     |
+| wasmtime-py raw bench (2 MB, 3 iterations)       | ~1.7 MB/s      | bench/python-host/        |
+| wasmtime-rs typed bindings (2 MB, 3 iterations)  | ~9,100 MB/s    | bench/rust-host/          |
+| Per-step copies (any Wasm codec, any host)       | 2              | DATA_COPIES.md            |
+| Per-step copies after static composition (Component Model, fixed topology) | 2 total (all N steps) | DATA_COPIES.md |
+
+Full benchmark setup, methodology, and per-codec/per-size tables are in
+`docs/design/CANONICAL_ABI_PERF.md` and `docs/design/DATA_COPIES.md`.
diff --git a/docs/guides/COMPONENT_MODEL_C.md b/docs/guides/COMPONENT_MODEL_C.md
new file mode 100644
index 0000000..83245e3
--- /dev/null
+++ b/docs/guides/COMPONENT_MODEL_C.md
@@ -0,0 +1,271 @@
+# Building a Component Model Codec in C
+
+This guide walks through building a Component Model codec using C and the Zig build system, following the structure of the [identity-c](../../codec/identity-c/) codec.
+
+For the contract specification, see [reference/codec-contract/COMPONENT_MODEL.md](../reference/codec-contract/COMPONENT_MODEL.md).
+
+## Prerequisites
+
+- [Zig](https://ziglang.org/) (provides a C cross-compiler targeting `wasm32-wasi`)
+- [wasm-tools](https://github.com/bytecodealliance/wasm-tools) (lifts core wasm to Component Model)
+- chonkle CLI (for signature embedding): `pip install chonkle`
+
+## Project structure
+
+```
+codec/my-codec-c/
+  build.zig          # two-step build: compile C, then lift to Component Model
+  codec_impl.c       # your encode/decode implementation
+  signature.json     # port/parameter declarations embedded into the .wasm
+```
+
+Your codec also depends on shared infrastructure in `codec/shared/`:
+
+- `codec.h` / `codec.c` — WIT-generated canonical ABI bindings (types, string helpers, weak `cabi_realloc`)
+- `codec_component_type.o` — component type object from `wit-bindgen`, linked into the binary
+- `../wasi_snapshot_preview1.reactor.wasm` — WASI preview1-to-preview2 adapter used during the Component Model lift step
+
+## Step 1: Understand the generated types
+
+`codec.h` is generated by `wit-bindgen` from the WIT interface. The key types you'll work with:
+
+```c
+// A byte buffer (ptr + length).
+typedef struct { uint8_t *ptr; size_t len; } codec_list_u8_t;
+
+// A single port-map entry: (name, bytes).
+typedef struct {
+    codec_string_t f0;    // port name
+    codec_list_u8_t f1;   // port data
+} codec_tuple2_port_name_list_u8_t;
+
+// The full port-map: array of entries.
+typedef struct {
+    codec_tuple2_port_name_list_u8_t *ptr;
+    size_t len;
+} exports_chonkle_codec_transform_port_map_t;
+```
+
+String helpers from `codec.c`:
+- `codec_string_set(ret, "literal")` — point at a string literal (no copy)
+- `codec_string_dup(ret, "text")` — allocate a copy
+
+## Step 2: Write codec_impl.c
+
+Your implementation file must export two functions with these exact signatures:
+
+```c
+bool exports_chonkle_codec_transform_encode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err);
+
+bool exports_chonkle_codec_transform_decode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err);
+```
+
+Return `true` and populate `ret` on success. Return `false` and populate `err` on failure.
+
+Here is the complete `codec_impl.c` from the identity codec:
+
+```c
+#include "../shared/codec.h"
+#include <stdlib.h>
+#include <string.h>
+
+/*
+ * Override the weak cabi_realloc from codec.c so that OOM calls
+ * __builtin_trap() instead of abort(), avoiding WASI stderr imports.
+ */
+__attribute__((__export_name__("cabi_realloc")))
+void *cabi_realloc(void *ptr, size_t old_size, size_t align, size_t new_size) {
+    (void)old_size;
+    if (new_size == 0) return (void *)align;
+    void *ret = realloc(ptr, new_size);
+    if (!ret) __builtin_trap();
+    return ret;
+}
+
+/* Find a named port in the input port-map; returns NULL if absent. */
+static codec_list_u8_t *find_port(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    const char *name)
+{
+    size_t name_len = strlen(name);
+    for (size_t i = 0; i < inputs->len; i++) {
+        codec_tuple2_port_name_list_u8_t *e = &inputs->ptr[i];
+        if (e->f0.len == name_len && memcmp(e->f0.ptr, name, name_len) == 0)
+            return &e->f1;
+    }
+    return NULL;
+}
+
+/*
+ * Allocate a port-map with a single "bytes" port pointing to buf/buf_len.
+ * Ownership of buf transfers to the caller's port-map.
+ */
+static bool make_bytes_result(
+    uint8_t *buf, size_t buf_len,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    codec_tuple2_port_name_list_u8_t *entry = malloc(sizeof(*entry));
+    if (!entry) {
+        codec_string_set(err, "out of memory");
+        return false;
+    }
+    codec_string_dup(&entry->f0, "bytes");
+    entry->f1.ptr = buf;
+    entry->f1.len = buf_len;
+    ret->ptr = entry;
+    ret->len = 1;
+    return true;
+}
+
+bool exports_chonkle_codec_transform_encode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    codec_list_u8_t *bytes_port = find_port(inputs, "bytes");
+    if (!bytes_port) { codec_string_set(err, "missing port: bytes"); return false; }
+
+    size_t input_len = bytes_port->len;
+    uint8_t *out = malloc(input_len);
+    if (!out) { codec_string_set(err, "out of memory"); return false; }
+    memcpy(out, bytes_port->ptr, input_len);
+
+    return make_bytes_result(out, input_len, ret, err);
+}
+
+bool exports_chonkle_codec_transform_decode(
+    exports_chonkle_codec_transform_port_map_t *inputs,
+    exports_chonkle_codec_transform_port_map_t *ret,
+    codec_string_t *err)
+{
+    codec_list_u8_t *bytes_port = find_port(inputs, "bytes");
+    if (!bytes_port) { codec_string_set(err, "missing port: bytes"); return false; }
+
+    size_t input_len = bytes_port->len;
+    uint8_t *out = malloc(input_len);
+    if (!out) { codec_string_set(err, "out of memory"); return false; }
+    memcpy(out, bytes_port->ptr, input_len);
+
+    return make_bytes_result(out, input_len, ret, err);
+}
+```
+
+The pattern:
+
+1. **`cabi_realloc` override**: The default `cabi_realloc` in `codec.c` calls `abort()` on OOM, which requires WASI stderr imports. This override uses `__builtin_trap()` instead, avoiding that dependency. Copy this verbatim into every C codec.
+2. **`find_port` helper**: Looks up a named port in the input port-map by comparing byte strings. Copy and reuse.
+3. **`make_bytes_result` helper**: Builds a one-entry output port-map with a `"bytes"` port. Copy and reuse.
+4. **`encode` / `decode`**: Extract ports, do your transformation, return the result via `make_bytes_result`. This is the only part that changes per codec.
+
+To replace the identity passthrough with real codec logic, swap the `memcpy` with your transformation. For a codec with multiple input ports and integer parameters, see the [tiff-predictor-2-c](../../codec/tiff-predictor-2-c/) example.
+
+## Step 3: Write build.zig
+
+The build is a two-step process: compile C to a core wasm32-wasi reactor module, then lift it to a Component Model binary.
+
+```zig
+const std = @import("std");
+
+pub fn build(b: *std.Build) void {
+    // Step 1: compile C sources to a core wasm32-wasi reactor module.
+    const compile = b.addSystemCommand(&.{
+        b.graph.zig_exe,
+        "cc",
+        "--target=wasm32-wasi",
+        "-mexec-model=reactor",
+        "-fvisibility=hidden",
+        "-lc",
+        "-O2",
+    });
+
+    compile.addArg("-o");
+    const core_wasm = compile.addOutputFileArg("my-codec-core.wasm");
+    compile.addFileArg(b.path("../shared/codec.c"));
+    compile.addFileArg(b.path("codec_impl.c"));
+    // Add additional .c files here if your codec has separate algorithm files
+    compile.addFileArg(b.path("../shared/codec_component_type.o"));
+
+    // Step 2: lift the core module to a Component Model .wasm via the WASI
+    // preview1 → preview2 adapter.
+    const componentize = b.addSystemCommand(&.{ "wasm-tools", "component", "new" });
+    componentize.addFileArg(core_wasm);
+    componentize.addArg("--adapt");
+    componentize.addPrefixedFileArg(
+        "wasi_snapshot_preview1=",
+        b.path("../wasi_snapshot_preview1.reactor.wasm"),
+    );
+    componentize.addArg("-o");
+    const component = componentize.addOutputFileArg("my-codec.wasm");
+
+    b.getInstallStep().dependOn(
+        &b.addInstallFile(component, "my-codec.wasm").step,
+    );
+}
+```
+
+Key flags:
+- `--target=wasm32-wasi` — cross-compile C to WebAssembly
+- `-mexec-model=reactor` — produce a reactor (library) rather than a command (main())
+- `-fvisibility=hidden` — only explicitly exported symbols are visible
+- `codec_component_type.o` — a `wit-bindgen` artifact that embeds the component type metadata
+
+To adapt this for your codec, change the output filenames (`my-codec-core.wasm`, `my-codec.wasm`) and add any additional `.c` source files your codec needs.
+
+## Step 4: Write the signature
+
+Create `signature.json`:
+
+```json
+{
+  "codec_id": "my-codec",
+  "implementation": "my-codec-c",
+  "inputs": {"bytes": {"type": "bytes", "required": true}},
+  "outputs": {"bytes": {"type": "bytes"}}
+}
+```
+
+See [reference/codec-contract/README.md](../reference/codec-contract/README.md) for the full signature specification, including optional parameters, defaults, and `encode_only`.
+
+## Step 5: Build
+
+```bash
+cd codec/my-codec-c
+zig build
+```
+
+For optimized builds: `zig build -Doptimize=ReleaseFast` (or `ReleaseSmall`, `ReleaseSafe`).
+
+The output lands in `zig-out/my-codec.wasm`.
+
+## Step 6: Embed the signature
+
+```bash
+cp zig-out/my-codec.wasm my-codec.wasm
+chonkle embed-signature my-codec.wasm signature.json
+```
+
+## Step 7: Verify
+
+```bash
+chonkle install my-codec.wasm
+chonkle list-codecs
+```
+
+## Going further: multiple ports and integer parameters
+
+The [tiff-predictor-2-c](../../codec/tiff-predictor-2-c/) codec demonstrates a more complex interface with three input ports (`bytes`, `bytes_per_sample`, `width`). It adds a `parse_int` helper for parsing integer parameters from UTF-8 byte strings, separates the algorithm into its own `.c` file, and passes different function pointers to encode vs. decode. This is the pattern to follow for any codec with configuration parameters.
+
+## Regenerating WIT bindings
+
+The files in `codec/shared/` (`codec.h`, `codec.c`, `codec_component_type.o`) are generated by `wit-bindgen` and should only be regenerated if the WIT interface definition changes:
+
+```bash
+wit-bindgen c codec/wit/ --world codec --out-dir codec/shared/
+```
diff --git a/docs/guides/COMPONENT_MODEL_RUST.md b/docs/guides/COMPONENT_MODEL_RUST.md
new file mode 100644
index 0000000..bdfccfd
--- /dev/null
+++ b/docs/guides/COMPONENT_MODEL_RUST.md
@@ -0,0 +1,209 @@
+# Building a Component Model Codec in Rust
+
+This guide walks through building a Component Model codec using Rust and `cargo-component`, following the structure of the [zlib-rs](../../codec/zlib-rs/) codec.
+
+For the contract specification, see [reference/codec-contract/COMPONENT_MODEL.md](../reference/codec-contract/COMPONENT_MODEL.md).
+
+## Prerequisites
+
+- Rust (edition 2024) with the `wasm32-wasip2` target: `rustup target add wasm32-wasip2`
+- [cargo-component](https://github.com/bytecodealliance/cargo-component): `cargo install cargo-component`
+- chonkle CLI (for signature embedding): `pip install chonkle`
+
+## Project structure
+
+```
+codec/my-codec-rs/
+  Cargo.toml        # crate config: cdylib + component metadata
+  src/lib.rs        # codec implementation
+  wit/world.wit     # WIT interface definition (same for all codecs)
+  signature.json    # port/parameter declarations embedded into the .wasm
+```
+
+## Step 1: Create the WIT world file
+
+Create `wit/world.wit` with the canonical codec interface. This file is identical for every Component Model codec:
+
+```wit
+package chonkle:codec@0.1.0;
+
+interface transform {
+    /// A named byte buffer — one port in the port map.
+    type port-name = string;
+    type port-map = list<tuple<port-name, list<u8>>>;
+
+    encode: func(inputs: port-map) -> result<port-map, string>;
+    decode: func(inputs: port-map) -> result<port-map, string>;
+}
+
+world codec {
+    export transform;
+}
+```
+
+This declares two functions, `encode` and `decode`, that take a list of named byte buffers (the port-map) and return a transformed port-map or an error string. See [docs/wasm/WIT.md](../wasm/WIT.md) for WIT syntax background.
+
+## Step 2: Configure Cargo.toml
+
+```toml
+[package]
+name = "my-codec"
+version = "0.1.0"
+edition = "2024"
+
+[lib]
+crate-type = ["cdylib"]
+
+[dependencies]
+wit-bindgen-rt = { version = "0.44.0", features = ["bitflags"] }
+# Add your codec library here, e.g.:
+# flate2 = { version = "1", default-features = false, features = ["rust_backend"] }
+
+[package.metadata.component]
+package = "chonkle:codec"
+
+[package.metadata.component.dependencies]
+```
+
+Key points:
+- `crate-type = ["cdylib"]` produces a WebAssembly dynamic library.
+- `wit-bindgen-rt` provides the runtime support for the generated WIT bindings.
+- `[package.metadata.component]` tells `cargo-component` which WIT package this crate implements.
+- Use `default-features = false` and pure-Rust backends for dependencies where possible, since native C libraries won't cross-compile to wasm32.
+
+## Step 3: Implement encode and decode
+
+`cargo-component` auto-generates a `bindings` module from your WIT file. Your code imports types from it and implements the `Guest` trait.
+
+Here is the complete `src/lib.rs` from the zlib codec:
+
+```rust
+#[allow(warnings)]
+mod bindings;
+
+use bindings::exports::chonkle::codec::transform::{Guest, PortMap};
+use flate2::read::ZlibDecoder;
+use flate2::write::ZlibEncoder;
+use flate2::Compression;
+use std::io::{Read, Write};
+
+struct Component;
+
+fn find_port<'a>(inputs: &'a PortMap, name: &str) -> Option<&'a Vec<u8>> {
+    inputs.iter().find(|(k, _)| k == name).map(|(_, v)| v)
+}
+
+impl Guest for Component {
+    fn encode(inputs: PortMap) -> Result<PortMap, String> {
+        let bytes = find_port(&inputs, "bytes")
+            .ok_or_else(|| "missing port: bytes".to_string())?;
+
+        let level = find_port(&inputs, "level")
+            .and_then(|v| std::str::from_utf8(v).ok())
+            .and_then(|s| s.trim().parse::<u32>().ok())
+            .unwrap_or(6)
+            .min(9);
+
+        let mut enc = ZlibEncoder::new(Vec::new(), Compression::new(level));
+        enc.write_all(bytes).map_err(|e| e.to_string())?;
+        let compressed = enc.finish().map_err(|e| e.to_string())?;
+
+        Ok(vec![("bytes".to_string(), compressed)])
+    }
+
+    fn decode(inputs: PortMap) -> Result<PortMap, String> {
+        let bytes = find_port(&inputs, "bytes")
+            .ok_or_else(|| "missing port: bytes".to_string())?;
+
+        let mut dec = ZlibDecoder::new(bytes.as_slice());
+        let mut out = Vec::new();
+        dec.read_to_end(&mut out).map_err(|e| e.to_string())?;
+
+        Ok(vec![("bytes".to_string(), out)])
+    }
+}
+
+bindings::export!(Component with_types_in bindings);
+```
+
+The pattern:
+
+1. **Boilerplate**: `mod bindings`, `struct Component`, `bindings::export!()` — same for every codec.
+2. **`find_port` helper**: Looks up a named port in the input port-map. You'll reuse this pattern in every codec.
+3. **`encode`**: Extract the required `bytes` port. Read optional parameters (here, `level`). Parameters arrive as UTF-8 byte strings and need parsing. Do the transformation. Return the output port-map.
+4. **`decode`**: Extract `bytes`, reverse the transformation, return.
+5. **Error handling**: Return `Err(String)` to signal failure. The host surfaces this as a codec error.
+
+## Step 4: Write the signature
+
+Create `signature.json` declaring the codec's ports and parameters:
+
+```json
+{
+  "codec_id": "my-codec",
+  "implementation": "my-codec-rs",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 6, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+
+- `codec_id` — logical name used in pipeline JSON to reference this codec.
+- `implementation` — identifies this specific build. Used as the filename in the codec store.
+- `inputs` — declares each input port. `required: false` ports can be omitted by pipeline steps; the host uses `default` when they are. `encode_only: true` ports are ignored during decode.
+- `outputs` — declares each output port.
+
+See [reference/codec-contract/README.md](../reference/codec-contract/README.md) for the full signature specification.
+
+## Step 5: Build
+
+```bash
+cd codec/my-codec-rs
+cargo component build --target wasm32-wasip2 --release
+```
+
+This compiles the Rust code, generates WIT bindings, and produces a Component Model `.wasm` binary. The output path is printed by cargo; for zlib-rs it lands at `target/wasm32-wasip1/release/zlib.wasm`.
+
+## Step 6: Embed the signature
+
+```bash
+cp target/wasm32-wasip1/release/my-codec.wasm my-codec.wasm
+chonkle embed-signature my-codec.wasm signature.json
+```
+
+This writes `signature.json` into the `.wasm` binary as a `chonkle:signature` custom section. The host reads this section at runtime to discover the codec's ports and parameters.
+
+## Step 7: Verify
+
+Install the codec into the local codec store and confirm it appears:
+
+```bash
+chonkle install my-codec.wasm
+chonkle list-codecs
+```
+
+To test a roundtrip, create a minimal pipeline JSON and run it:
+
+```json
+{
+  "steps": [
+    {
+      "codec_id": "my-codec",
+      "inputs": {"bytes": "input"},
+      "outputs": {"bytes": "output"}
+    }
+  ]
+}
+```
+
+## Common patterns
+
+**Optional parameters with defaults**: Declare the port with `required: false` and a `default` value in the signature. In the Rust code, use `unwrap_or()` when parsing the port value.
+
+**Encode-only parameters**: Parameters like compression level only apply during encoding. Mark them `encode_only: true` in the signature; the host strips them from the port-map during decode, so your decode function never sees them.
+
+**Multiple output ports**: Return multiple entries in the output `Vec`. Each `(name, bytes)` pair becomes a named output port that downstream steps can wire to.
diff --git a/docs/guides/CORE.md b/docs/guides/CORE.md
new file mode 100644
index 0000000..ab28c48
--- /dev/null
+++ b/docs/guides/CORE.md
@@ -0,0 +1,231 @@
+# Building a Core Wasm Codec
+
+This guide walks through building a Core Wasm codec using C and the Zig build system, following the structure of the [identity-core-c](../../codec/identity-core-c/) codec.
+
+For the contract specification, see [reference/codec-contract/CORE.md](../reference/codec-contract/CORE.md).
+
+## When to use Core Wasm
+
+Core Wasm bypasses the Component Model canonical ABI. In the Python host, this means ~10 GB/s host-to-module transfer vs ~1.7 MB/s for Component Model codecs, because data is written directly into linear memory instead of going through the canonical ABI's lifting and lowering. The trade-off is a lower-level interface: you manage linear memory allocation and work with a binary port-map wire format.
+
+Choose Core Wasm when transfer throughput matters (large buffers, hot paths). Choose Component Model when development simplicity matters. See [design/DATA_COPIES.md](../design/DATA_COPIES.md) and [design/CANONICAL_ABI_PERF.md](../design/CANONICAL_ABI_PERF.md) for the full performance analysis.
+
+## Prerequisites
+
+- [Zig](https://ziglang.org/) (provides a C cross-compiler targeting `wasm32-wasi`)
+- chonkle CLI (for signature embedding): `pip install chonkle`
+
+Note: no `wasm-tools` needed — Core Wasm codecs are not lifted to Component Model.
+
+## Project structure
+
+```
+codec/my-codec-core-c/
+  build.zig          # single-step build: compile C to wasm32-wasi reactor
+  codec_impl.c       # your alloc/dealloc/encode/decode exports
+  signature.json     # port/parameter declarations embedded into the .wasm
+```
+
+Your codec depends on shared infrastructure in `codec/shared/`:
+
+- `core_abi.h` / `core_abi.c` — port-map parse, serialize, find, and pack helpers
+
+## Step 1: Understand the required exports
+
+A Core Wasm codec must export five symbols:
+
+| Export | Signature | Description |
+|---|---|---|
+| `memory` | Memory | Linear memory (provided by the wasm toolchain) |
+| `alloc` | `(size: i32) -> i32` | Allocate `size` bytes, return pointer (or 0) |
+| `dealloc` | `(ptr: i32, size: i32)` | Free a buffer |
+| `encode` | `(port_map_ptr: i32, port_map_len: i32) -> i64` | Encode operation |
+| `decode` | `(port_map_ptr: i32, port_map_len: i32) -> i64` | Decode operation |
+
+The host calls `alloc` to place input data in the module's memory. `encode`/`decode` take ownership of the input buffer (must free it) and return a packed `i64` of `(output_ptr << 32) | output_len`. The host reads the output, then calls `dealloc` to free it.
+
+## Step 2: Understand the port-map wire format
+
+Inputs and outputs are serialized as a binary port-map:
+
+```
+u32: entry_count
+For each entry:
+  u32: name_len
+  u8[name_len]: name (UTF-8)
+  u32: data_len
+  u8[data_len]: data
+```
+
+All integers are little-endian. No padding between fields.
+
+The `core_abi.h` helpers handle this for you:
+
+```c
+// Parse a serialized port-map. Returned entries point into the input buffer
+// (zero-copy). Returns count=0 on error.
+core_abi_port_map_t core_abi_parse_port_map(const uint8_t *buf, uint32_t len);
+
+// Look up a port by name.
+const core_abi_port_t *core_abi_find_port(const core_abi_port_map_t *pm, const char *name);
+
+// Serialize a port-map to a new malloc'd buffer.
+uint8_t *core_abi_serialize_port_map(const core_abi_port_map_t *pm, uint32_t *out_len);
+
+// Pack (ptr, len) into a single i64 return value.
+int64_t core_abi_pack_result(uint32_t ptr, uint32_t len);
+
+// Free a parsed port-map's entries array.
+void core_abi_free_port_map(core_abi_port_map_t *pm);
+```
+
+## Step 3: Write codec_impl.c
+
+Here is the complete `codec_impl.c` from the identity-core codec:
+
+```c
+#include "../shared/core_abi.h"
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+
+__attribute__((__export_name__("alloc")))
+int32_t codec_alloc(int32_t size) {
+    void *ptr = malloc((size_t)size);
+    return (int32_t)(uintptr_t)ptr;
+}
+
+__attribute__((__export_name__("dealloc")))
+void codec_dealloc(int32_t ptr, int32_t size) {
+    (void)size;
+    free((void *)(uintptr_t)ptr);
+}
+
+static int64_t transform(int32_t pm_ptr, int32_t pm_len) {
+    uint8_t *input_buf = (uint8_t *)(uintptr_t)pm_ptr;
+    core_abi_port_map_t pm = core_abi_parse_port_map(input_buf, (uint32_t)pm_len);
+
+    if (pm.count == 0) {
+        free(input_buf);
+        return CORE_ABI_ERROR;
+    }
+
+    const core_abi_port_t *bytes_port = core_abi_find_port(&pm, "bytes");
+    if (!bytes_port) {
+        core_abi_free_port_map(&pm);
+        free(input_buf);
+        return CORE_ABI_ERROR;
+    }
+
+    /* Copy input bytes before freeing the input buffer. */
+    uint32_t out_data_len = bytes_port->data_len;
+    uint8_t *out_data = malloc(out_data_len);
+    if (!out_data) {
+        core_abi_free_port_map(&pm);
+        free(input_buf);
+        return CORE_ABI_ERROR;
+    }
+    memcpy(out_data, bytes_port->data, out_data_len);
+
+    core_abi_free_port_map(&pm);
+    free(input_buf);
+
+    /* Build output port-map with a single "bytes" port. */
+    core_abi_port_t out_entry;
+    out_entry.name = "bytes";
+    out_entry.name_len = 5;
+    out_entry.data = out_data;
+    out_entry.data_len = out_data_len;
+
+    core_abi_port_map_t out_pm;
+    out_pm.entries = &out_entry;
+    out_pm.count = 1;
+
+    uint32_t ser_len;
+    uint8_t *ser_buf = core_abi_serialize_port_map(&out_pm, &ser_len);
+    free(out_data);
+
+    if (!ser_buf) return CORE_ABI_ERROR;
+
+    return core_abi_pack_result((uint32_t)(uintptr_t)ser_buf, ser_len);
+}
+
+__attribute__((__export_name__("encode")))
+int64_t encode(int32_t pm_ptr, int32_t pm_len) {
+    return transform(pm_ptr, pm_len);
+}
+
+__attribute__((__export_name__("decode")))
+int64_t decode(int32_t pm_ptr, int32_t pm_len) {
+    return transform(pm_ptr, pm_len);
+}
+```
+
+The pattern:
+
+1. **`alloc` / `dealloc`**: Thin wrappers around `malloc`/`free` with the `__export_name__` attribute. Copy verbatim.
+2. **Parse input**: Cast the pointer, call `core_abi_parse_port_map`, look up ports with `core_abi_find_port`.
+3. **Memory ownership**: The codec owns the input buffer — copy any data you need, then `free(input_buf)` and `core_abi_free_port_map`. The parsed port entries point into the input buffer (zero-copy parse), so you must copy data before freeing.
+4. **Do the transformation**: Replace the `memcpy` with your codec logic.
+5. **Build output**: Construct a `core_abi_port_map_t`, serialize it with `core_abi_serialize_port_map`, free intermediate buffers, and return `core_abi_pack_result(ptr, len)`.
+6. **Error handling**: Return `CORE_ABI_ERROR` (0) on any failure.
+
+## Step 4: Write build.zig
+
+The build is a single step — no Component Model lifting:
+
+```zig
+const std = @import("std");
+
+pub fn build(b: *std.Build) void {
+    const compile = b.addSystemCommand(&.{
+        b.graph.zig_exe,
+        "cc",
+        "--target=wasm32-wasi",
+        "-mexec-model=reactor",
+        "-fvisibility=hidden",
+        "-lc",
+        "-O2",
+    });
+
+    compile.addArg("-o");
+    const core_wasm = compile.addOutputFileArg("my-codec-core.wasm");
+    compile.addFileArg(b.path("../shared/core_abi.c"));
+    compile.addFileArg(b.path("codec_impl.c"));
+
+    b.getInstallStep().dependOn(
+        &b.addInstallFile(core_wasm, "my-codec-core.wasm").step,
+    );
+}
+```
+
+Compared to the Component Model C build: no `codec.c`, no `codec_component_type.o`, no `wasm-tools component new` step, no WASI adapter. Just `core_abi.c` + your codec.
+
+## Step 5: Write the signature
+
+```json
+{
+  "codec_id": "my-codec",
+  "implementation": "my-codec-core-c",
+  "inputs": {"bytes": {"type": "bytes", "required": true}},
+  "outputs": {"bytes": {"type": "bytes"}}
+}
+```
+
+Same format as Component Model signatures.
+
+## Step 6: Build and embed
+
+```bash
+cd codec/my-codec-core-c
+zig build
+cp zig-out/my-codec-core.wasm my-codec-core.wasm
+chonkle embed-signature my-codec-core.wasm signature.json
+```
+
+## Step 7: Verify
+
+```bash
+chonkle install my-codec-core.wasm
+chonkle list-codecs
+```
diff --git a/docs/guides/NATIVE.md b/docs/guides/NATIVE.md
new file mode 100644
index 0000000..cf408a6
--- /dev/null
+++ b/docs/guides/NATIVE.md
@@ -0,0 +1,169 @@
+# Adding a Native (numcodecs) Codec
+
+Native codecs wrap existing [numcodecs](https://numcodecs.readthedocs.io/) codec objects. You write a JSON signature file; chonkle handles the rest. No build step, no Wasm.
+
+For the contract specification, see [reference/codec-contract/NATIVE.md](../reference/codec-contract/NATIVE.md).
+
+## When to use native codecs
+
+When numcodecs already implements the codec you need and you don't need the portability or sandboxing of Wasm. Native codecs run as regular Python/C code in-process.
+
+## Prerequisites
+
+- chonkle with the native extra: `pip install chonkle[native]` (brings `numcodecs` and `numpy`)
+
+## What you create
+
+A single JSON file at `src/chonkle/signatures/numcodecs/<codec_id>.json`. That is all.
+
+## Step 1: Determine the calling convention for each direction
+
+The `native` block in the signature JSON specifies a calling recipe for each direction (`encode` and `decode`) independently. Each recipe has a `mode`:
+
+**`"bytes"`** — The `bytes` port value is passed directly to `codec.encode()` / `codec.decode()` as raw bytes. Use this for compression codecs that operate on byte streams (zlib, gzip, bz2, lzma, zstd, lz4, blosc, checksum codecs).
+
+**`"ndarray"`** — The `bytes` port is converted to a numpy ndarray using a dtype from a named port before being passed to the codec. The result is converted back to bytes afterward so it conforms to the port-map interface shared by all codec backends (`list[tuple[str, bytes]]`). When two native ndarray codecs are chained, this means a redundant `tobytes()` / `frombuffer()` round-trip at each step boundary; `frombuffer` is zero-copy but `tobytes()` allocates and copies. In practice this is cheap relative to the codec work itself. Use this mode for array-level codecs whose numcodecs `encode()` expects an ndarray (delta, shuffle, bitround, astype, packbits, categorize).
+
+Most codecs use the same mode for both directions, but this is not required — a codec can use `"bytes"` for one direction and `"ndarray"` for the other. Each direction can also reference a different dtype port, which is how asymmetric-dtype codecs (astype, packbits, categorize) work.
+
+To decide the mode for each direction: check the numcodecs documentation for your codec. If its `encode()` / `decode()` method expects `bytes` or `bytearray`, use `"bytes"` for that direction. If it expects an `ndarray`, use `"ndarray"`.
+
+## Step 2: Write the signature JSON
+
+Every signature has two parts: the standard port-map interface (`codec_id`, `implementation`, `inputs`, `outputs`) shared with wasm codecs, and the `native` block that controls the calling convention.
+
+### Signature fields
+
+Standard fields:
+
+- **`codec_id`** — logical codec identifier used in pipeline JSON (e.g. `"zlib"`, `"delta"`).
+- **`implementation`** — must be `"numcodecs.<id>"` where `<id>` is the numcodecs codec ID passed to `numcodecs.get_codec()`.
+- **`inputs`** — must include a `"bytes"` port (`required: true`). Additional ports become parameters available to the `native` block.
+- **`outputs`** — must include a `"bytes"` port.
+
+Port descriptor fields: `type` (required), `required` (default `true`), `default`, `encode_only` (default `false`), `decode_only` (default `false`). See [reference/codec-contract/README.md](../reference/codec-contract/README.md) for the full specification.
+
+Native block fields:
+
+- **`constructor_ports`** — list of input port names whose JSON-decoded values become kwargs to `numcodecs.get_codec()`. Ports not in this list are not passed to the constructor. Ports listed here but absent from the port-map (e.g. encode-only ports during decode) are silently skipped.
+- **`encode`** / **`decode`** — calling recipe per direction. `mode` is `"bytes"` or `"ndarray"`. When mode is `"ndarray"`, `dtype_port` (required) names the input port supplying the numpy dtype string for `np.frombuffer`.
+
+### Examples
+
+#### Bytes-mode codec (zlib)
+
+`src/chonkle/signatures/numcodecs/zlib.json`:
+
+```json
+{
+  "codec_id": "zlib",
+  "implementation": "numcodecs.zlib",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["level"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
+```
+
+`constructor_ports` lists the input port names whose values become kwargs to `numcodecs.get_codec()`. Here, `level` is passed as `numcodecs.get_codec({"id": "zlib", "level": 6})`. The `bytes` port is never a constructor kwarg.
+
+#### Symmetric ndarray-mode codec (delta)
+
+`src/chonkle/signatures/numcodecs/delta.json`:
+
+```json
+{
+  "codec_id": "delta",
+  "implementation": "numcodecs.delta",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
+```
+
+`dtype_port` names the input port supplying the numpy dtype string for `np.frombuffer`. Both directions use the same dtype, so both point to `"dtype"`.
+
+#### Ndarray-mode codec where dtype is not a constructor arg (shuffle)
+
+`src/chonkle/signatures/numcodecs/shuffle.json`:
+
+```json
+{
+  "codec_id": "shuffle",
+  "implementation": "numcodecs.shuffle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": [],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
+```
+
+`Shuffle()` takes no constructor args — `constructor_ports` is empty. The `dtype` port is used only for the `np.frombuffer` conversion, not passed to the constructor.
+
+#### Asymmetric-dtype codec (astype)
+
+`src/chonkle/signatures/numcodecs/astype.json`:
+
+```json
+{
+  "codec_id": "astype",
+  "implementation": "numcodecs.astype",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "encode_dtype": {"type": "string", "required": true},
+    "decode_dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["encode_dtype", "decode_dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "decode_dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "encode_dtype"}
+  }
+}
+```
+
+AsType converts between dtypes. `encode_dtype` is the output dtype of encode; `decode_dtype` is the output dtype of decode (matching the numcodecs constructor args). The `dtype_port` references are swapped: during encode, the input buffer has the decode dtype (the original type); during decode, the input buffer has the encode dtype (the encoded type).
+
+## Step 3: Verify
+
+```bash
+chonkle codecs
+```
+
+The new codec should appear in the output. To test a roundtrip, use the Python API directly:
+
+```python
+from chonkle.codecs.native import NativeCodec
+
+codec = NativeCodec("zlib")
+encoded = codec.call("encode", [("bytes", b"hello world"), ("level", b"6")])
+decoded = codec.call("decode", encoded)
+assert decoded[0][1] == b"hello world"
+```
diff --git a/docs/guides/README.md b/docs/guides/README.md
new file mode 100644
index 0000000..1b7883c
--- /dev/null
+++ b/docs/guides/README.md
@@ -0,0 +1,20 @@
+# Guides
+
+Step-by-step guides for building chonkle codecs. For contract specifications, see [reference/codec-contract/](../reference/codec-contract/).
+
+## Building a codec
+
+- [Component Model codec in Rust](COMPONENT_MODEL_RUST.md) — recommended starting point
+- [Component Model codec in C](COMPONENT_MODEL_C.md) — C with Zig build system
+- [Core Wasm codec](CORE.md) — C with Zig build system, no Component Model
+- [Native codec](NATIVE.md) — numcodecs wrapper (no build step)
+
+## Which type should I build?
+
+**Component Model (Rust)** if you're writing a new codec from scratch. Rust's `cargo-component` handles all the WIT bindings and Component Model lifting automatically, so you write idiomatic Rust and get a `.wasm` binary out. This is the simplest path.
+
+**Component Model (C)** if you have existing C code you want to wrap, or need to avoid Rust. The Zig build system cross-compiles C to `wasm32-wasi` and `wasm-tools` lifts the result to a Component Model binary. You get the same Component Model interface as the Rust path but manage memory and ABI glue manually.
+
+**Core Wasm** if you need maximum host-to-module transfer throughput in the Python host. Core Wasm bypasses the Component Model canonical ABI, which is the primary bottleneck in the Python host's data path. The trade-off is a lower-level interface: you implement `alloc`/`dealloc` and work with a binary port-map wire format directly. See [design/DATA_COPIES.md](../design/DATA_COPIES.md) and [design/CANONICAL_ABI_PERF.md](../design/CANONICAL_ABI_PERF.md) for the performance context.
+
+**Native** if [numcodecs](https://numcodecs.readthedocs.io/) already implements the codec you need. You write a JSON signature file and chonkle wraps the numcodecs codec automatically. No build step, no Wasm. Requires `pip install chonkle[native]`.
diff --git a/docs/reference/CODEC_RESOLUTION.md b/docs/reference/CODEC_RESOLUTION.md
new file mode 100644
index 0000000..4c10602
--- /dev/null
+++ b/docs/reference/CODEC_RESOLUTION.md
@@ -0,0 +1,30 @@
+# Codec Resolution
+
+The `Resolver` maps codec_ids to `Codec` instances via a resolution chain. Each step is tried in order; the first match wins.
+
+## Resolution chain
+
+1. **Explicit paths** — a `--codec-path` CLI flag or `paths` dict mapping a codec_id directly to a `.wasm` file. Useful for testing.
+2. **Force sources** — a `--source ID=URI` CLI flag or `force_sources` dict. Downloads the binary from the URI unconditionally, installs it into the store (overwriting any existing entry with the same implementation name), and returns the newly installed codec. Use this to fetch a specific remote build regardless of what is already in the store.
+3. **Per-codec overrides** — an `overrides` dict mapping a codec_id to a specific implementation name, bypassing the preference system entirely.
+4. **Local codec store and native** — all implementations for the codec_id are collected from the store and from bundled native signatures, then selected by backend preference.
+5. **Pipeline sources** — if no local match is found, a `sources` URI from the pipeline JSON is downloaded, installed into the store, and instantiated.
+
+## Backend preference
+
+When step 4 finds candidates, the resolver walks an ordered preference list (default: `["native", "core", "component"]`) and returns the first candidate whose backend matches. This selects between backend *types*, not between implementations of the same type.
+
+Preference is strict: if no available backend matches the preference list, the resolver raises `ValueError`.
+
+## Multiple implementations of the same backend
+
+If the store contains more than one implementation of the same backend for a given codec_id (e.g. two different core wasm builds), the resolver returns whichever appears first in the store index. The store index is built by scanning `.wasm` files in sorted filename order, so the implementation whose filename sorts first alphabetically wins. This is a deterministic but arbitrary tiebreak — use an explicit override to select a specific implementation when it matters.
+
+## Binary detection
+
+`detect_codec_type()` reads the 8-byte wasm header to route `.wasm` files to the correct backend:
+
+- Core module: layer field `01 00 00 00`
+- Component: layer field `0d 00 01 00`
+
+See [codec-contract/COMPONENT_MODEL.md](codec-contract/COMPONENT_MODEL.md) and [codec-contract/CORE.md](codec-contract/CORE.md) for per-backend details.
diff --git a/docs/reference/PIPELINE_SCHEMA.md b/docs/reference/PIPELINE_SCHEMA.md
new file mode 100644
index 0000000..b3b97f7
--- /dev/null
+++ b/docs/reference/PIPELINE_SCHEMA.md
@@ -0,0 +1,169 @@
+# Pipeline JSON Schema
+
+A pipeline is a DAG of codec steps defined in a JSON file. A pipeline JSON is an *implementation* -- it describes the DAG, the wiring, and the baked-in constants. It is not a codec signature, and its schema is not the same as a signature's.
+
+However, a pipeline is conceptually a codec: a codec signature for the pipeline can be derived from the pipeline JSON. The input names and types come directly from the `inputs` field; constants appear as optional inputs with their baked-in value as the default; and output types are inferred by tracing each output wiring reference back through the step signatures.
+
+## Pipeline Object
+
+```json
+{
+  "codec_id": "cog-decode",
+  "direction": "decode",
+  "sources": {
+    "tiff-predictor-2": "oci://ghcr.io/nasa-impact/codec-tiff-predictor-2-c:v1.0"
+  },
+  "inputs": {"bytes": {"type": "bytes"}},
+  "constants": {
+    "bytes_per_sample": {"type": "int", "value": 2},
+    "width": {"type": "int", "value": 1024}
+  },
+  "outputs": {"bytes": "predictor2.bytes"},
+  "steps": {
+    "zlib": {
+      "codec_id": "zlib",
+      "inputs": {"bytes": "input.bytes"}
+    },
+    "predictor2": {
+      "codec_id": "tiff-predictor-2",
+      "inputs": {
+        "bytes": "zlib.bytes",
+        "bytes_per_sample": "constant.bytes_per_sample",
+        "width": "constant.width"
+      }
+    }
+  }
+}
+```
+
+### `codec_id` (string, required)
+
+The canonical identifier for this pipeline. A pipeline is itself a codec per protospec. Used in any derived codec signature and in a future registry for resolving this pipeline as a step in outer pipelines.
+
+### `direction` (string, required)
+
+`"encode"` or `"decode"`. Declares which direction this pipeline definition represents. Pass a `direction` to `run()` that differs from `pipeline.direction` to execute the pipeline in the opposite direction (inverted step order, opposite encode/decode function).
+
+### `sources` (object, optional)
+
+Maps codec_ids to download URIs for codec artifacts. Supported URI schemes: `file://`, `https://`, `oci://` (`http://` raises `ValueError`). These are advisory fetch hints for the `Resolver`: if a codec_id is not found in the local store or via other resolution mechanisms, the resolver downloads from the URI listed here.
+
+### `inputs` (object, required)
+
+The pipeline's public input ports -- what callers must supply. Each key is a port name; each value is a descriptor:
+
+- `type` (string, required): the semantic type of the port. Valid values follow the protospec type vocabulary (see the [Types section of the codec inventory](https://github.com/cylf-dev/protospec/blob/main/codec-inventory.md#types)): `"bytes"`, `"string"`, `"int"`, `"uint"`, `"float"`, `"bool"`, `"uint[]"`, `"dtype_desc"`.
+- `encode_only` (bool, optional): if true, this input is only meaningful during encoding. It is omitted from the port-map during decode, and any derived codec signature will carry this annotation so that outer pipelines can handle it correctly.
+- `decode_only` (bool, optional): if true, this input is only meaningful during decoding. It is omitted from the port-map during encode. The symmetric counterpart to `encode_only`.
+
+`required` and `default` are absent (unlike codec signature inputs, which carry both). All pipeline inputs are caller-supplied -- there is no fallback at the pipeline boundary. Parameters with baked-in values belong in `constants`.
+
+### `constants` (object, optional)
+
+Values baked into the pipeline definition. Constants are never supplied by the caller. Each key is a constant name; each value is a descriptor:
+
+- `type` (string, required): determines how the value is serialized into the port-map. Valid values: `"bytes"`, `"string"`, `"int"`, `"uint"`, `"float"`, `"bool"`, `"uint[]"`, `"dtype_desc"`.
+- `value` (required): the constant's value.
+
+Constants use the `constant.<name>` wiring namespace, which is distinct from `input.<name>`. During pipeline inversion, pipeline inputs become pipeline outputs (and vice versa), but constants remain constants -- available as configuration in both directions.
+
+### `outputs` (object, required)
+
+Maps each pipeline output port name to the step port that produces it. The key is the name callers use to retrieve results from the returned port-map; the value is a wiring reference of the form `step_name.port_name`. An output port may be renamed relative to the source step's port name. Types are not declared here -- they are derived from the source step's codec signature during validation.
+
+### `steps` (object, required)
+
+A named map of step objects. Keys are step names (unique DAG node identifiers used in wiring references). Order in JSON does not matter -- topological sort determines execution order. See [Steps](#steps) below.
+
+## Steps
+
+Each step invokes a codec. The step's `codec_id` resolves to a codec signature, which defines the port names, types, and encode-only annotations that the step's wiring must satisfy.
+
+```json
+"steps": {
+  "zlib": {
+    "codec_id": "zlib",
+    "inputs": {"bytes": "input.bytes"}
+  },
+  "predictor2": {
+    "codec_id": "tiff-predictor-2",
+    "inputs": {
+      "bytes": "zlib.bytes",
+      "bytes_per_sample": "constant.bytes_per_sample",
+      "width": "constant.width"
+    }
+  }
+}
+```
+
+Each key in the `steps` object is the step name -- the unique DAG node identifier used in wiring references (`step_name.port_name`). Multiple steps may share a `codec_id` but must have distinct keys.
+
+### `codec_id` (string, required)
+
+The logical codec identifier. The `Resolver` maps this to a `Codec` instance. Matched against the codec signature's `codec_id` during validation.
+
+### `inputs` (object, required)
+
+Maps each codec input port name to a wiring reference. The key is the port name as declared in the codec signature; the value is a wiring reference (see [Wiring References](#wiring-references)). No type information is included—types come from the codec signature.
+
+## Wiring References
+
+A wiring reference is a dot-notation string identifying the source of a value:
+
+| Form | Meaning |
+|---|---|
+| `input.<port>` | A pipeline-level input port |
+| `constant.<name>` | A pipeline-level constant |
+| `<step_name>.<port>` | An output port of a named step |
+
+Wiring references are parsed into `WiringRef` objects at pipeline construction time. Step existence and input/constant refs are validated at parse time. Step output port refs (references to `<step_name>.<port>`) are validated at `prepare()` time against codec signatures.
+
+## Codec Type Detection
+
+The codec type for each step is not declared in the pipeline schema. It is detected from the resolved binary by reading the 8-byte wasm header:
+
+- `.wasm` with core version header (`01 00 00 00`) -- `CoreWasmCodec`
+- `.wasm` with component version header (`0d 00 01 00`) -- `ComponentCodec`
+- No wasm binary (numcodecs registry) -- `NativeCodec`
+
+## Divergences from the Protospec
+
+The protospec pipeline format is the upstream reference. The following changes were made deliberately.
+
+### `sources` (chonkle addition)
+
+**Protospec**: no equivalent (codec resolution is implicit via registry)
+
+**Ours**: `"sources": {"codec_id": "oci://..."}` at the pipeline level
+
+The protospec assumes codecs are resolved from a registry by `codec_id`. Chonkle doesn't have a registry yet, so `sources` gives the resolver advisory download hints for codecs not available locally.
+
+### `"codec"` renamed to `"codec_id"` in steps
+
+**Protospec**: `"codec": "zstd"`
+
+**Ours**: `"codec_id": "zstd"`
+
+`codec_id` is the consistent term for a codec identifier throughout the rest of the schema (top-level `codec_id`, codec signatures). Using `codec` in steps would be the only place the shorter form appeared.
+
+### No `outputs` on steps
+
+**Protospec**: steps have an `"outputs"` field that aliases output port names, e.g. `"outputs": {"bytes": "raw_uints"}`. Other steps reference the alias (`decode_varint.raw_uints`), not the codec signature name.
+
+**Ours**: steps have only `inputs`. Output ports are referenced by their codec signature names directly (e.g., `decode_varint.bytes`). Output port names are validated at `prepare()` time against the codec signature.
+
+### `encode_only` on pipeline inputs
+
+**Protospec**: pipeline inputs have only `type`; `encode_only` does not appear on pipeline inputs in any protospec example
+
+**Ours**: `encode_only` is permitted on pipeline inputs
+
+A pipeline's derived codec signature needs `encode_only` annotations so that outer pipelines know which inputs to omit during decode. The protospec does not show this on pipeline inputs, but it is a logical extension of the same property on codec signature inputs.
+
+### `decode_only` on codec signature and pipeline inputs
+
+**Protospec**: no `decode_only` property exists in the signature model
+
+**Ours**: `decode_only` is supported on codec signature inputs and pipeline inputs
+
+The protospec has `encode_only` but no symmetric counterpart for decode. Chonkle adds `decode_only` to handle codecs with inputs meaningful only during decoding (e.g. packbits, where `decode_dtype` is only needed during decode). The executor omits `decode_only` ports from the port-map during encode, mirroring the `encode_only` behavior during decode. See the [signature model gap analysis](protospec/PROTOSPEC_NOTES.md#no-decode_only-property) for catalog examples where the protospec lacks this capability.
diff --git a/docs/reference/codec-contract/COMPONENT_MODEL.md b/docs/reference/codec-contract/COMPONENT_MODEL.md
new file mode 100644
index 0000000..4cc27c4
--- /dev/null
+++ b/docs/reference/codec-contract/COMPONENT_MODEL.md
@@ -0,0 +1,33 @@
+# Component Model Wasm
+
+Component Model codecs implement the `chonkle:codec/transform@0.1.0` WIT interface. For build guides, see [Rust](../../guides/COMPONENT_MODEL_RUST.md) or [C](../../guides/COMPONENT_MODEL_C.md). The toolchain (e.g. `cargo-component`, `wit-bindgen`, `wasm-tools component new`) generates all memory management and canonical ABI glue. Codec authors work only with high-level types.
+
+## WIT interface
+
+A component must export the `chonkle:codec/transform` interface. The full WIT definition (from `codec/wit/codec.wit`):
+
+```wit
+package chonkle:codec@0.1.0;
+
+interface transform {
+    type port-name = string;
+    type port-map = list<tuple<port-name, list<u8>>>;
+
+    encode: func(inputs: port-map) -> result<port-map, string>;
+    decode: func(inputs: port-map) -> result<port-map, string>;
+}
+
+world codec {
+    export transform;
+}
+```
+
+The toolchain embeds this under the key `"chonkle:codec/transform@0.1.0"` — the host looks up this exact key in the component's exports. Components must target **WASIp2**. Return `Err(string)` to signal failure; return `Ok(port-map)` on success.
+
+## Signature embedding
+
+The codec's signature must be embedded in the `.wasm` binary as a `chonkle:signature` custom section at build time. The `chonkle` CLI provides a helper command to do this:
+
+```bash
+chonkle embed-signature codec.wasm signature.json
+```
diff --git a/docs/reference/codec-contract/CORE.md b/docs/reference/codec-contract/CORE.md
new file mode 100644
index 0000000..2c2c57a
--- /dev/null
+++ b/docs/reference/codec-contract/CORE.md
@@ -0,0 +1,75 @@
+# Core Wasm
+
+Core Wasm codecs are wasm32-wasi reactor modules. For a build guide, see [Building a Core Wasm Codec](../../guides/CORE.md). The host and module communicate through direct reads and writes to the module's linear memory, using a binary serialization format for port-maps.
+
+## Required Exports
+
+| Export | Type | Description |
+|---|---|---|
+| `memory` | Memory | Linear memory (provided automatically by the wasm toolchain) |
+| `alloc` | `func(size: i32) -> i32` | Allocate `size` bytes; return pointer or 0 on failure |
+| `dealloc` | `func(ptr: i32, size: i32)` | Free a buffer previously returned by `alloc` or by `encode`/`decode` |
+| `encode` | `func(port_map_ptr: i32, port_map_len: i32) -> i64` | Encode operation |
+| `decode` | `func(port_map_ptr: i32, port_map_len: i32) -> i64` | Decode operation |
+
+## Port-Map Binary Format
+
+All multi-byte integers are little-endian.
+
+```
+u32: entry_count
+For each entry:
+  u32: name_len
+  u8[name_len]: name (UTF-8, not null-terminated)
+  u32: data_len
+  u8[data_len]: data
+```
+
+No padding or alignment between fields. Entries are serialized in order.
+
+### Example
+
+A port-map with one entry `("bytes", [0xDE, 0xAD])`:
+
+```
+01 00 00 00    entry_count = 1
+05 00 00 00    name_len = 5
+62 79 74 65 73 name = "bytes"
+02 00 00 00    data_len = 2
+DE AD          data
+```
+
+Total: 19 bytes.
+
+## Calling Convention
+
+The host allocates the input buffer in the module's linear memory via `alloc()`, then writes the serialized port-map. The codec takes ownership of this buffer and is responsible for freeing it.
+
+`encode` and `decode` return a packed `i64`: `(output_ptr << 32) | output_len`. The host reads the output port-map from linear memory, then calls `dealloc(output_ptr, output_len)` to free it. Return `0` to signal an error.
+
+```
+Host                              Module
+  |                                  |
+  |-- alloc(input_size) ------------>|
+  |<------------ input_ptr ----------|
+  |                                  |
+  |-- Memory.write(input_ptr, data)->|
+  |                                  |
+  |-- encode(input_ptr, input_size)->|
+  |           [module frees input]   |
+  |           [module allocates out] |
+  |<------ (output_ptr, output_len) -|
+  |                                  |
+  |-- Memory.read(output_ptr, ...) ->|
+  |                                  |
+  |-- dealloc(output_ptr, out_len) ->|
+  |                                  |
+```
+
+## Signature Embedding
+
+The codec's signature must be embedded in the `.wasm` binary as a `chonkle:signature` custom section at build time. The `chonkle` CLI provides a helper command to do this:
+
+```bash
+chonkle embed-signature codec.wasm signature.json
+```
diff --git a/docs/reference/codec-contract/NATIVE.md b/docs/reference/codec-contract/NATIVE.md
new file mode 100644
index 0000000..aadfde9
--- /dev/null
+++ b/docs/reference/codec-contract/NATIVE.md
@@ -0,0 +1,90 @@
+# Native (numcodecs)
+
+Native codecs wrap [numcodecs](https://numcodecs.readthedocs.io/) codec objects. For a step-by-step guide, see [Adding a Native Codec](../../guides/NATIVE.md) — there is no `.wasm` binary. Using native codecs requires `numcodecs` and `numpy`: `pip install chonkle[native]`.
+
+## Signature source
+
+Signatures are bundled as JSON files in `src/chonkle/signatures/numcodecs/`, one file per supported codec (e.g. `zlib.json`, `delta.json`). Adding support for a new numcodecs codec requires adding a JSON file to this directory.
+
+Native signatures use the same port-map interface as wasm codec signatures (`codec_id`, `implementation`, `inputs`, `outputs`), with an additional `native` block that controls the calling convention.
+
+## The `native` block
+
+The `native` block is a required top-level field in native signature JSON files. It has three fields:
+
+- **`constructor_ports`** — list of input port names whose JSON-decoded values become keyword arguments to `numcodecs.get_codec()`. Ports not in this list are not passed to the constructor. Ports listed here but absent from the port-map at call time (e.g. encode-only ports during decode) are silently skipped.
+- **`encode`** — calling recipe for the encode direction.
+- **`decode`** — calling recipe for the decode direction.
+
+Each recipe has:
+
+- **`mode`** — `"bytes"` or `"ndarray"`.
+- **`dtype_port`** — (required when mode is `"ndarray"`) name of the input port supplying the numpy dtype string for `np.frombuffer`.
+
+### Bytes mode
+
+The `bytes` port value is passed directly to `codec.encode()` / `codec.decode()`. Used by compression codecs (zlib, gzip, bz2, lzma, zstd, lz4, blosc) and checksum codecs (adler32, crc32, crc32c, fletcher32, jenkins_lookup3).
+
+```json
+"native": {
+    "constructor_ports": ["level"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+}
+```
+
+### Ndarray mode (symmetric dtype)
+
+The `bytes` port is converted to a numpy ndarray via `np.frombuffer(data, dtype=...)` before being passed to the codec. The dtype comes from the port named by `dtype_port`. Both directions use the same dtype port.
+
+```json
+"native": {
+    "constructor_ports": ["dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+}
+```
+
+### Ndarray mode (asymmetric dtype)
+
+Some codecs use different dtypes for encode and decode. Each direction references a different port for the `np.frombuffer` dtype. The `dtype_port` should reference the port whose value matches the *input* dtype for that direction.
+
+```json
+"native": {
+    "constructor_ports": ["encode_dtype", "decode_dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "decode_dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "encode_dtype"}
+}
+```
+
+In this example (astype), `encode_dtype` is the output type of encode and `decode_dtype` is the output type of decode (matching the numcodecs constructor args). The `dtype_port` references are swapped because `np.frombuffer` needs the *input* type: during encode, input data has the decode dtype; during decode, input data has the encode dtype.
+
+### Mixed mode (ndarray encode, bytes decode)
+
+Some codecs consume an ndarray for encode but raw bytes for decode. Each direction uses a different mode. The `dtype` port is `encode_only` because only the encode direction needs it for `np.frombuffer`.
+
+```json
+"native": {
+    "constructor_ports": [],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "bytes"}
+}
+```
+
+In this example (json2), encode converts the byte buffer to an ndarray and serializes it to JSON bytes. Decode receives JSON bytes directly and returns the deserialized ndarray (numcodecs JSON embeds dtype and shape in the JSON output, so the round-trip is byte-identical).
+
+## Parameter handling
+
+Port-map entries other than `bytes` are JSON-decoded. Ports listed in `constructor_ports` become keyword arguments to `numcodecs.get_codec()`. For example, a zlib codec receiving `("level", b"6")` in its port-map and `constructor_ports: ["level"]` is instantiated as `numcodecs.get_codec({"id": "zlib", "level": 6})`.
+
+## Direction-aware ports
+
+Input ports can be marked `encode_only: true` or `decode_only: true` in the signature. The executor omits `encode_only` ports from the port-map during decode and `decode_only` ports during encode. This is useful for codecs like packbits where `encode_dtype` is only needed during encode and `decode_dtype` only during decode.
+
+## Unsupported numcodecs codecs
+
+Three numcodecs codecs cannot be supported by NativeCodec: `vlen-array`, `vlen-bytes`, and `vlen-utf8`.
+
+All three encode/decode arrays of variable-length elements (sub-arrays, byte strings, or UTF-8 strings). `encode` expects a 1-D numpy **object array** where each element is a variable-length Python object. `decode` expects raw bytes and returns a 1-D object array. `np.frombuffer` can only produce flat numeric arrays with fixed-element-size dtypes — there is no dtype that reconstructs variable-length Python objects from a flat byte buffer. The encode path has no way to receive the correct input type.
+
+Supporting these codecs would require a custom serialization format for object arrays in the bytes port, plus NativeCodec logic to deserialize bytes into an object array before calling encode.
diff --git a/docs/reference/codec-contract/README.md b/docs/reference/codec-contract/README.md
new file mode 100644
index 0000000..753f417
--- /dev/null
+++ b/docs/reference/codec-contract/README.md
@@ -0,0 +1,51 @@
+# Codec Contract
+
+These documents specify the interface contracts between chonkle (the host) and codec implementations. Three codec backends are supported:
+
+- [Core Wasm](CORE.md)
+- [Component Model Wasm](COMPONENT_MODEL.md)
+- [Native (numcodecs)](NATIVE.md)
+
+For step-by-step build guides, see [docs/guides/](../../guides/).
+
+All three are wrapped by the `Codec` ABC and present a uniform interface to the executor: `call(direction, port_map)` and `signature()`.
+
+## Call interface
+
+Every codec exposes an `encode` and a `decode` operation. The `direction` argument to `call()` determines which is invoked. Both operations receive named inputs and produce named outputs, corresponding to the `inputs` and `outputs` declared in the codec's signature. Concretely, the inputs and outputs are passed as a **port-map**: a list of `(name, bytes)` pairs.
+
+## Signature
+
+Every codec must have a signature, returned by `signature()`: a JSON object declaring `codec_id`, `implementation`, `inputs`, and `outputs`. For wasm codecs, the signature is embedded in the `.wasm` binary as a `chonkle:signature` custom section. For native codecs, the signature is a bundled JSON file in `src/chonkle/signatures/numcodecs/`.
+
+### Format
+
+```json
+{
+  "codec_id": "zstd",
+  "implementation": "zstd-rs",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 3, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+
+Fields:
+
+- `codec_id` — logical codec identifier (e.g. `"zlib"`, `"tiff-predictor-2"`)
+- `implementation` — identifies the specific build or project that produced the codec (e.g. `"zlib-rs"`, `"numcodecs.zlib"`)
+- `inputs` — map of input port descriptors. Each has `type` (required), and optional `required` (default `true`), `default`, `encode_only` (default `false`), and `decode_only` (default `false`).
+- `outputs` — map of output port descriptors. Each has `type`.
+
+Native codec signatures also include a `native` block that controls the calling convention (constructor args, bytes vs. ndarray mode, per-direction dtype). See [NATIVE.md](NATIVE.md).
+
+### Validation
+
+The host validates signatures during pipeline preparation:
+
+- Both inputs and outputs use **subset checks** — a step need not wire every port the codec declares. For example, a codec may declare an optional `level` input with a default value; a pipeline step that is happy with the default simply omits it from its `inputs`.
+- Input validation is **direction-aware**: ports marked `encode_only: true` are excluded during decode; ports marked `decode_only: true` are excluded during encode. Both are omitted from validation and from the port-map passed to the codec in the excluded direction.
diff --git a/docs/reference/protospec/PROTOSPEC.md b/docs/reference/protospec/PROTOSPEC.md
new file mode 100644
index 0000000..33e9a68
--- /dev/null
+++ b/docs/reference/protospec/PROTOSPEC.md
@@ -0,0 +1,2351 @@
+# Codec Invetory
+
+Codec inventory for chunked array and columnar data formats. Each codec is a
+named, versioned transformation with typed inputs and outputs. Codecs compose
+into pipelines; a pipeline is itself a codec.
+
+Entries flagged ⚠️ have lower confidence and should be verified against primary
+sources before finalizing.
+
+Note this document in its current form has been compiled entirely by an LLM,
+and is still in progress.
+
+---
+
+## Codec Signature
+
+A codec signature defines the transformation interface: typed inputs and typed
+outputs. Every codec in this catalog has a signature. A pipeline is also a codec
+and has the same signature structure.
+
+### Types
+
+| Type         | Description                                         |
+|--------------|-----------------------------------------------------|
+| `bytes`      | A byte stream                                       |
+| `uint`       | Unsigned integer scalar                              |
+| `int`        | Signed integer scalar                                |
+| `float`      | Floating point scalar                                |
+| `bool`       | Boolean scalar                                       |
+| `string`     | String value (e.g. dtype name, encoding mode)        |
+| `uint[]`     | Array of unsigned integers (e.g. shape, permutation) |
+| `dtype_desc` | Dtype descriptor (base type, width, signedness)      |
+
+### Input properties
+
+| Property      | Description                                                     |
+|---------------|-----------------------------------------------------------------|
+| `type`        | One of the types above                                          |
+| `required`    | Whether the input must be provided (default: `true`)            |
+| `default`     | Default value if not provided                                   |
+| `encode_only` | If `true`, only needed during encoding, not decoding            |
+
+### Codec identification
+
+Each codec is identified by a URI that includes a version. The URI scheme is
+TBD; identifiers used in this catalog are short slugs (e.g. `zstd`, `shuffle`)
+that do not yet conform to the final URI scheme.
+
+### Example signature
+
+```json
+{
+  "codec_id": "zstd",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 3, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+
+`zstd` takes one byte stream and an optional compression level (encode-only),
+and produces one byte stream. On decode, only `bytes` is required.
+
+---
+
+## Pipeline Model
+
+A pipeline is a codec. It has typed inputs and outputs, and internally is a
+directed acyclic graph (DAG) of steps. Each step invokes a codec and wires its
+inputs/outputs to other steps or to the pipeline boundary.
+
+### Wiring rules
+
+- Step inputs reference either `input.<n>` (pipeline input),
+  `constant.<n>` (pipeline constant), or `<step_name>.<output_name>`
+  (output of a previous step).
+- Pipeline outputs reference `<step_name>.<output_name>`.
+- Steps execute in dependency order (topological sort of the DAG).
+- A value can be referenced by multiple steps (fan-out / reuse).
+
+### Direction and inversion
+
+A pipeline declares a `direction`: `"decode"` or `"encode"`. This tells the
+executor which way the DAG was written. The examples in this catalog use the
+decode direction. A codec executor must be able to invert a pipeline regardless
+of which direction is declared.
+
+Individual codecs do not have a direction — every codec is bidirectional by
+definition. A codec encodes and decodes. The direction flag exists only on the
+pipeline to specify how to read the DAG.
+
+To run a pipeline in the opposite direction from the one declared:
+
+1. Reverse the DAG edges (each step's outputs become its inputs and vice versa).
+2. Pipeline inputs become pipeline outputs and vice versa.
+3. Constants remain constants.
+4. `encode_only` inputs are active when encoding, ignored when decoding.
+
+The executor calls each codec's encode or decode operation depending on which
+direction is needed. No per-step inversion logic is required beyond this — each
+codec already knows how to operate in both directions.
+
+### Constants
+
+Constants are values fixed for a given pipeline definition. They differ from
+inputs in that they are not supplied by the caller — they are baked into the
+pipeline spec. Constants have a type and a value.
+
+### Self-describing vs externally-parameterized codecs
+
+A codec whose decode parameters are embedded in the chunk bytes (self-describing)
+is atomic — it cannot be decomposed into a pipeline, because the format driver
+would need to parse the codec's internal header to supply parameters to sub-steps.
+The chunk goes into the codec as-is.
+
+A codec whose decode parameters come from external metadata (the format driver
+supplies them) can be a step in a pipeline, because the format driver provides
+the parameters as pipeline inputs.
+
+Example: Blosc embeds its compressor name, shuffle mode, and element size in a
+header within the chunk bytes → atomic codec. Zstd does not embed any metadata
+in its frame that the format driver needs to extract → pipeline step.
+
+### Pipeline examples
+
+Composite codecs that can be decomposed into pipelines of generic codecs
+have their pipeline definition colocated with their catalog entry. See:
+
+- [`orc-varint`](#orc-varint) — linear pipeline: `varint → zigzag`
+- [`parquet-page-v2-split`](#parquet-page-v2-split) — DAG with fan-out, multiple inputs, constants
+
+---
+
+## Catalog Fields
+
+Each codec entry includes:
+
+- **canonical-id**: Primary key slug (will become a versioned URI; see above)
+- **aliases**: `{format, identifier}` pairs
+- **awareness**: `pure-bytes` | `stride-aware` | `type-aware` | `structure-aware`
+- **lossiness**: `lossless` | `lossy` | `conditional`
+- **signature**: JSON codec signature (typed inputs and outputs)
+- **pipeline**: (where applicable) pipeline decomposition into generic codecs
+- **conceptual decomposition**: (where applicable) decomposition not directly expressible as a pipeline
+- **spec**: Link(s) to canonical specification
+- **implementations**: Known implementations
+- **licensing**: Any restrictions
+- **notes**: Caveats, open questions
+
+---
+
+## Codec Inventory
+
+### Awareness Categories
+
+Awareness describes what a codec needs to know about the data it operates on,
+beyond the raw bytes.
+
+#### [Pure-Bytes](#pure-bytes-codecs)
+
+Codec treats input as opaque bytes. No knowledge of element boundaries, data
+types, or dimensional structure.
+
+- [`zstd`](#zstd) · [`lz4`](#lz4) · [`deflate`](#deflate) · [`gzip`](#gzip) · [`bzip2`](#bzip2) · [`lzma`](#lzma) · [`snappy`](#snappy) · [`brotli`](#brotli) · [`lzo`](#lzo) · [`lzw`](#lzw) · [`lzw-tiff`](#lzw-tiff) · [`packbits`](#packbits) · [`szip`](#szip) · [`blosc`](#blosc) · [`jpeg-xl`](#jpeg-xl) · [`jpeg`](#jpeg) · [`jpeg2000`](#jpeg2000) · [`webp`](#webp) · [`jpeg-xr`](#jpeg-xr) · [`no-compression`](#no-compression) · [`fletcher32`](#fletcher32) · [`crc32c`](#crc32c) · [`crc32`](#crc32) · [`orc-boolean-rle`](#orc-boolean-rle)
+
+#### [Stride-Aware](#stride-aware-codecs)
+
+Codec needs to know element width (byte stride or bit width) to operate on
+element boundaries, but does not need to know the semantic type.
+
+- [`shuffle`](#shuffle) · [`bitshuffle`](#bitshuffle) · [`byte-stream-split`](#byte-stream-split) · [`rle-parquet`](#rle-parquet) · [`rle`](#rle)
+
+#### [Type-Aware](#type-aware-codecs)
+
+Codec needs to know the semantic data type (signed vs unsigned, float vs int,
+physical type) to correctly encode or decode values.
+
+- [`bitpack`](#bitpack) · [`fsst`](#fsst) · [`nbit`](#nbit) · [`scale-offset`](#scale-offset) · [`lerc`](#lerc) · [`delta-parquet`](#delta-parquet) · [`delta-length-byte-array`](#delta-length-byte-array) · [`delta-byte-array`](#delta-byte-array) · [`delta`](#delta) · [`fixedscaleoffset`](#fixedscaleoffset) · [`quantize`](#quantize) · [`bitround`](#bitround) · [`astype`](#astype) · [`bytes`](#bytes) · [`plain-parquet`](#plain-parquet) · [`plain-dictionary`](#plain-dictionary) · [`orc-rle-v1`](#orc-rle-v1) · [`orc-rle-v2`](#orc-rle-v2) · [`orc-varint`](#orc-varint) · [`orc-string-direct`](#orc-string-direct) · [`orc-string-dictionary`](#orc-string-dictionary) · [`vlen-utf8`](#vlen-utf8) · [`vlen-bytes`](#vlen-bytes) · [`categorize`](#categorize) · [`zigzag`](#zigzag) · [`varint`](#varint) · [`zfp`](#zfp) · [`sz2`](#sz2) · [`sz3`](#sz3)
+
+#### [Structure-Aware](#structure-aware-codecs)
+
+Codec needs format-specific structural knowledge — dimensional layout, page
+structure, image geometry, or format-specific metadata beyond simple type info.
+
+- [`transpose`](#transpose) · [`tiff-predictor-2`](#tiff-predictor-2) · [`tiff-predictor-3`](#tiff-predictor-3) · [`png-filter`](#png-filter) · [`parquet-page-v1-split`](#parquet-page-v1-split) · [`parquet-page-v2-split`](#parquet-page-v2-split) · [`arrow-ipc-buffer`](#arrow-ipc-buffer) · [`tiff-old-jpeg`](#tiff-old-jpeg) · [`ccitt-group3`](#ccitt-group3) · [`ccitt-group4`](#ccitt-group4) · [`grib-grid-packing`](#grib-grid-packing) · [`grib-ccsds`](#grib-ccsds) · [`grib-jpeg2000`](#grib-jpeg2000)
+
+---
+
+## Pure-Bytes Codecs
+
+---
+
+### zstd
+
+- **canonical-id**: `zstd`
+- **aliases**: `{zarr-v2, "numcodecs.Zstd"}`, `{zarr-v3, "numcodecs.zstd"}`, `{hdf5, 32015}`, `{parquet, "ZSTD"}`, `{orc, "ZSTD"}`, `{arrow-ipc, "ZSTD"}`, `{tiff, 50000}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "zstd",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 3, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://facebook.github.io/zstd/zstd_manual.html
+- **implementations**: [zstd (C reference)](https://github.com/facebook/zstd), [python-zstd](https://github.com/sergey-dryabzhinsky/python-zstd), [zstd-rs](https://github.com/gyscos/zstd-rs)
+- **licensing**: BSD 2-Clause / GPLv2 dual license
+- **notes**: Compression level is encode-only; decode does not require it. TIFF tag 50000 is self-assigned by the libtiff project; Adobe's official registration process is defunct, making libtiff the de facto registry for newer TIFF compression IDs.
+
+---
+
+### lz4
+
+- **canonical-id**: `lz4`
+- **aliases**: `{zarr-v2, "numcodecs.LZ4"}`, `{hdf5, 32004}`, `{parquet, "LZ4_RAW"}`, `{orc, "LZ4"}`, `{blosc2, "lz4"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "lz4",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "acceleration": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://github.com/lz4/lz4/blob/dev/doc/lz4_Frame_format.md
+- **implementations**: [lz4 (C reference)](https://github.com/lz4/lz4), [lz4-rs](https://github.com/10xGenomics/lz4)
+- **licensing**: BSD 2-Clause
+- **notes**: Parquet previously used a framed LZ4 variant (`LZ4`) that was inconsistently implemented; `LZ4_RAW` is the unframed standard and preferred in newer Parquet writers. ⚠️ Verify HDF5 filter 32004 uses framed or raw variant.
+
+---
+
+### deflate
+
+- **canonical-id**: `deflate`
+- **aliases**: `{zarr-v2, "numcodecs.Zlib"}`, `{zarr-v3, "numcodecs.zlib"}`, `{hdf5, 1}`, `{tiff, 8}`, `{tiff, 32946}`, `{parquet, "GZIP"}`, `{orc, "ZLIB"}`, `{netcdf-classic, "deflate"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "deflate",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://www.rfc-editor.org/rfc/rfc1951 (DEFLATE), https://www.rfc-editor.org/rfc/rfc1950 (zlib framing)
+- **implementations**: [zlib](https://zlib.net), [miniz](https://github.com/richgel999/miniz), [zlib-ng](https://github.com/zlib-ng/zlib-ng)
+- **licensing**: zlib license
+- **notes**: "deflate" and "zlib" are often conflated. DEFLATE is the algorithm; zlib adds a 2-byte header and Adler-32 checksum. TIFF compression=8 uses zlib framing. NetCDF classic uses zlib framing. Parquet labels this `GZIP` but uses zlib framing, not gzip framing. ⚠️ Verify exact framing used per format alias. NetCDF classic has no codec layer beyond deflate applied per-variable. NetCDF-4 inherits the full HDF5 filter pipeline.
+
+---
+
+### gzip
+
+- **canonical-id**: `gzip`
+- **aliases**: `{zarr-v2, "numcodecs.GZip"}`, `{arrow-ipc, "GZIP"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "gzip",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://www.rfc-editor.org/rfc/rfc1952
+- **implementations**: [gzip](https://www.gnu.org/software/gzip/), [zlib](https://zlib.net)
+- **licensing**: GPL (gzip tool); zlib license (zlib library implementation)
+- **notes**: Gzip framing = DEFLATE + gzip header/trailer with CRC-32. Distinct from zlib framing. Not the same as `deflate` despite shared underlying algorithm.
+
+---
+
+### bzip2
+
+- **canonical-id**: `bzip2`
+- **aliases**: `{zarr-v2, "numcodecs.BZ2"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "bzip2",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://sourceware.org/bzip2/
+- **implementations**: [bzip2](https://sourceware.org/bzip2/), [bzip2-rs](https://github.com/alexcrichton/bzip2-rs)
+- **licensing**: BSD-like (bzip2 license)
+
+---
+
+### lzma
+
+- **canonical-id**: `lzma`
+- **aliases**: `{zarr-v2, "numcodecs.LZMA"}`, `{tiff, 34925}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "lzma",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "format": {"type": "int", "required": false, "default": 1, "encode_only": true},
+    "check": {"type": "int", "required": false, "default": -1, "encode_only": true},
+    "preset": {"type": "int", "required": false, "encode_only": true},
+    "filters": {"type": "string", "required": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://tukaani.org/xz/xz-file-format.txt
+- **implementations**: [xz-utils](https://tukaani.org/xz/), [lzma-rs](https://github.com/gendx/lzma-rs)
+- **licensing**: Public domain / 0BSD
+- **notes**: TIFF (tag 34925) uses LZMA2 via liblzma. The framing difference (XZ container vs raw LZMA2 stream) is handled transparently by liblzma depending on invocation context.
+
+---
+
+### snappy
+
+- **canonical-id**: `snappy`
+- **aliases**: `{parquet, "SNAPPY"}`, `{orc, "SNAPPY"}`, `{arrow-ipc, "SNAPPY"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "snappy",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://github.com/google/snappy/blob/main/format_description.txt
+- **implementations**: [snappy (C++ reference)](https://github.com/google/snappy), [snap (Rust)](https://github.com/BurntSushi/rust-snappy)
+- **licensing**: BSD 3-Clause
+- **notes**: No HDF5 registered filter. Prioritizes speed over compression ratio. No configuration parameters.
+
+---
+
+### brotli
+
+- **canonical-id**: `brotli`
+- **aliases**: `{parquet, "BROTLI"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "brotli",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "quality": {"type": "int", "required": false, "default": 11, "encode_only": true},
+    "lgwin": {"type": "int", "required": false, "default": 22, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://www.rfc-editor.org/rfc/rfc7932
+- **implementations**: [brotli (C reference)](https://github.com/google/brotli)
+- **licensing**: MIT
+
+---
+
+### lzo ⚠️
+
+- **canonical-id**: `lzo`
+- **aliases**: `{orc, "LZO"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "lzo",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://www.oberhumer.com/opensource/lzo/ ⚠️
+- **implementations**: [lzo](https://www.oberhumer.com/opensource/lzo/)
+- **licensing**: GPL v2 ⚠️ — licensing may be a concern for some uses
+- **notes**: ⚠️ Verify exact LZO variant and framing used in ORC. GPL license may be a concern.
+
+---
+
+### lzw
+
+- **canonical-id**: `lzw`
+- **aliases**: none in target formats
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "lzw",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: Welch, T. (1984). "A Technique for High-Performance Data Compression". IEEE Computer.
+- **implementations**: [compress (Unix)](https://en.wikipedia.org/wiki/Compress), various
+- **licensing**: Unisys patent expired; now unencumbered
+- **notes**: The canonical LZW algorithm. Included to document the distinction from `lzw-tiff`. TIFF uses an incompatible variant — a generic LZW implementation will not correctly decode TIFF LZW data. See `lzw-tiff`.
+
+---
+
+### lzw-tiff
+
+- **canonical-id**: `lzw-tiff`
+- **aliases**: `{tiff, 5}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "lzw-tiff",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: TIFF 6.0 specification section on LZW compression
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [golang/image tiff/lzw](https://github.com/golang/image/tree/master/tiff/lzw)
+- **licensing**: Unencumbered
+- **notes**: TIFF's LZW is incompatible with standard LZW due to an "off by one" in code length transitions. A generic LZW codec cannot decode TIFF LZW data. Additionally, libtiff includes a backwards-compatibility decode path (`LZW_COMPAT`) for older files with reversed bit order. This is a concrete example of why format libraries cannot simply be replaced by a generic codec — the TIFF format driver must invoke `lzw-tiff` specifically.
+
+---
+
+### packbits
+
+- **canonical-id**: `packbits`
+- **aliases**: `{tiff, 32773}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "packbits",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: TIFF 6.0 specification section 9
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+
+---
+
+### szip ⚠️
+
+- **canonical-id**: `szip`
+- **aliases**: `{hdf5, 4}`, `{netcdf-4, "szip"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless ⚠️ verify for float data
+- **signature**:
+```json
+{
+  "codec_id": "szip",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "coding": {"type": "int", "required": true},
+    "pixels_per_block": {"type": "int", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://support.hdfgroup.org/doc_resource/SZIP/ ⚠️
+- **implementations**: [libaec](https://gitlab.dkrz.de/k202009/libaec) (open-source reimplementation), [szip (original)](https://support.hdfgroup.org/doc_resource/SZIP/)
+- **licensing**: ⚠️ Original SZIP has proprietary licensing restrictions. `libaec` is BSD licensed.
+- **notes**: Significant licensing history. libaec is a compatible open-source reimplementation. HDF5 bundles libaec in recent versions. Verify losslessness for all dtypes.
+
+---
+
+### blosc
+
+- **canonical-id**: `blosc`
+- **aliases**: `{zarr-v2, "numcodecs.Blosc"}`, `{hdf5, 32001}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "blosc",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "cname": {"type": "string", "required": false, "default": "lz4", "encode_only": true},
+    "clevel": {"type": "int", "required": false, "default": 5, "encode_only": true},
+    "shuffle": {"type": "int", "required": false, "default": 1, "encode_only": true},
+    "blocksize": {"type": "int", "required": false, "default": 0, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://github.com/Blosc/c-blosc/blob/master/README.rst
+- **implementations**: [c-blosc](https://github.com/Blosc/c-blosc), [python-blosc](https://github.com/Blosc/python-blosc)
+- **licensing**: BSD 3-Clause
+- **notes**: Self-describing: the Blosc header embedded in the chunk carries the compressor name, shuffle mode, element size, and block size. All encode parameters are encode-only because decode reads them from the header. This makes Blosc atomic — it cannot be decomposed into a pipeline without the format driver parsing the Blosc header, which violates the principle that the chunk enters the codec pipeline as-is. In Zarr v3, the equivalent operation is decomposed into explicit shuffle + compressor pipeline steps with parameters supplied externally by the format driver.
+
+---
+
+### jpeg-xl
+
+- **canonical-id**: `jpeg-xl`
+- **aliases**: `{tiff, 50002}` ⚠️ verify tag number, `{gdal-gtiff, "JXL"}`
+- **awareness**: pure-bytes
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "jpeg-xl",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "distance": {"type": "float", "required": false, "default": 1.0, "encode_only": true},
+    "effort": {"type": "int", "required": false, "default": 7, "encode_only": true},
+    "lossless": {"type": "bool", "required": false, "default": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://jpeg.org/jpegxl/ ; ISO/IEC 18181
+- **implementations**: [libjxl (C++ reference)](https://github.com/libjxl/libjxl), [jxl-rs](https://github.com/libjxl/jxl-rs), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: BSD 3-Clause (libjxl); royalty-free format
+- **notes**: Supports both lossy and lossless compression. Notably supports lossless JPEG recompression. Internal stages (VarDCT, modular prediction, ANS entropy coding) are not decomposed — the intermediate representations are JXL-internal and not usefully recombineable with other codec steps. TIFF tag number needs verification.
+
+---
+
+### jpeg
+
+- **canonical-id**: `jpeg`
+- **aliases**: `{tiff, 7}`, `{numcodecs, "imagecodecs.jpeg8"}` ⚠️ verify
+- **awareness**: pure-bytes
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "jpeg",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "quality": {"type": "int", "required": false, "default": 75, "encode_only": true},
+    "lossless": {"type": "bool", "required": false, "default": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: ITU-T T.81 / ISO/IEC 10918-1
+- **implementations**: [libjpeg-turbo](https://github.com/libjpeg-turbo/libjpeg-turbo), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered (format); libjpeg-turbo BSD/zlib
+- **notes**: Internally applies DCT → quantization → Huffman coding. Internal stages are not decomposed. TIFF compression=7 is new-style JPEG with self-contained strips; see `tiff-old-jpeg` (compression=6) for legacy variant.
+
+---
+
+### jpeg2000
+
+- **canonical-id**: `jpeg2000`
+- **aliases**: `{tiff, 34712}`, `{numcodecs, "imagecodecs.jpeg2k"}` ⚠️ verify
+- **awareness**: pure-bytes
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "jpeg2000",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "lossless": {"type": "bool", "required": false, "default": false, "encode_only": true},
+    "level": {"type": "int", "required": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: ISO/IEC 15444-1
+- **implementations**: [OpenJPEG](https://github.com/uclouvain/openjpeg), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: BSD 2-Clause (OpenJPEG); royalty-free format
+- **notes**: Internally applies DWT → quantization → EBCOT. Lossless mode uses reversible 5/3 wavelet; lossy uses 9/7 wavelet. Stages not decomposed. Used in GRIB2 (see `grib-jpeg2000`).
+
+---
+
+### webp
+
+- **canonical-id**: `webp`
+- **aliases**: `{tiff, 50001}`
+- **awareness**: pure-bytes
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "webp",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "quality": {"type": "int", "required": false, "default": 75, "encode_only": true},
+    "lossless": {"type": "bool", "required": false, "default": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://developers.google.com/speed/webp/docs/riff_container
+- **implementations**: [libwebp](https://chromium.googlesource.com/webm/libwebp), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: BSD 3-Clause
+- **notes**: Lossy mode based on VP8 intra-frame coding. Lossless mode uses spatial prediction + LZ77+Huffman. Internal stages not decomposed. TIFF tag 50001 is self-assigned by libtiff.
+
+---
+
+### jpeg-xr ⚠️
+
+- **canonical-id**: `jpeg-xr`
+- **aliases**: `{tiff, 34892}` ⚠️ verify
+- **awareness**: pure-bytes
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "jpeg-xr",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "quality": {"type": "float", "required": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: ITU-T T.832 / ISO/IEC 29199-2
+- **implementations**: [jxrlib](https://github.com/4creators/jxrlib), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: BSD (jxrlib) ⚠️ verify
+- **notes**: ⚠️ Low confidence. Microsoft-developed HDR-capable image codec. Internal stages not decomposed.
+
+---
+
+### no-compression
+
+- **canonical-id**: `no-compression`
+- **aliases**: `{tiff, 1}`, `{parquet, "UNCOMPRESSED"}`, `{orc, "NONE"}`, `{hdf5, "no filter"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "no-compression",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: N/A
+- **implementations**: N/A — identity transform
+- **notes**: Included for completeness. Zarr v3's `"bytes"` codec is a separate concept (serialization with explicit endianness) — not aliased here.
+
+---
+
+### fletcher32
+
+- **canonical-id**: `fletcher32`
+- **aliases**: `{hdf5, 3}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "fletcher32",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "error_on_fail": {"type": "bool", "required": false, "default": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://support.hdfgroup.org/documentation/hdf5/latest/group___d_a_p_l.html ⚠️ verify
+- **implementations**: [HDF5 (C)](https://github.com/HDFGroup/hdf5)
+- **licensing**: BSD-like (HDF5 license)
+- **notes**: Appends 4-byte Fletcher-32 checksum on encode; verifies and strips on decode. `error_on_fail` is a runtime parameter (not stored in data). Output is 4 bytes shorter on decode.
+
+---
+
+### crc32c
+
+- **canonical-id**: `crc32c`
+- **aliases**: `{zarr-v3, "crc32c"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "crc32c",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "error_on_fail": {"type": "bool", "required": false, "default": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://zarr-specs.readthedocs.io/en/latest/v3/codecs/crc32c/v1.0.html
+- **implementations**: [zarr-python](https://github.com/zarr-developers/zarr-python), [zarrs (Rust)](https://github.com/LDeakin/zarr_tools), [tensorstore](https://github.com/google/tensorstore)
+- **licensing**: Unencumbered (algorithm); implementations vary
+- **notes**: Appends 4-byte CRC-32C (Castagnoli) checksum on encode; verifies and strips on decode. CRC-32C uses a different polynomial (0x1EDC6F41) than standard CRC-32 (0x04C11DB7). Recommended by the Zarr v3 sharding spec for shard index integrity. No configuration parameters beyond the runtime `error_on_fail`.
+
+---
+
+### crc32
+
+- **canonical-id**: `crc32`
+- **aliases**: `{parquet, "page-crc32"}` ⚠️ verify identifier
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "crc32",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "error_on_fail": {"type": "bool", "required": false, "default": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://www.rfc-editor.org/rfc/rfc3720#appendix-B (CRC-32 / ISO 3309)
+- **implementations**: Widely available in standard libraries (zlib crc32, Python zlib.crc32)
+- **licensing**: Unencumbered
+- **notes**: ⚠️ Standard CRC-32 (polynomial 0x04C11DB7, ISO 3309). Parquet page headers include an optional CRC-32 checksum for page data integrity. Verify whether Parquet's CRC is appended to the page bytes (making it a codec in the pipeline sense) or stored in the page header metadata (making it a format-driver concern outside the codec layer). If the latter, this entry should be revised or removed.
+
+---
+
+### orc-boolean-rle
+
+- **canonical-id**: `orc-boolean-rle`
+- **aliases**: `{orc, "BOOLEAN_RLE"}`
+- **awareness**: pure-bytes
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "orc-boolean-rle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://orc.apache.org/specification/ORCv1/
+- **implementations**: [orc (Java)](https://github.com/apache/orc), [arrow (C++)](https://github.com/apache/arrow)
+- **licensing**: Apache 2.0
+- **notes**: Used exclusively for PRESENT (validity/null) streams. Packs 8 bits per byte.
+
+---
+
+
+## Stride-Aware Codecs
+
+---
+
+### shuffle
+
+- **canonical-id**: `shuffle`
+- **aliases**: `{hdf5, 2}`, `{numcodecs, "numcodecs.Shuffle"}`, `{blosc2, "shuffle"}`
+- **awareness**: stride-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "shuffle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "element_size": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://support.hdfgroup.org/documentation/hdf5/latest/_f_i_l_t_e_r.html
+- **implementations**: [HDF5 (C)](https://github.com/HDFGroup/hdf5), [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: Unencumbered (algorithm); HDF5 BSD-like
+
+---
+
+### bitshuffle
+
+- **canonical-id**: `bitshuffle`
+- **aliases**: `{hdf5, 32008}`, `{numcodecs, "numcodecs.BitShuffle"}`, `{blosc2, "bitshuffle"}`
+- **awareness**: stride-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "bitshuffle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "element_size": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://github.com/kiyo-masui/bitshuffle
+- **implementations**: [bitshuffle](https://github.com/kiyo-masui/bitshuffle), [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: The HDF5 filter variant (32008) optionally bundles LZ4 compression as an atomic operation. In a decomposed pipeline this should be modeled as `bitshuffle → lz4`.
+
+---
+
+### byte-stream-split
+
+- **canonical-id**: `byte-stream-split`
+- **aliases**: `{parquet, "BYTE_STREAM_SPLIT"}`
+- **awareness**: stride-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "byte-stream-split",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "element_size": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#byte-stream-split-byte_stream_split--11
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Identical in effect to HDF5 shuffle. Transposes byte matrix: all byte-0s contiguous, then all byte-1s, etc.
+
+---
+
+### rle-parquet
+
+- **canonical-id**: `rle-parquet`
+- **aliases**: `{parquet, "RLE"}`, `{parquet, "RLE_DICTIONARY"}` (index stream only)
+- **awareness**: stride-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "rle-parquet",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "bit_width": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#run-length-encoding--bit-packing-hybrid-rle--3
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Hybrid RLE and bit-packing. Self-selects per block between RLE runs and bit-packed runs via header byte. Not decomposable into a pipeline of `rle → bitpack` — the per-block adaptive selection is integral to the encoding.
+
+---
+
+### rle
+
+- **canonical-id**: `rle`
+- **aliases**: `{numcodecs, "numcodecs.RunLengthEncoding"}` ⚠️
+- **awareness**: stride-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "rle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "element_size": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/ ⚠️ verify
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs) ⚠️
+- **licensing**: MIT
+- **notes**: ⚠️ Verify numcodecs exposes a general RLE codec.
+
+---
+
+
+## Type-Aware Codecs
+
+---
+
+### bitpack
+
+- **canonical-id**: `bitpack`
+- **aliases**: `{parquet, "BIT_PACKED"}` (deprecated encoding 4), `{lance, "bitpacking"}`, `{tiff, "sub-byte-packing"}` ⚠️ verify TIFF modeling
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "bitpack",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "bit_width": {"type": "uint", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: No single canonical spec. Parquet encoding spec (encoding 4); Fastlanes paper (Afroozeh & Boncz, 2023).
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs), [lance](https://github.com/lancedb/lance) (Fastlanes-based), [bitpacking crate](https://github.com/tantivy-search/bitpacking)
+- **licensing**: Unencumbered (algorithm)
+- **notes**: Packs integer values into `bit_width` bits per value. Decode produces native-width integers. Parquet's `BIT_PACKED` (encoding 4) is deprecated in favor of the RLE/bit-packing hybrid. `nbit` (HDF5) performs a related operation — strips padding bits based on HDF5 dtype precision and bit offset — but requires full HDF5 dtype context; see `nbit` entry. TIFF sub-byte sample packing (e.g. 12-bit samples) is effectively `bitpack(bit_width=bits_per_sample)`. ⚠️ Verify TIFF modeling.
+
+---
+
+### fsst
+
+- **canonical-id**: `fsst`
+- **aliases**: `{lance, "fsst"}`, `{duckdb, "fsst"}` ⚠️ verify DuckDB identifier
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "fsst",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "symbol_table": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: Boncz, P., Neumann, T., & Raasveldt, M. (2020). "FSST: Fast Random Access String Compression." VLDB 2020. https://www.vldb.org/pvldb/vol13/p2649-boncz.pdf
+- **implementations**: [fsst (C++ reference)](https://github.com/cwida/fsst), [lance](https://github.com/lancedb/lance)
+- **licensing**: MIT ⚠️ verify
+- **notes**: Compresses strings by building a 256-entry symbol table from input data, then replacing matching substrings with 1-byte codes. Escape byte (0xFF) for literals. The `symbol_table` is a data-derived input: built during encoding, required for decoding, stored in codec metadata alongside the chunk. Compare with dictionary encoding (maps whole values to codes) — FSST maps substrings to codes and is effective on high-cardinality columns. ⚠️ Verify framing conventions for symbol table storage across implementations.
+
+---
+
+### nbit
+
+- **canonical-id**: `nbit`
+- **aliases**: `{hdf5, 5}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "nbit",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "dtype_desc", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://support.hdfgroup.org/documentation/hdf5/latest/
+- **implementations**: [HDF5 (C)](https://github.com/HDFGroup/hdf5)
+- **licensing**: BSD-like (HDF5 license)
+- **notes**: Strips padding bits based on declared precision and offset in the HDF5 dtype. Requires full HDF5 dtype descriptor. Conceptually related to `bitpack` — both reduce values to fewer bits — but `nbit` derives bit width from dtype metadata rather than taking an explicit `bit_width` parameter. An implementation could potentially delegate to `bitpack` internally.
+
+---
+
+### scale-offset
+
+- **canonical-id**: `scale-offset`
+- **aliases**: `{hdf5, 6}`
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "scale-offset",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "dtype_desc", "required": true},
+    "fill_value": {"type": "float", "required": true},
+    "scale_type": {"type": "int", "required": true},
+    "scale_factor": {"type": "int", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://support.hdfgroup.org/documentation/hdf5/latest/
+- **implementations**: [HDF5 (C)](https://github.com/HDFGroup/hdf5)
+- **licensing**: BSD-like (HDF5 license)
+- **notes**: Lossless for integer types. Lossy for floating point (quantization). Fill values require special handling.
+
+---
+
+### lerc
+
+- **canonical-id**: `lerc`
+- **aliases**: `{tiff, 34887}`, `{numcodecs, "numcodecs.LERC"}` ⚠️
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "lerc",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "shape": {"type": "uint[]", "required": true},
+    "nodata_value": {"type": "float", "required": false},
+    "max_z_error": {"type": "float", "required": false, "default": 0.0, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://github.com/Esri/lerc/blob/master/doc/LERC_WhitePaper.pdf
+- **implementations**: [lerc (C++, Esri)](https://github.com/Esri/lerc), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Apache 2.0
+- **notes**: Esri geospatial codec. Natively 2D-aware. Lossless when `max_z_error=0`. Handles NoData internally. GDAL's `LERC_DEFLATE` and `LERC_ZSTD` are pipelines: `lerc → deflate` or `lerc → zstd`.
+
+---
+
+### delta-parquet
+
+- **canonical-id**: `delta-parquet`
+- **aliases**: `{parquet, "DELTA_BINARY_PACKED"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "delta-parquet",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "block_size": {"type": "int", "required": false, "default": 128, "encode_only": true},
+    "miniblock_count": {"type": "int", "required": false, "default": 4, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **conceptual decomposition**: `delta(dtype) → fixedscaleoffset(scale=1, offset=per_block_min) → bitpack(bit_width=per_miniblock)`. Not directly expressible as a pipeline because parameters vary per block within the stream.
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#delta-encoding-delta_binary_packed--5
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+
+---
+
+### delta-length-byte-array
+
+- **canonical-id**: `delta-length-byte-array`
+- **aliases**: `{parquet, "DELTA_LENGTH_BYTE_ARRAY"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "delta-length-byte-array",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **conceptual decomposition**: `variable-width-split() → { delta-parquet(lengths), identity(values) }`. Fan-out, then per-stream codec.
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#delta-encoding-of-string-lengths-delta_length_byte_array--6
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Stores lengths via delta-binary-packed encoding, followed by concatenated raw bytes.
+
+---
+
+### delta-byte-array
+
+- **canonical-id**: `delta-byte-array`
+- **aliases**: `{parquet, "DELTA_BYTE_ARRAY"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "delta-byte-array",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **conceptual decomposition**: `front-coding-split() → { delta-parquet(prefix_lengths), identity(suffixes) }`. Fan-out into two streams with delta on prefix lengths.
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#incremental-encoding-for-byte-arrays-aka-front-coding-delta_byte_array--7
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Front coding: encodes shared prefixes between successive byte arrays.
+
+---
+
+### delta
+
+- **canonical-id**: `delta`
+- **aliases**: `{zarr-v3, "numcodecs.delta"}`, `{numcodecs, "numcodecs.Delta"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "delta",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "element_size": {"type": "uint", "required": true},
+    "shape": {"type": "uint[]", "required": false},
+    "order": {"type": "string", "required": false, "default": "C"}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/delta.html
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Differences between successive elements. Unlike tiff-predictor-2, operates on the full array (not row-by-row) unless shape/order constrain traversal. `dtype` affects overflow behavior.
+
+---
+
+### fixedscaleoffset
+
+- **canonical-id**: `fixedscaleoffset`
+- **aliases**: `{numcodecs, "numcodecs.FixedScaleOffset"}`, `{zarr-v3, "numcodecs.fixedscaleoffset"}`
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "fixedscaleoffset",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "offset": {"type": "float", "required": true},
+    "scale": {"type": "float", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "astype": {"type": "string", "required": false}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/fixedscaleoffset.html
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Lossless if `astype` can represent all scaled values. Lossy otherwise. With `scale=1` this is frame-of-reference encoding (subtract offset only).
+
+---
+
+### quantize
+
+- **canonical-id**: `quantize`
+- **aliases**: `{numcodecs, "numcodecs.Quantize"}`
+- **awareness**: type-aware
+- **lossiness**: lossy
+- **signature**:
+```json
+{
+  "codec_id": "quantize",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "digits": {"type": "int", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/quantize.html
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+
+---
+
+### bitround
+
+- **canonical-id**: `bitround`
+- **aliases**: `{numcodecs, "numcodecs.BitRound"}`
+- **awareness**: type-aware
+- **lossiness**: lossy
+- **signature**:
+```json
+{
+  "codec_id": "bitround",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "keepbits": {"type": "int", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/bitround.html
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Zeroes low-order mantissa bits. Often followed by zstd or lz4.
+
+---
+
+### astype
+
+- **canonical-id**: `astype`
+- **aliases**: `{numcodecs, "numcodecs.AsType"}`
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "astype",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "encode_dtype": {"type": "string", "required": true},
+    "decode_dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/astype.html
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Lossless if encode_dtype can represent all values exactly. Lossy otherwise.
+
+---
+
+### bytes
+
+- **canonical-id**: `bytes`
+- **aliases**: `{zarr-v3, "bytes"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "bytes",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "shape": {"type": "uint[]", "required": true},
+    "endian": {"type": "string", "required": false, "default": "little"},
+    "order": {"type": "string", "required": false, "default": "C"}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://zarr-specs.readthedocs.io/en/latest/v3/codecs/bytes/v1.0.html
+- **implementations**: [zarr-python](https://github.com/zarr-developers/zarr-python), [zarr-rs](https://github.com/LDeakin/zarr_tools)
+- **licensing**: MIT
+- **notes**: Boundary codec in Zarr v3. Explicit endianness and layout order.
+
+---
+
+### plain-parquet
+
+- **canonical-id**: `plain-parquet`
+- **aliases**: `{parquet, "PLAIN"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "plain-parquet",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "physical_type": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#plain-plain--0
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Encoding varies by physical type. Fixed-width: packed little-endian. BYTE_ARRAY: 4-byte length + data. BOOLEAN: 8 per byte LSB-first.
+
+---
+
+### plain-dictionary
+
+- **canonical-id**: `plain-dictionary`
+- **aliases**: `{parquet, "PLAIN_DICTIONARY"}`, `{parquet, "RLE_DICTIONARY"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "plain-dictionary",
+  "inputs": {
+    "indices": {"type": "bytes", "required": true},
+    "dictionary": {"type": "bytes", "required": true},
+    "physical_type": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/encodings/#dictionary-encoding-plain_dictionary--2-and-rle_dictionary--8
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: `dictionary` is a second `bytes` input supplied by the format driver from a separate dictionary page. `PLAIN_DICTIONARY` (deprecated) and `RLE_DICTIONARY` are functionally equivalent.
+
+---
+
+### orc-rle-v1
+
+- **canonical-id**: `orc-rle-v1`
+- **aliases**: `{orc-v0, "RLE_V1"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "orc-rle-v1",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://orc.apache.org/specification/ORCv0/
+- **implementations**: [orc (Java)](https://github.com/apache/orc), [arrow (C++)](https://github.com/apache/arrow)
+- **licensing**: Apache 2.0
+- **notes**: Per-block adaptive selection between Run (fixed-delta) and Literals (varint). Signed integers use zigzag encoding internally. Not decomposable into a flat pipeline due to per-block strategy selection.
+
+---
+
+### orc-rle-v2
+
+- **canonical-id**: `orc-rle-v2`
+- **aliases**: `{orc-v1, "RLE_V2"}`, `{orc-v2, "RLE_V2"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "orc-rle-v2",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://orc.apache.org/specification/ORCv1/
+- **implementations**: [orc (Java)](https://github.com/apache/orc), [arrow (C++)](https://github.com/apache/arrow)
+- **licensing**: Apache 2.0
+- **notes**: Per-block adaptive selection from four sub-encodings via 2-bit header: Short Repeat, Direct (bitpacked), Patched Base, Delta. Not decomposable. Signed integers use zigzag internally.
+
+---
+
+### orc-varint
+
+- **canonical-id**: `orc-varint`
+- **aliases**: `{orc, "VARINT"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "orc-varint",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **pipeline** (decode, signed types):
+```json
+{
+  "codec_id": "orc-varint",
+  "direction": "decode",
+
+  "inputs": {
+    "bytes": {"type": "bytes"},
+    "dtype": {"type": "string"}
+  },
+
+  "outputs": {
+    "bytes": "unzigzag.signed_ints"
+  },
+
+  "steps": {
+    "decode_varint": {
+      "codec": "varint",
+      "inputs": {
+        "bytes": "input.bytes"
+      },
+      "outputs": {
+        "bytes": "raw_uints"
+      }
+    },
+    "unzigzag": {
+      "codec": "zigzag",
+      "inputs": {
+        "bytes": "decode_varint.raw_uints",
+        "dtype": "input.dtype"
+      },
+      "outputs": {
+        "bytes": "signed_ints"
+      }
+    }
+  }
+}
+```
+Note: For unsigned types, the zigzag step is skipped. Modeling of conditional steps is TBD.
+- **spec**: https://orc.apache.org/specification/ORCv1/
+- **implementations**: [orc (Java)](https://github.com/apache/orc)
+- **licensing**: Apache 2.0
+- **notes**: Base-128 variable-length integer encoding with zigzag for signed types. See generic `varint` and `zigzag` codecs.
+
+---
+
+### orc-string-direct
+
+- **canonical-id**: `orc-string-direct`
+- **aliases**: `{orc, "DIRECT"}` (string columns)
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "orc-string-direct",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "lengths": {"type": "bytes"},
+    "data": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://orc.apache.org/specification/ORCv1/
+- **implementations**: [orc (Java)](https://github.com/apache/orc), [arrow (C++)](https://github.com/apache/arrow)
+- **licensing**: Apache 2.0
+- **notes**: Fan-out: produces two output streams. Lengths stream uses orc-rle-v1 or orc-rle-v2.
+
+---
+
+### orc-string-dictionary
+
+- **canonical-id**: `orc-string-dictionary`
+- **aliases**: `{orc, "DICTIONARY"}`, `{orc, "DICTIONARY_V2"}` (string columns)
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "orc-string-dictionary",
+  "inputs": {
+    "indices": {"type": "bytes", "required": true},
+    "dictionary": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "data": {"type": "bytes"},
+    "lengths": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://orc.apache.org/specification/ORCv1/
+- **implementations**: [orc (Java)](https://github.com/apache/orc), [arrow (C++)](https://github.com/apache/arrow)
+- **licensing**: Apache 2.0
+- **notes**: Dictionary is per-stripe, supplied by format driver as a second `bytes` input. Multiple byte-stream inputs + multiple byte-stream outputs.
+
+---
+
+### vlen-utf8
+
+- **canonical-id**: `vlen-utf8`
+- **aliases**: `{numcodecs, "numcodecs.VLenUTF8"}`, `{zarr-v2, "numcodecs.VLenUTF8"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "vlen-utf8",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/vlen.html ⚠️
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Serializes variable-length UTF-8 strings into a flat bytestream. Output length not predictable from dtype/shape. Breaks fixed-stride assumption. Currently Zarr/Python-specific framing.
+
+---
+
+### vlen-bytes
+
+- **canonical-id**: `vlen-bytes`
+- **aliases**: `{numcodecs, "numcodecs.VLenBytes"}`, `{zarr-v2, "numcodecs.VLenBytes"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "vlen-bytes",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/vlen.html ⚠️
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Same as `vlen-utf8` but for arbitrary byte sequences. Same constraints apply.
+
+---
+
+### categorize
+
+- **canonical-id**: `categorize`
+- **aliases**: `{numcodecs, "numcodecs.Categorize"}`
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "categorize",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "labels": {"type": "string", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "astype": {"type": "string", "required": false, "default": "u1"}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://numcodecs.readthedocs.io/en/stable/categorize.html
+- **implementations**: [numcodecs](https://github.com/zarr-developers/numcodecs)
+- **licensing**: MIT
+- **notes**: Maps categorical values to integer codes. Analogous to dictionary encoding. Currently Python-specific in framing. Lossless if all values in labels; error on unknown values.
+
+---
+
+### zigzag
+
+- **canonical-id**: `zigzag`
+- **aliases**: embedded inside ORC and Protocol Buffers encodings
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "zigzag",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://protobuf.dev/programming-guides/encoding/#signed-ints
+- **implementations**: Trivially implementable; included in protobuf, ORC, and Arrow libraries.
+- **licensing**: Unencumbered
+- **notes**: Maps signed integers to unsigned: 0→0, -1→1, 1→2, -2→3, etc. Improves subsequent varint or bitpack encoding. Used implicitly inside `orc-rle-v1`, `orc-rle-v2`, `orc-varint`.
+
+---
+
+### varint
+
+- **canonical-id**: `varint`
+- **aliases**: `{protobuf, "varint"}`, `{orc, "base-128-varint"}`
+- **awareness**: type-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "varint",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://protobuf.dev/programming-guides/encoding/#varints
+- **implementations**: Included in all protobuf implementations; trivially implementable.
+- **licensing**: Unencumbered
+- **notes**: Base-128 variable-length integer encoding. 7 data bits + 1 continuation bit per byte. Handles any unsigned integer type — small values use fewer bytes regardless of declared type width. Signed types should be zigzag-encoded first. See `orc-varint` pipeline decomposition.
+
+---
+
+### zfp
+
+- **canonical-id**: `zfp`
+- **aliases**: `{hdf5, 32013}`, `{hdf5plugin, "hdf5plugin.Zfp"}` ⚠️ verify numcodecs identifier
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "zfp",
+  "inputs": {
+    "bytes":      {"type": "bytes", "required": true},
+    "dtype":      {"type": "string", "required": true},
+    "shape":      {"type": "uint[]", "required": true},
+    "mode":       {"type": "string", "required": true},
+    "rate":       {"type": "float", "required": false, "encode_only": true},
+    "precision":  {"type": "int", "required": false, "encode_only": true},
+    "accuracy":   {"type": "float", "required": false, "encode_only": true},
+    "minbits":    {"type": "int", "required": false, "encode_only": true},
+    "maxbits":    {"type": "int", "required": false, "encode_only": true},
+    "maxprec":    {"type": "int", "required": false, "encode_only": true},
+    "minexp":     {"type": "int", "required": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: Lindstrom, P. (2014). "Fixed-Rate Compressed Floating-Point Arrays." IEEE TVCG 20(12). https://zfp.readthedocs.io/
+- **implementations**: [zfp (C/C++, LLNL)](https://github.com/LLNL/zfp), [H5Z-ZFP (HDF5 filter)](https://github.com/LLNL/H5Z-ZFP), [hdf5plugin (Python)](https://github.com/silx-kit/hdf5plugin)
+- **licensing**: BSD 3-Clause
+- **notes**: Transform-based lossy compressor for spatially correlated floating-point and integer arrays. Partitions data into 4^d blocks, applies block-floating-point conversion, a custom near-orthogonal decorrelation transform (similar to DCT), coefficient reordering, and embedded bitplane coding with truncation. Supports 1D–4D arrays of float32, float64, int32, int64; works best on 2D+ data with spatial correlation. Five compression modes selected by `mode`: `rate` (fixed bits per value), `precision` (fixed bit planes), `accuracy` (absolute error tolerance), `expert` (direct control of minbits/maxbits/maxprec/minexp), `reversible` (lossless). Only the parameters relevant to the selected mode are used. All mode-specific parameters are encode-only — the compressed stream is self-describing for decompression. Internal stages are tightly integrated and not decomposable — truncation-based coding is fundamental to the algorithm and cannot be separated from the transform.
+
+---
+
+### sz2 ⚠️
+
+- **canonical-id**: `sz2`
+- **aliases**: `{hdf5, 32017}`, `{hdf5plugin, "hdf5plugin.SZ"}`
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "sz2",
+  "inputs": {
+    "bytes":               {"type": "bytes", "required": true},
+    "dtype":               {"type": "string", "required": true},
+    "shape":               {"type": "uint[]", "required": true},
+    "absolute":            {"type": "float", "required": false, "encode_only": true},
+    "relative":            {"type": "float", "required": false, "encode_only": true},
+    "pointwise_relative":  {"type": "float", "required": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: Di, S. & Cappello, F. (2016). "Fast Error-bounded Lossy HPC Data Compression with SZ." IEEE IPDPS. https://github.com/szcompressor/SZ2
+- **implementations**: [SZ2 (C)](https://github.com/szcompressor/SZ2), [H5Z-SZ (HDF5 filter)](https://github.com/disheng222/H5Z-SZ), [hdf5plugin (Python)](https://github.com/silx-kit/hdf5plugin)
+- **licensing**: BSD
+- **notes**: ⚠️ Prediction-based lossy compressor from Argonne National Laboratory. Predicts each value from its neighbors using Lorenzo predictor or linear regression, quantizes prediction errors within the error bound, and entropy-codes the residuals (Huffman + Zstd). Designed for large-scale scientific simulation output. Supports float32, float64, and integers at arbitrary dimensionality. While the backend (Huffman + Zstd) is conceptually generic compression, the prediction-quantization-entropy pipeline is co-designed with internal intermediate formats — not decomposable into catalog codecs. Error bound parameters are encode-only — the compressed stream stores the mode and tolerance. Superseded by SZ3; cannot decode SZ3 data, and SZ3 cannot decode SZ2 data. ⚠️ Verify exact parameter set for HDF5 filter interface.
+
+---
+
+### sz3 ⚠️
+
+- **canonical-id**: `sz3`
+- **aliases**: `{hdf5, 32024}`, `{hdf5plugin, "hdf5plugin.SZ3"}`
+- **awareness**: type-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "sz3",
+  "inputs": {
+    "bytes":                      {"type": "bytes", "required": true},
+    "dtype":                      {"type": "string", "required": true},
+    "shape":                      {"type": "uint[]", "required": true},
+    "absolute":                   {"type": "float", "required": false, "encode_only": true},
+    "relative":                   {"type": "float", "required": false, "encode_only": true},
+    "norm2":                      {"type": "float", "required": false, "encode_only": true},
+    "peak_signal_to_noise_ratio": {"type": "float", "required": false, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: Zhao et al. (2021). "Optimizing Error-Bounded Lossy Compression for Scientific Data by Dynamic Spline Interpolation." IEEE ICDE. https://github.com/szcompressor/SZ3
+- **implementations**: [SZ3 (C++)](https://github.com/szcompressor/SZ3), [hdf5plugin (Python)](https://github.com/silx-kit/hdf5plugin)
+- **licensing**: BSD
+- **notes**: ⚠️ Successor to SZ2 with a modular framework and improved spline interpolation predictor. Higher compression ratios than SZ2 in most cases. Incompatible compressed format with SZ2. Same algorithmic approach (prediction → quantization → entropy coding) but with better predictors and auto-tuning. Not decomposable into catalog codecs for the same reasons as SZ2. ⚠️ Verify exact parameter interface for the HDF5 filter.
+
+---
+
+
+## Structure-Aware Codecs
+
+---
+
+### transpose
+
+- **canonical-id**: `transpose`
+- **aliases**: `{zarr-v3, "transpose"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "transpose",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "shape": {"type": "uint[]", "required": true},
+    "order": {"type": "uint[]", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://zarr-specs.readthedocs.io/en/latest/v3/codecs/transpose/v1.0.html
+- **implementations**: [zarr-python](https://github.com/zarr-developers/zarr-python)
+- **licensing**: MIT
+- **notes**: Reorders axes before serialization. Equivalent to numpy transpose.
+
+---
+
+### tiff-predictor-2
+
+- **canonical-id**: `tiff-predictor-2`
+- **aliases**: `{tiff, "predictor=2"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "tiff-predictor-2",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "element_size": {"type": "uint", "required": true},
+    "row_width": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: TIFF 6.0 specification, Section 14
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+- **notes**: Horizontal differencing row-by-row. First sample per row stored as-is; subsequent samples store difference from previous. Requires `row_width` to know row boundaries.
+
+---
+
+### tiff-predictor-3
+
+- **canonical-id**: `tiff-predictor-3`
+- **aliases**: `{tiff, "predictor=3"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "tiff-predictor-3",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "element_size": {"type": "uint", "required": true},
+    "row_width": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: TIFF Technical Note 3 ⚠️ verify
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+- **notes**: Byte-level shuffle specific to IEEE 754 float layout, followed by horizontal differencing, row-by-row. ⚠️ Verify exact byte shuffle and delta order.
+
+---
+
+### png-filter
+
+- **canonical-id**: `png-filter`
+- **aliases**: `{numcodecs, "imagecodecs.pngfilter"}` ⚠️ verify
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "png-filter",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "element_size": {"type": "uint", "required": true},
+    "row_width": {"type": "uint", "required": true},
+    "filter_type": {"type": "string", "required": false, "default": "adaptive", "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://www.w3.org/TR/PNG/#9Filters
+- **implementations**: [libpng](http://www.libpng.org/pub/png/libpng.html), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+- **notes**: Five filter types: None, Sub, Up, Average, Paeth. On decode, filter type is read from per-scanline byte; `filter_type` only affects encoding. A genuinely separable precompression step — can be paired with any entropy coder.
+
+---
+
+### parquet-page-v1-split
+
+- **canonical-id**: `parquet-page-v1-split`
+- **aliases**: `{parquet, "data-page-v1"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "parquet-page-v1-split",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "rep_bytes": {"type": "bytes"},
+    "def_bytes": {"type": "bytes"},
+    "value_bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Self-delimiting — decoder parses through RLE-framed level sections to find boundaries. No external length metadata. Fragile; prefer v2.
+
+---
+
+### parquet-page-v2-split
+
+- **canonical-id**: `parquet-page-v2-split`
+- **aliases**: `{parquet, "data-page-v2"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "parquet-page-v2-split",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "rep_length": {"type": "uint", "required": true},
+    "def_length": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "rep_bytes": {"type": "bytes"},
+    "def_bytes": {"type": "bytes"},
+    "value_bytes": {"type": "bytes"}
+  }
+}
+```
+- **pipeline** (decode, dictionary-encoded column with zstd):
+```json
+{
+  "codec_id": "parquet-v2-dict-decode",
+  "direction": "decode",
+
+  "inputs": {
+    "data_page":       {"type": "bytes"},
+    "dict_page":       {"type": "bytes"},
+    "rep_length":      {"type": "uint"},
+    "def_length":      {"type": "uint"},
+    "index_bit_width": {"type": "uint"},
+    "physical_type":   {"type": "string"}
+  },
+
+  "constants": {
+    "def_bit_width": {"type": "uint", "value": 1},
+    "rep_bit_width": {"type": "uint", "value": 1}
+  },
+
+  "outputs": {
+    "values": "dict_lookup.decoded_values",
+    "def":    "decode_def.decoded_def",
+    "rep":    "decode_rep.decoded_rep"
+  },
+
+  "steps": {
+    "decompress_page": {
+      "codec": "zstd",
+      "inputs": {
+        "bytes": "input.data_page"
+      },
+      "outputs": {
+        "bytes": "decompressed_page"
+      }
+    },
+    "split_page": {
+      "codec": "parquet-page-v2-split",
+      "inputs": {
+        "bytes":      "decompress_page.decompressed_page",
+        "rep_length": "input.rep_length",
+        "def_length": "input.def_length"
+      },
+      "outputs": {
+        "rep_bytes":   "raw_rep",
+        "def_bytes":   "raw_def",
+        "value_bytes": "raw_indices"
+      }
+    },
+    "decode_indices": {
+      "codec": "rle-parquet",
+      "inputs": {
+        "bytes":     "split_page.raw_indices",
+        "bit_width": "input.index_bit_width"
+      },
+      "outputs": {
+        "bytes": "decoded_indices"
+      }
+    },
+    "dict_lookup": {
+      "codec": "plain-dictionary",
+      "inputs": {
+        "indices":       "decode_indices.decoded_indices",
+        "dictionary":    "input.dict_page",
+        "physical_type": "input.physical_type"
+      },
+      "outputs": {
+        "bytes": "decoded_values"
+      }
+    },
+    "decode_def": {
+      "codec": "rle-parquet",
+      "inputs": {
+        "bytes":     "split_page.raw_def",
+        "bit_width": "constant.def_bit_width"
+      },
+      "outputs": {
+        "bytes": "decoded_def"
+      }
+    },
+    "decode_rep": {
+      "codec": "rle-parquet",
+      "inputs": {
+        "bytes":     "split_page.raw_rep",
+        "bit_width": "constant.rep_bit_width"
+      },
+      "outputs": {
+        "bytes": "decoded_rep"
+      }
+    }
+  }
+}
+```
+Note: `def_bit_width` and `rep_bit_width` are constants for non-nested schemas; for nested schemas they become pipeline inputs derived from schema depth.
+- **spec**: https://parquet.apache.org/docs/file-format/data-pages/
+- **implementations**: [parquet-cpp](https://github.com/apache/arrow/tree/main/cpp/src/parquet), [parquet-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: `rep_length` and `def_length` from page header enable clean splitting. Preferred over v1.
+
+---
+
+### arrow-ipc-buffer
+
+- **canonical-id**: `arrow-ipc-buffer`
+- **aliases**: `{arrow-ipc, "buffer-layout"}`, `{lance, "arrow-buffer-layout"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "arrow-ipc-buffer",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "arrow_type": {"type": "string", "required": true},
+    "nullable": {"type": "bool", "required": true}
+  },
+  "outputs": {
+    "validity": {"type": "bytes"},
+    "offsets": {"type": "bytes"},
+    "values": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://arrow.apache.org/docs/format/Columnar.html
+- **implementations**: [arrow-cpp](https://github.com/apache/arrow), [arrow-rs](https://github.com/apache/arrow-rs)
+- **licensing**: Apache 2.0
+- **notes**: Output streams vary by type. Non-nullable primitive: only `values`. Nullable: `validity` + `values`. Variable-width: `validity` + `offsets` + `values`. Lance inherits this layout.
+
+---
+
+### tiff-old-jpeg
+
+- **canonical-id**: `tiff-old-jpeg`
+- **aliases**: `{tiff, 6}`
+- **awareness**: structure-aware
+- **lossiness**: lossy
+- **signature**:
+```json
+{
+  "codec_id": "tiff-old-jpeg",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "jpeg_tables": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: TIFF 6.0 specification (compression=6)
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+- **notes**: Old-style JPEG in TIFF. JPEG tables stored in IFD tags, supplied as second `bytes` input. Strongly prefer new-style JPEG (compression=7).
+
+---
+
+### ccitt-group3
+
+- **canonical-id**: `ccitt-group3`
+- **aliases**: `{tiff, 3}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "ccitt-group3",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "image_width": {"type": "uint", "required": true},
+    "t4_options": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: ITU-T T.4
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+- **notes**: Bilevel (1-bit) image data only. Row-oriented.
+
+---
+
+### ccitt-group4
+
+- **canonical-id**: `ccitt-group4`
+- **aliases**: `{tiff, 4}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "ccitt-group4",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "image_width": {"type": "uint", "required": true},
+    "t6_options": {"type": "uint", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: ITU-T T.6
+- **implementations**: [libtiff](https://libtiff.gitlab.io/libtiff/), [imagecodecs](https://github.com/cgohlke/imagecodecs)
+- **licensing**: Unencumbered
+- **notes**: Bilevel (1-bit) image data only. Pure 2D encoding — each row relative to previous.
+
+---
+
+### grib-grid-packing ⚠️
+
+- **canonical-id**: `grib-grid-packing`
+- **aliases**: `{grib-edition-1, "simple-packing"}`
+- **awareness**: structure-aware
+- **lossiness**: lossy
+- **signature**:
+```json
+{
+  "codec_id": "grib-grid-packing",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: https://library.wmo.int/records/item/35625-manual-on-codes-volume-i-2 ⚠️
+- **implementations**: [eccodes](https://github.com/ecmwf/eccodes), [cfgrib](https://github.com/ecmwf/cfgrib)
+- **licensing**: Apache 2.0 (eccodes)
+- **notes**: ⚠️ Low confidence. Scale factor and reference value in GRIB headers. Verify details.
+
+---
+
+### grib-ccsds ⚠️
+
+- **canonical-id**: `grib-ccsds`
+- **aliases**: `{grib-edition-2, "ccsds"}`, `{grib2, "template-42"}`
+- **awareness**: structure-aware
+- **lossiness**: lossless
+- **signature**:
+```json
+{
+  "codec_id": "grib-ccsds",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: CCSDS 121.0-B; WMO GRIB2 template 42 ⚠️
+- **implementations**: [eccodes](https://github.com/ecmwf/eccodes), [libaec](https://gitlab.dkrz.de/k202009/libaec)
+- **licensing**: Apache 2.0 (eccodes); BSD (libaec)
+- **notes**: ⚠️ Low confidence. Adaptive entropy coding from space mission data.
+
+---
+
+### grib-jpeg2000 ⚠️
+
+- **canonical-id**: `grib-jpeg2000`
+- **aliases**: `{grib-edition-2, "jpeg2000"}`, `{grib2, "template-40"}`
+- **awareness**: structure-aware
+- **lossiness**: conditional
+- **signature**:
+```json
+{
+  "codec_id": "grib-jpeg2000",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  }
+}
+```
+- **spec**: WMO GRIB2 template 40 ⚠️
+- **implementations**: [eccodes](https://github.com/ecmwf/eccodes)
+- **licensing**: Apache 2.0 (eccodes)
+- **notes**: ⚠️ Low confidence. JPEG2000 lossless or lossy.
+
+---
+
+
+## Explicitly Excluded
+
+The following appear in codec registries but are excluded from this catalog for
+the reasons given.
+
+### Serialization bridges
+
+These are object serialization formats, not typed array codecs.
+
+#### json (excluded)
+
+- **identifier**: `numcodecs.JSON`
+- **reason**: Encodes arbitrary Python objects to JSON bytes. Decode output is runtime-dependent, not a typed numeric array.
+
+#### msgpack (excluded)
+
+- **identifier**: `numcodecs.MsgPack`
+- **reason**: Binary serialization for arbitrary object graphs. Not a typed array codec.
+
+#### pickle (excluded)
+
+- **identifier**: `numcodecs.Pickle`
+- **reason**: Python-specific. Security concerns (arbitrary code execution on unpickle). Excluded on portability and security grounds.
+
+### Storage/access layer operations
+
+These are registered as codecs in their respective formats but do not perform
+data transformation in the sense used by this catalog. They reorganize how
+chunks are stored and accessed — a concern of the storage/access layer above
+the codec layer.
+
+#### sharding-indexed (excluded)
+
+- **identifier**: `{zarr-v3, "sharding_indexed"}`
+- **spec**: https://zarr-specs.readthedocs.io/en/latest/v3/codecs/sharding-indexed/index.html
+- **reason**: Sharding splits a chunk ("shard") into inner sub-chunks, applies a
+  nested codec pipeline to each sub-chunk independently, concatenates the
+  compressed sub-chunks, and appends an index (offset + length per sub-chunk)
+  with its own codec pipeline. This is a storage organization operation — it
+  determines how data is laid out within a storage object for efficient partial
+  access — not a data transformation. No byte of actual array data is
+  transformed by sharding itself; the transformation is delegated to the inner
+  codec pipelines, which are composed of codecs from this catalog.
+
+  Zarr v3 registers sharding as a codec (specifically an `array → bytes` codec)
+  because Zarr's codec pipeline is the only extension point available for
+  operations between the array and storage layers. This is an overloading of the
+  codec concept. A more precise model would place sharding in a separate
+  storage/access layer that sits above the codec layer — responsible for
+  chunking strategy, sub-chunk indexing, and partial I/O — while the codec layer
+  handles the actual data transformations applied to each chunk or sub-chunk.
+
+  The storage/access layer could potentially be modeled with a similar signature
+  and pipeline structure to what this catalog uses for codecs, but that design
+  is out of scope here.
+
+## Open Questions and Issues
+
+1. ~~**deflate framing variants (resolved)**~~: Handled by underlying zlib library.
+
+2. **lzo licensing**: GPL v2 may be problematic. Worth flagging.
+
+3. **szip losslessness for floats**: Verify whether SZIP guarantees losslessness for float data.
+
+4. **TIFF predictor-3 traversal order**: Verify row-by-row vs full flattened array.
+
+5. **ORC v2 codec evolution**: ORC v2 proposes removing RLEv1/v2. Monitor.
+
+6. **GRIB entries**: All three are low confidence.
+
+7. ~~**netcdf-classic-record (resolved)**~~: No separate entry needed.
+
+8. **variable-width-split pattern**: `orc-string-direct`, `arrow-ipc-buffer`, and `delta-length-byte-array` share a lengths/values split pattern. A generic `variable-width-split` codec may unify these, but input framing varies across formats. TBD.
+
+9. **Data-derived inputs**: Should data-derived inputs (built during encode, required for decode, stored in metadata — e.g. FSST `symbol_table`) have an explicit flag in the signature?
+
+10. ~~**Encode-only flag (resolved)**~~: Implemented as `encode_only: true` in signatures.
+
+11. **Format-specific algorithm variants**: `lzw-tiff` pattern. Check when expanding to new formats.
+
+12. **Checksum codec scope**: `fletcher32`, `crc32c`, and `crc32` are now included. Determine whether additional checksum variants exist in target formats. Adler-32 (used in zlib framing) is embedded in the deflate codec's framing, not a separate codec.
+
+13. ~~**Pipeline direction and inversion (resolved)**~~: Formalized in the Pipeline Model section. Reverse the DAG edges, swap pipeline inputs/outputs, constants unchanged, `encode_only` inputs active only when encoding.
+
+14. **Conditional pipeline steps**: `orc-varint` zigzag step is conditional on signedness. How to model conditional steps is TBD.
+
+15. **Per-block parameter variation**: `delta-parquet` uses per-block/per-miniblock parameters. These vary within a single stream, preventing direct pipeline decomposition. Consider block-wise parameterization.
+
+16. **TIFF bitpack alias**: Verify whether TIFF sub-byte sample packing is correctly modeled as `bitpack(bit_width=bits_per_sample)`.
+
+17. **Lance codec compatibility**: Verify whether Lance's RLE, dictionary, and delta are byte-compatible with Parquet/ORC variants.
+
+18. **Codec URI scheme**: Codec IDs will become versioned URIs. Scheme TBD.
+
+19. ~~**ZFP / SZ lossy compressors (resolved)**~~: Added as `zfp`, `sz2`, `sz3`. All are atomic (not decomposable). Transform-based (ZFP) vs prediction-based (SZ) approaches to error-bounded lossy compression.
+
+20. ~~**Zarr v3 sharding codec (resolved)**~~: Excluded. Sharding is a storage/access layer concern, not a data transformation. See the `sharding-indexed` entry in Explicitly Excluded.
+
+21. **Self-describing codec boundary**: Blosc is atomic because its header is embedded in the chunk. Formalize the criterion.
+
+22. **Awareness classification edge cases**: `tiff-predictor-2` and `png-filter` moved from stride-aware to structure-aware because they require `row_width` (dimensional layout knowledge). `delta` moved from stride-aware to type-aware because it requires `dtype` for overflow semantics. These reclassifications should be verified.
diff --git a/docs/reference/protospec/PROTOSPEC_NOTES.md b/docs/reference/protospec/PROTOSPEC_NOTES.md
new file mode 100644
index 0000000..0985db6
--- /dev/null
+++ b/docs/reference/protospec/PROTOSPEC_NOTES.md
@@ -0,0 +1,37 @@
+# Protospec Notes
+
+Commentary and clarifications on `PROTOSPEC.md`. That document is an exact copy of an upstream reference and is not modified here.
+
+## Codec signatures are written in decode direction by convention
+
+The protospec states "Individual codecs do not have a direction — every codec is bidirectional by definition." This is defensible but incomplete without qualification.
+
+"Bidirectional" means: a codec implements both encode and decode operations. It does not mean the signature is direction-neutral. All codec signatures in the catalog are written in **decode direction by convention**. By this convention, inputs represent the encoded/compressed form and outputs represent the decoded/raw form. The encode direction is the implied inverse. For `bytes → bytes` codecs this convention is not visible in the signature itself — both sides have the same type — it is only apparent from the semantic descriptions in the catalog entries. The only structural asymmetry the signature model captures is the `encode_only` property on inputs, which marks parameters that are active only during encoding (e.g. a compression level).
+
+For pipelines, the `direction` field makes the decode-first convention explicit: it tells the executor which way the DAG was written so it can traverse or invert it correctly. Individual codec signatures have no equivalent field — the decode-first convention is implicit and unstated in the protospec. The protospec should state that codec signatures describe the decode interface by convention. Without that, the claim that "codecs do not have a direction" is only coherent if the reader already knows the convention.
+
+## Signature model gaps
+
+### No `decode_only` property
+
+The protospec signature model has `encode_only` to mark inputs active only during encoding, but no `decode_only` counterpart. At least two codecs in the catalog have inputs meaningful only during decode:
+
+- **parquet-page-v2-split**: `rep_length` and `def_length` tell the decoder where to split the byte stream into three output streams. In the encode direction these lengths are not needed — they follow from the sizes of the three input buffers. The signature does not mark them as decode-only, and the model has no way to express this.
+
+- **fletcher32, crc32c, crc32**: `error_on_fail` controls whether to raise on checksum mismatch during decode. It is meaningless during encode (which unconditionally appends the checksum). It is not marked `encode_only` in the catalog — that would be wrong, it is the opposite — and the gap is unaddressed.
+
+Chonkle implements `decode_only` as the symmetric counterpart to `encode_only`. It is used by native codecs like packbits, where `encode_dtype` is encode-only and `decode_dtype` is decode-only. The executor omits `decode_only` ports from the port-map during encode, mirroring the existing `encode_only` handling during decode. See the `decode_only` divergence in [PIPELINE_SCHEMA.md](../PIPELINE_SCHEMA.md#decode_only-on-codec-signature-and-pipeline-inputs).
+
+### No variable-arity outputs
+
+`arrow-ipc-buffer` produces a different number of output buffers depending on `arrow_type` and `nullable`: non-nullable primitives produce only `values`; nullable types add `validity`; variable-width types add `offsets`. The signature declares all three outputs statically, but not all are produced for every call. The signature model has no mechanism to express conditional or variable-arity outputs.
+
+## Implications of "a pipeline is a codec"
+
+The protospec states that a pipeline is a codec, but does not spell out what that requires. Three things follow:
+
+**Derived signatures.** A pipeline's boundary must satisfy the same contract as a leaf codec signature: typed ports in, typed ports out. The protospec does not specify how to derive this signature from the pipeline JSON. In particular, it does not say how constants appear in the derived signature, and pipeline outputs are wiring references (`step_name.port_name`) rather than type descriptors — output types must be traced through step signatures.
+
+**Bidirectionality.** Since every codec implements both `encode` and `decode`, a pipeline must also be runnable in either direction. A pipeline that can only run in one direction is not a codec. This is why the executor must support DAG inversion.
+
+**Nesting.** A pipeline can appear as a step inside another pipeline. Real cases exist in the catalog (`orc-varint` is a two-step pipeline; `parquet-page-v2-split` appears as a step in the Parquet v2 dictionary-decode pipeline). Zarr v3 sharding is a case not in the catalog: the sharding codec wraps an inner codec pipeline (e.g. shuffle + zstd) applied per-chunk within the shard. The executor currently assumes every `codec_id` resolves to a leaf codec. Supporting nesting means handling the case where the resolver finds a pipeline JSON rather than a leaf codec for a given `codec_id`, and recursing into `run()`. No new concepts are required — the same DAG inversion and step execution mechanisms apply at both levels.
diff --git a/docs/wasm/COMPONENT_MODEL.md b/docs/wasm/COMPONENT_MODEL.md
new file mode 100644
index 0000000..fe98ce8
--- /dev/null
+++ b/docs/wasm/COMPONENT_MODEL.md
@@ -0,0 +1,52 @@
+# The WebAssembly Component Model
+
+The **Component Model** is a higher-level standard built on top of Core WebAssembly. It defines a binary format for *components* — self-describing Wasm binaries that declare their typed imports and exports in a machine-readable way — along with a composition model and an ABI for marshalling rich types across component boundaries.
+
+The Component Model is a Bytecode Alliance specification, separate from (but layered on top of) the Core Wasm spec. See [component-model.bytecodealliance.org](https://component-model.bytecodealliance.org/) for the full specification.
+
+## Core modules vs. components
+
+A **Core Wasm module** (`.wasm` with magic bytes `01 00 00 00`) exposes only Wasm primitives at its boundary: integers (`i32`, `i64`) and floats (`f32`, `f64`). To exchange bytes or strings with a host, the module and host must agree on a manual convention — pointers and lengths passed as integers, manual memory management on both sides. There is no machine-readable description of what the module expects or produces.
+
+A **component** (`.wasm` with magic bytes `0d 00 01 00`) wraps one or more Core modules and adds a typed interface layer. Its imports and exports are described using [WIT](WIT.md) (WebAssembly Interface Types), so tooling can verify the interface at load time and generate marshalling glue automatically. The host calls the component's exported functions passing real typed values, not raw pointers — the Canonical ABI handles the serialization.
+
+The two formats are mutually exclusive but complementary: components are built *from* Core modules (or from source via language toolchains), and the host runtime detects the format automatically by reading the magic bytes.
+
+## Why the Component Model exists
+
+Core Wasm's primitive-only boundary works, but it puts the burden on every codec author and every host to agree on the same manual convention. If the convention changes, or if two independently developed modules use slightly different conventions, things break silently. There is no tooling support, no interface verification, and no code generation.
+
+The Component Model solves this by making the interface explicit and verifiable:
+
+- **Interface verification at load time** — Wasmtime rejects a component whose exports don't match the expected WIT signature immediately, before any code runs.
+- **Code generation** — Language toolchains (Rust via `wit-bindgen`, Python via `componentize-py`, C via `wit-bindgen-c`) generate the marshalling glue from the WIT file. Component authors write functions against the WIT interface in their native language; the toolchain handles the rest.
+- **Ecosystem interoperability** — Any language with a Component Model toolchain can implement the same WIT interface, so components authored in different languages compose without manual coordination.
+
+## Component composition
+
+The Component Model's design intent is to enable *composition*: combining components by wiring one component's exports to another's imports, producing a new component. This is the "lego bricks" promise — build a pipeline by composing individually authored codec components.
+
+Current composition tooling:
+
+- **`wasm-tools compose`** — static composition tool: takes a root component and a set of dependencies, wires imports to exports, and produces a single composed component.
+- **`wac`** (WebAssembly Compositions) — a higher-level composition language that compiles to the same output as `wasm-tools compose`.
+
+**Current limitation**: these tools support static, single-topology linking. They wire one component's exports to another's imports in a fixed data-flow graph. They cannot generate two distinct wiring topologies within one component (e.g., a codec pipeline where encoding chains components in one order and decoding chains them in reverse).
+
+## Toolchain for authoring components
+
+| Language | Tool | Notes |
+| --- | --- | --- |
+| Python | `componentize-py` | Generates a WASIp2 component from a Python class; requires a `.wit` file |
+| Rust | `cargo-component` | Cargo subcommand; generates bindings from WIT |
+| Rust | `wit-bindgen` | Lower-level binding generator; used directly or via `cargo-component` |
+| C / C++ | `wit-bindgen-c` | Generates C headers from WIT |
+| Go | `wit-bindgen-go` | Generates Go bindings from WIT |
+
+## Further reading
+
+- [Component Model specification](https://component-model.bytecodealliance.org/)
+- [WIT format reference](https://component-model.bytecodealliance.org/design/wit.html) — see also [WIT.md](WIT.md) in this knowledge base
+- [Canonical ABI spec](https://github.com/WebAssembly/component-model/blob/main/design/mvp/CanonicalABI.md)
+- [componentize-py](https://github.com/bytecodealliance/componentize-py)
+- [wasm-tools](https://github.com/bytecodealliance/wasm-tools)
diff --git a/docs/wasm/EXECUTION_MODEL.md b/docs/wasm/EXECUTION_MODEL.md
new file mode 100644
index 0000000..90c1c20
--- /dev/null
+++ b/docs/wasm/EXECUTION_MODEL.md
@@ -0,0 +1,61 @@
+# Wasm Execution Model
+
+This document covers what actually happens when a Wasm module is compiled, instantiated, and called from a host program. The existing docs explain what Wasm is and how linear memory works; this one explains how execution works at runtime.
+
+## Compilation: .wasm → native machine code
+
+A `.wasm` file contains portable bytecode — instructions for the virtual CPU described in [OVERVIEW.md](OVERVIEW.md). Before any of it can run, a runtime like Wasmtime translates that bytecode into native machine code for the host CPU (ARM64, x86-64, etc.). This translation is called JIT (just-in-time) compilation, and Wasmtime does it using its Cranelift compiler.
+
+The result is native machine code — the same kind your OS runs from any compiled binary. Wasmtime can save this compiled output to disk as a `.cwasm` file. On subsequent runs, Wasmtime loads the `.cwasm` directly and skips recompilation. This is the compilation cache; it is why the first call to a codec is slow and subsequent calls are fast.
+
+## Instantiation: allocating runtime state
+
+Compiling a module produces code. Instantiating it produces a running copy of that code with its own state.
+
+Before instantiation, three objects must already exist. An `Engine` holds the compiler configuration and is typically created once per process. A `Store` is the container for all runtime state — instances, memories, tables — and is tied to an engine; it is the thing that actually owns memory allocations and gets passed to every Wasmtime call. A `Linker` is a registry of host-provided imports: you register host functions, a shared memory object, and WASI bindings into the linker before instantiation so it knows how to satisfy each import the module declares.
+
+With those in place, `linker.instantiate(store, module)` wires a compiled module to its imports and produces a live instance. Wasmtime does the following:
+
+1. **Allocates linear memory.** A contiguous byte array is allocated (via `mmap` or equivalent). Its initial size is specified in the module's binary. This step is skipped if the host pre-registered a memory object with the linker: the module's memory import is satisfied by that existing allocation instead of a fresh one. That is how multiple module instances can share the same backing buffer.
+
+2. **Writes data segments.** The module carries a data section — static content like string literals, lookup tables, and initialized global variables. Wasmtime copies these into the linear memory at the addresses the compiler chose. This is why two instances of the same binary cannot safely share a memory object: the destination offsets for data segments are fixed at compile time by the linker, not controlled by the host at instantiation. The second instantiation unconditionally reinitializes those same low addresses, stomping the first instance's data. Two instances of different binaries compiled to use non-overlapping regions could coexist in shared memory, but that requires deliberate coordination at compile time — it is not something Wasmtime or the host can arrange at runtime.
+
+3. **Sets up the function table.** Wasm supports indirect function calls (function pointers). The module's table lists which functions can be called indirectly. Wasmtime populates this table with references to the compiled native functions.
+
+4. **Resolves imports.** The module declares what it needs from the host: memory, functions, globals. Wasmtime connects each declared import to whatever the host provided via the linker — the shared memory object, the `request_output_buffer` callback, WASI functions, etc. Unresolved imports are an error at this step.
+
+5. **Returns an instance.** The instance is a handle to all of the above. Its exports (functions, memory, globals) are accessible to the host.
+
+## Calling a Wasm function
+
+There is no interpreter loop. When you call an exported Wasm function, you are calling a native function compiled from the Wasm bytecode. It runs on the host CPU in the same OS process as your Python program. The host thread that made the call is the thread the Wasm code runs on.
+
+Locals and function call frames live on the host's native call stack, the same stack your Python program uses. Values that do not fit in registers or whose address is taken spill into the module's linear memory via a shadow stack managed by the compiled code (usually pointed to by a global called `__stack_pointer`). This is distinct from the heap that `malloc` manages — it is the equivalent of the C call stack, but placed inside linear memory rather than on the OS stack.
+
+There is no context switch, no system call, and no process boundary when calling a Wasm function. The overhead is comparable to calling a native shared library function.
+
+## The sandbox: how isolation is enforced
+
+Wasm code has no pointers. All memory accesses use 32-bit integers as offsets into the module's linear memory array. The compiled native code translates every load and store into a bounds-checked array access: if the offset is out of range, execution traps immediately (an exception is thrown to the host). The module cannot construct an address to anything outside its linear memory — not the Python heap, not Wasmtime's internals, not the OS.
+
+On 64-bit systems, Wasmtime uses a virtual memory trick to make this essentially free: it `mmap`s a 4 GiB guard region around the linear memory. An out-of-bounds access faults at the OS level (SIGSEGV/SIGBUS) and is caught by Wasmtime's signal handler, converting it into a trap. No explicit bounds-check instruction is needed in the hot path.
+
+System calls are equally unavailable. Wasm has no `syscall` instruction. The only way a module can interact with the outside world is through imported functions the host explicitly provided. If the host did not give the module `fd_write`, the module cannot write to stdout. If the host did not provide `request_output_buffer`, the module cannot call it. The host controls the entire interface surface.
+
+## Linear memory from the host's perspective
+
+From the host's side, the linear memory is just a byte array. Wasmtime exposes it through methods like `mem.write(store, data, offset)` and `mem.read(store, start, end)`, which are single C-level `memcpy` calls into or out of the backing buffer.
+
+There is no marshalling, no type conversion, and no protocol. Writing bytes at an offset makes those bytes visible to the Wasm module at that same offset. The module reads them with a load instruction; the host reads them back with `mem.read`. Both sides see the same flat byte array.
+
+## Shared memory and zero-copy
+
+Normally each module instance has its own linear memory, allocated during instantiation. Wasmtime also supports importing an externally created memory object. When multiple module instances import the same memory object, they all operate on the same underlying byte array.
+
+A Wasm codec writing output to offset 0x500000 in a shared memory makes those bytes immediately visible to any other module instance that imports the same memory — at offset 0x500000, with no copy. The host can also read those bytes at any time via `mem.read`. This is the mechanism that makes zero-copy pipelines possible: the output offset from step N is passed directly as the input offset to step N+1.
+
+## What "VM" means and does not mean
+
+Wasm is called a virtual machine because it defines a portable bytecode format that any conforming runtime can execute. The portability is real: the same `.wasm` file runs on any OS and CPU.
+
+What "VM" does not imply here: a separate process, an interpreter running alongside your program, or a hypervisor. After compilation, Wasm code is native machine code executing on real hardware in the same process as the host. The "virtual" part is the instruction set the source language compiled to (the portable bytecode), not the thing that runs it.
diff --git a/docs/wasm/MEMORY.md b/docs/wasm/MEMORY.md
index 3002e7b..4557745 100644
--- a/docs/wasm/MEMORY.md
+++ b/docs/wasm/MEMORY.md
@@ -4,36 +4,37 @@ Wasm modules have their own linear memory — a flat, contiguous byte array. The
 
 ## Memory growth
 
-Linear memory starts at a fixed initial size (typically a few 64 KiB pages). When the module needs more memory — for example, when `malloc` can't satisfy a request from the existing heap — the memory must grow.
+Linear memory starts at a fixed initial size (typically a few 64 KiB pages). When the module needs more memory — for example, when a memory allocator can't satisfy a request from the existing heap — the memory must grow.
 
 ### How it works
 
-The WASI libc bundled into the `.wasm` binary includes a `dlmalloc` allocator. It manages a freelist within linear memory and handles growth automatically:
+Each module includes an allocator (provided by its language runtime) that manages a heap within linear memory. When the allocator cannot satisfy a request from existing space, it calls `memory.grow(n)` — a WebAssembly instruction that appends `n` 64 KiB pages to the end of linear memory. Those pages are always blank: the Wasm spec guarantees a module can never read data left behind by prior allocations or by the host process. If the module declares a maximum memory size and the request would exceed it, `memory.grow` returns -1 and the allocation fails.
 
-1. C code calls `malloc(size)`.
-2. `dlmalloc` checks its freelist. If it can satisfy the request from already-available space, it does.
-3. If not, `dlmalloc` calls `memory.grow(n_pages)` — a WebAssembly instruction that appends `n` 64 KiB pages to the end of linear memory.
-4. The Wasm runtime (wasmtime, wasmer, etc.) fulfills the growth. If the module declares a maximum memory size and the request exceeds it, `memory.grow` returns -1 and `malloc` returns `NULL`.
+This mechanism is the same regardless of source language. Rust, C, Go, and others each bring their own allocator, but all ultimately rely on `memory.grow` when the heap needs to expand.
 
-Because a WASI module is built with libc, all of this is handled transparently — the C code just calls `malloc`/`free` and never deals with `memory.grow` directly. Without WASI libc (a bare `wasm32-unknown` target), you would need to call `__builtin_wasm_memory_grow(0, n_pages)` yourself or have the host grow the memory from outside.
+### Initial memory size
+
+The initial number of pages is not set by the host at runtime — it is declared in the Wasm binary's `memory` section, written there by the codec's build toolchain at compile time. When wasmtime instantiates the component, it allocates exactly that many pages upfront.
+
+Most toolchains default to a small initial size (typically 1–16 pages, or 64 KiB–1 MiB). This means the first call to a codec with a large input will almost always trigger `memory.grow` to accommodate the data being passed in. A codec author can raise the initial size via build flags or linker configuration if the typical input size is known, avoiding that first growth at the cost of higher baseline memory use.
 
 ### Growth is contiguous and append-only
 
 `memory.grow` always appends pages to the end of linear memory. The entire address space remains one flat, contiguous byte array. Existing addresses and pointers are never invalidated by growth — new pages simply appear at higher addresses. This is simpler than native virtual memory, where `mmap` might return non-contiguous regions.
 
-However, linear memory **never shrinks** during the lifetime of an instance. `free` returns memory to `dlmalloc`'s freelist for reuse, but no pages are released back to the host. This means a module's memory high-water mark stays elevated after processing a large input. Long-lived instances can accumulate memory over time, which is one reason some designs spin up fresh instances per task.
+However, linear memory **never shrinks** during the lifetime of an instance. Freed memory returns to the allocator's internal freelist for reuse, but no pages are released back to the host. This means a module's memory high-water mark stays elevated after processing a large input. Long-lived instances can accumulate memory over time, which is one reason some designs spin up fresh instances per task.
 
 ## Performance considerations
 
-- **`memory.grow` is not free** — the runtime may need to zero-initialize new pages or update internal bookkeeping. Growing one page at a time in a loop is slower than growing many pages at once. `dlmalloc` handles this reasonably by requesting pages in batches.
-- **`realloc` may copy** — if `dlmalloc` can't extend an allocation in place, `realloc` allocates a new block, copies the data, and frees the old one. For large buffers this is expensive and temporarily uses 2x the memory.
-- **Preallocation helps** — when the output size is known, allocating once at the correct size avoids both realloc copies and incremental `memory.grow` calls.
+- **`memory.grow` has a fixed cost per call** — many small growths add up.
+- **Reallocating may copy** — if an allocation cannot be extended in place, the allocator allocates a new block, copies the data, and frees the old one. For large buffers this is expensive and temporarily uses 2x the memory.
+- **Preallocation helps** — when the output size is known, a codec can allocate the output buffer at the right size before processing begins, avoiding both copy-on-realloc and repeated `memory.grow` calls. The codec computes the size from inputs available in the port-map (dimensions, sample size, etc.) before any processing begins.
 
 ## Strategies for dynamically growing output buffers
 
 Some codecs — particularly compression or variable-length encoding — don't know the output size in advance. In those cases, the buffer must grow dynamically during processing. Common strategies:
 
-- **Geometric growth** — start with a reasonable initial size and double the buffer (via `realloc`) each time it fills up. This amortizes the cost of copies to O(1) per element over time, at the expense of up to 2x memory overallocation. This is the standard approach used by dynamic arrays (e.g., C++ `std::vector`). Each `realloc` that can't extend in place will copy the entire buffer to a new location, so the tradeoff is fewer, larger copies versus many small allocations.
-- **Worst-case preallocation** — if you can bound the maximum output size (e.g., LZ4 defines a maximum expansion ratio), allocate that size upfront and truncate or `realloc` down at the end. This avoids all intermediate copies but may waste memory if the bound is loose.
+- **Geometric growth** — start with a reasonable initial size and double the buffer each time it fills up. Because copies happen rarely, the total copying work stays proportional to the final output size, at the expense of up to 2x memory overallocation.
+- **Worst-case preallocation** — if you can bound the maximum output size (e.g., LZ4 defines a maximum expansion ratio), allocate that size upfront and trim at the end. This avoids all intermediate copies but may waste memory if the bound is loose.
 
 Since a Wasm host reads output as a contiguous region at a single pointer+length, the output must ultimately be a single contiguous buffer in linear memory. Geometric growth is the typical choice when the final size isn't known upfront.
diff --git a/docs/wasm/MULTI_MEMORY.md b/docs/wasm/MULTI_MEMORY.md
new file mode 100644
index 0000000..35bfa09
--- /dev/null
+++ b/docs/wasm/MULTI_MEMORY.md
@@ -0,0 +1,63 @@
+# Multi-Memory
+
+The multi-memory feature (part of WASM 3.0, enabled by default in wasmtime) allows a module to declare and access more than one linear memory. Each memory is addressed by index in load/store instructions.
+
+A natural use case is separating private module state from a shared data plane:
+
+| Memory | Role | Contents |
+|--------|------|----------|
+| memory 0 | **private** (each module defines its own) | data segments, shadow stack, malloc heap |
+| memory 1 | **shared** (imported from host, same object for all modules) | data plane -- inputs, outputs, config |
+
+The WASM spec guarantees that imports receive indices before locally defined entities, so a module that imports a memory and also defines its own will always have the import as memory 0 and the local definition as memory 1 (or vice versa, depending on declaration order).
+
+## Multi-memory works in hand-written WAT
+
+Wasmtime's [multi-memory example](https://docs.wasmtime.dev/examples-multimemory.html) demonstrates this feature using hand-written WAT. The memory index is a literal in each load/store instruction:
+
+```wat
+(func (export "load0") (param i32) (result i32)
+  (i32.load (memory 0) (local.get 0)))
+
+(func (export "load1") (param i32) (result i32)
+  (i32.load (memory 1) (local.get 0)))
+```
+
+A hand-written codec that explicitly targets memory 1 for data plane access can coexist with other modules -- each has its own memory 0, and all share memory 1.
+
+## The toolchain gap: C and Rust cannot generate multi-memory code
+
+LLVM's WASM backend cannot emit load/store instructions targeting a non-zero memory index from C or Rust source code. All load/store instructions target memory 0 unconditionally, and no source-level annotation changes this.
+
+Specifically:
+
+- In C, **`__attribute__((address_space(N)))` is silently ignored** by the WASM backend. This was confirmed in [WebAssembly/multi-memory#45](https://github.com/WebAssembly/multi-memory/issues/45).
+- **No one is actively working on this in LLVM.** A WebAssembly contributor confirmed this in the same issue.
+- **`--import-memory` in wasm-ld** controls whether memory 0 is imported or defined. It has no concept of a second memory.
+- **`--global-base`** controls the offset within a single memory, not which memory index to target.
+
+The runtime (wasmtime) fully supports multi-memory. The spec fully defines it. The gap is in code generation: no mainstream compiler can produce WASM that uses multiple memories from C or Rust source.
+
+## Consequence for library-wrapping modules
+
+A module that wraps an existing C library (zlib, zstd, lz4, etc.) compiles that library to WASM. The library's pointer dereferences all target memory 0. If a shared data plane is memory 1, the library cannot read from or write to it.
+
+The module would need to:
+
+1. Copy input from shared memory (memory 1) to a local buffer (memory 0)
+2. Call the library (operates on memory 0 pointers)
+3. Copy output from local buffer (memory 0) to shared memory (memory 1)
+
+This adds intra-step copies that would not exist if the library operated directly on its only memory.
+
+## `--global-base` coordination
+
+An alternative to multi-memory: each module is compiled with a unique `--global-base` value so their data segments occupy non-overlapping regions of a single shared memory. This avoids multi-memory entirely and preserves zero-copy.
+
+Downsides:
+
+- Requires compile-time coordination -- you must know what other modules will share the space and assign non-overlapping regions
+- Fragile if a module's heap grows beyond its reserved region
+- Adding a new module to the pipeline could require recompiling existing ones
+
+This is the most practical path to simultaneous instantiation on a single shared memory, but it sacrifices the property that modules are independently compiled and composed at runtime.
diff --git a/docs/wasm/WASI.md b/docs/wasm/WASI.md
index b475a61..7125ec0 100644
--- a/docs/wasm/WASI.md
+++ b/docs/wasm/WASI.md
@@ -4,12 +4,27 @@
 
 A common reason to use WASI is that C code calls `malloc` and `free`. These libc functions need an underlying memory allocator, which WASI provides. Without WASI, you'd have to write your own allocator from scratch.
 
-## The `wasm32-wasi` target
+## WASI Preview 1 vs Preview 2
+
+WASI has two major versions with different foundations:
+
+- **Preview 1** (`wasm32-wasi`) — APIs defined as raw Core Wasm imports: plain function signatures using `i32`/`i64` primitives. The component model did not exist yet. This is what the `wasm32-wasi` build target produces.
+- **Preview 2** (WASIp2) — WASI redefined using WIT and the Component Model. The system interfaces (`wasi:filesystem/types`, `wasi:clocks/wall-clock`, etc.) are now WIT interfaces, not raw imports. A WASIp2 component declares which WASI interfaces it imports in its world; Wasmtime satisfies them at load time.
+
+The Component Model did not replace WASI — WASI Preview 2 *adopted* the Component Model as its foundation. This is why tools like `componentize-py` (which compiles Python source into a Wasm component) are described as producing "WASIp2 components": the output is a Component Model component that may import WASIp2 WASI interfaces.
+
+### The `wasm32-wasi` target
 
 When compiling to Wasm, the build target `wasm32-wasi` means:
 
 - `wasm32` — 32-bit WebAssembly (pointers are 32 bits)
-- `wasi` — use the WASI system interface (as opposed to a bare `wasm32-unknown` target with no system interface at all)
+- `wasi` — use the WASI Preview 1 system interface (as opposed to `wasm32-unknown-unknown`, which has no system interface and is used for browser WebAssembly where JavaScript is the host)
+
+This produces a Core Wasm module with Preview 1 imports.
+
+### The `wasm32-wasip2` target
+
+`wasm32-wasip2` is the Preview 2 counterpart. It produces a Component Model component with WASIp2 imports. Not all toolchains require you to name this target explicitly — `componentize-py`, for example, compiles Python source directly to a WASIp2 component with no build target flag involved. Rust toolchains (`cargo-component`) do specify `wasm32-wasip2` explicitly in their build configuration.
 
 ## Command vs reactor
 
@@ -18,6 +33,8 @@ Wasm modules come in two execution models:
 - **Command** — has a `_start` function (like `main()` in C). The runtime calls `_start`, the module runs to completion, and then it's done. This is the default and is analogous to running a CLI program.
 - **Reactor** — has no `_start`. Instead it exports individual functions that the host calls on demand, potentially many times over the module's lifetime. The module stays alive between calls.
 
-A module is built as a reactor (via `-mexec-model=reactor` in the build flags) when the host needs to call exported functions as separate steps. For example, a codec module exports `alloc`, `encode`/`decode`, and `dealloc` — the host calls these in sequence, and the module must stay alive between calls so that memory allocated by `alloc` is still valid when the codec function reads it.
+A module is built as a reactor (via `-mexec-model=reactor` in the build flags) when the host needs to call exported functions on demand, potentially many times. The module stays alive between calls, preserving any state it initializes at startup.
+
+Core Wasm WASI reactor modules export a function called `_initialize` (instead of `_start`). The host must call `_initialize` once after instantiation to set up the WASI environment before calling any other exported functions.
 
-WASI reactor modules export a function called `_initialize` (instead of `_start`). The runtime must call `_initialize` once after instantiation to set up the WASI environment before calling any other exported functions.
+This project's codecs are [Component Model](COMPONENT_MODEL.md) components, not bare Core modules. They are conceptually reactors — exporting `encode` and `decode` for the host to call repeatedly — but `_initialize` is never called explicitly. Wasmtime's Component Model linker (`linker.instantiate()`) handles all initialization automatically during instantiation.
diff --git a/docs/wasm/WIT.md b/docs/wasm/WIT.md
index 690a913..f94cd9c 100644
--- a/docs/wasm/WIT.md
+++ b/docs/wasm/WIT.md
@@ -1,15 +1,35 @@
 # WIT — WebAssembly Interface Types
 
-**WIT** (WebAssembly Interface Types) is the interface definition language for the Wasm Component Model. It lets you describe the types and functions a Wasm component exposes in a language-neutral `.wit` file, so that tooling can generate glue code for any language on either side of the boundary.
+**WIT** (WebAssembly Interface Types) is the interface definition language for the Wasm Component Model. It lets you describe the types and functions a Wasm component exposes in a language-neutral `.wit` file, so that tooling can generate marshalling glue for both the component (Wasm side) and the host (the runtime loading it), regardless of what language each is written in.
 
 A `.wit` file is distinct from the component itself — it's a declaration of what the component looks like, not an implementation of it.
 
-## The four building blocks
+## Building blocks
+
+A complete `.wit` file brings all the building blocks together:
+
+```wit
+package chonkle:codec@0.1.0;
+
+interface transform {
+    type port-name = string;
+    type port-map = list<tuple<port-name, list<u8>>>;
+
+    encode: func(inputs: port-map) -> result<port-map, string>;
+    decode: func(inputs: port-map) -> result<port-map, string>;
+}
+
+world codec {
+    export transform;
+}
+```
+
+Each element (`package`, `interface`, `world`) has a distinct role, explained below.
 
 ### `package`
 
 ```wit
-package cylf:tiff-predictor-2-python@0.1.0;
+package chonkle:codec@0.1.0;
 ```
 
 Declares a namespace, name, and version for this group of definitions. The format is `<namespace>:<name>@<version>`. It's the Component Model's equivalent of a package identifier — it scopes interface names so they don't collide with those from other packages, and tooling uses it to refer to this component as a dependency.
@@ -17,30 +37,33 @@ Declares a namespace, name, and version for this group of definitions. The forma
 ### `interface`
 
 ```wit
-interface codec {
-    encode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-    decode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
+interface transform {
+    ...
 }
 ```
 
-Declares a named set of functions and types. An interface is a reusable unit — it can be exported by one component and imported by another without being tied to a specific world or implementation.
+Declares a named set of functions and types. This is the shared contract: every codec component in this project is compiled against the same `codec/wit/codec.wit`, which means every one of them implements the *same* `transform` interface — not just functions that happen to look the same, but functions explicitly identified by name and package. The host uses the same file to know what to call. That shared identity is what lets the toolchain verify compatibility and lets the host treat all codecs uniformly.
 
 ### `world`
 
 ```wit
-world tiff-predictor-2-python {
-    export codec;
+world codec {
+    export transform;
 }
 ```
 
 Declares what a component exposes (`export`) and what it needs from the host (`import`). A world is the complete contract for a specific component. One package can define multiple worlds.
 
-- `export codec` — the component provides the functions in the `codec` interface; any host that loads it can call them.
-- `import ...` — the component expects the host to provide the named interface or function.
+- `export transform` — the component provides the functions in the `transform` interface; any host that loads it can call them.
+- `import ...` — the component expects the host to provide the named interface or function. Common uses: WASI interfaces (`wasi:filesystem/types`, `wasi:clocks/wall-clock`) for components that need I/O or system services; host-provided callbacks; or interfaces exported by another component in a composed pipeline, which the tooling wires together by name.
+
+`codec.wit` has no imports intentionally — codec components are pure data transformation with no side effects and no host services needed. The one exception is `componentize-py`: it may silently inject WASI imports into the compiled binary because the Python runtime requires them (clocks, environment, etc.), even though `codec.wit` doesn't declare them. This is a toolchain implementation detail, not something you control in the WIT file.
+
+### Types
 
-### WIT types
+WIT has its own type system, separate from any host language. Types appear in function signatures, type aliases, and composite structures. The two things to understand are the built-in types and how to name them with aliases.
 
-WIT has its own type system, separate from any host language. Common types:
+#### Built-in types
 
 | WIT type                   | Meaning                                 |
 |----------------------------|-----------------------------------------|
@@ -55,36 +78,31 @@ WIT has its own type system, separate from any host language. Common types:
 | `record`                   | Named struct                            |
 | `variant`                  | Tagged union (like Rust `enum`)         |
 | `tuple<T, U>`              | Fixed-length heterogeneous sequence     |
+| `resource`                 | Handle to an opaque object              |
 
-`list<u8>` is the idiomatic byte array in WIT. `result<list<u8>, string>` means "returns bytes on success or an error string on failure" — analogous to Rust's `Result<Vec<u8>, String>`.
+Notes:
 
-## The Canonical ABI
-
-WIT only defines the interface. The **Canonical ABI** is the separate specification that defines how these high-level types are serialized into Wasm's low-level number primitives (`i32`, `i64`, `f32`, `f64`) and linear memory. When you compile a component with a WIT file, the toolchain generates glue code that handles all the pointer/length/memory management according to the Canonical ABI — you call `encode(data, config)` passing a byte list and a string directly, rather than manually allocating memory and passing pointers.
-
-This is the key difference from Core Wasm: with Core modules you manage the low-level representation yourself; with Components the toolchain does it for you.
+- `list<u8>` is the idiomatic byte array in WIT.
+- `result<port-map, string>` means "returns the output port map on success or an error string on failure" — analogous to Rust's `Result<T, String>`.
 
-## How this project relates to WIT
+#### Type aliases
 
-chonkle supports both Core Wasm modules and Component Model components; `wasm_runner.py` detects the format automatically. See [WASM.md](../WASM.md#choosing-between-core-modules-and-components) for a tradeoffs comparison and [CODEC_CONTRACT.md](../CODEC_CONTRACT.md) for the interface contract each format must satisfy.
+```wit
+type port-name = string;
+type port-map = list<tuple<port-name, list<u8>>>;
+```
 
-The tiff-predictor-2-python codec is a Component — it's authored in Python via componentize-py, which generates a Component by necessity. It is used as a test fixture in `tests/fixtures/chunks/cog_wasm/` and exercised by `test_wasm_runner.py` via the Component path in `wasm_runner.py`. Its WIT file:
+`type` declares a named alias within an interface. `port-name = string` is a readable label; `port-map = list<tuple<port-name, list<u8>>>` names a more complex structure — a list of `(name, bytes)` pairs — so WIT function signatures can write `port-map` rather than spelling out `list<tuple<port-name, list<u8>>>` every time.
 
-```wit
-package cylf:tiff-predictor-2-python@0.1.0;
+## The Canonical ABI
 
-interface codec {
-    encode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-    decode: func(data: list<u8>, config: string) -> result<list<u8>, string>;
-}
+WIT only defines the interface. The **Canonical ABI** is the separate specification that defines how these high-level types are serialized into Wasm's low-level number primitives (`i32`, `i64`, `f32`, `f64`) and linear memory. When you compile a component with a WIT file, the toolchain generates glue code that handles all the pointer/length/memory management according to the Canonical ABI — you call exported functions passing real typed values directly, rather than manually allocating memory and passing pointers.
 
-world tiff-predictor-2-python {
-    export codec;
-}
-```
+This is the key difference from Core Wasm: with Core modules you manage the low-level representation yourself; with Components the toolchain does it for you.
 
 ## Further reading
 
+- [Component Model](COMPONENT_MODEL.md) — what components are, composition, toolchain (this knowledge base)
+- [Resources](WIT_RESOURCES.md) — handle types, own vs borrow, destructors, sharing rules
 - [WIT format reference](https://component-model.bytecodealliance.org/design/wit.html)
-- [Component Model overview](https://component-model.bytecodealliance.org/)
 - [Canonical ABI spec](https://github.com/WebAssembly/component-model/blob/main/design/mvp/CanonicalABI.md)
diff --git a/docs/wasm/WIT_RESOURCES.md b/docs/wasm/WIT_RESOURCES.md
new file mode 100644
index 0000000..adac314
--- /dev/null
+++ b/docs/wasm/WIT_RESOURCES.md
@@ -0,0 +1,242 @@
+# Component Model Resources
+
+Resources are a distinct category of WIT type. Unlike a `string` or a `list<u8>`,
+a resource is not data you copy and pass around — it is an object that lives in one
+place and you hold a *handle* to it. A handle is a 32-bit integer index into a private
+table maintained by the runtime. Passing handles between components is cheap; the
+underlying data does not move.
+
+## Defining a resource in WIT
+
+A resource is defined inside an interface, which is exported from a world:
+
+```wit
+package a:comp@0.1.0;
+
+interface blobs {
+    resource blob {
+        constructor(data: list<u8>);
+        as-bytes: func() -> list<u8>;
+        size: func() -> u64;
+    }
+}
+
+world a {
+    export blobs;
+}
+```
+
+When a function returns a `blob`, it returns a handle, not the
+bytes themselves. The bytes stay inside the component that owns the resource.
+
+## Own vs borrow
+
+Every handle is either *owned* or *borrowed*. This determines who is responsible for
+the resource's eventual cleanup.
+
+**Owning a handle** means you are responsible for eventually returning it. When an owned
+handle is dropped (goes out of scope or is explicitly released), the runtime calls the
+resource's destructor — the cleanup function that frees whatever the resource was holding.
+
+**Borrowing a handle** is a temporary loan for the duration of a function call. You can
+call methods on it, but you cannot keep it, transfer it, or drop the underlying resource.
+When the function call ends, the borrow automatically expires. The original owner retains
+the handle.
+
+In WIT, the distinction appears in function signatures:
+
+```wit
+// Returns an owned handle — the caller now owns this blob
+make-blob: func(data: list<u8>) -> blob;
+
+// Takes a borrowed handle — only needs it for this call
+measure: func(b: borrow<blob>) -> u64;
+
+// Takes ownership — the callee is now responsible for cleanup
+consume: func(b: blob);
+```
+
+Method signatures always desugar to `borrow<self>` — calling a method on a resource
+does not transfer ownership. Constructors always return an owned handle.
+
+## Destructors
+
+A resource can declare a destructor — a function the runtime calls automatically when
+an owned handle is dropped. This is where cleanup happens: closing a file, freeing
+memory, ending a network connection.
+
+If no destructor is declared, dropping the handle simply removes the entry from the
+handle table. Any memory the resource was holding in the component's linear memory is
+not reclaimed — the Component Model has no implicit garbage collection. For short-lived
+component instances this may be acceptable, but within a long-running instance a
+resource without a destructor that frees its backing memory is a leak.
+
+The runtime guarantees the destructor runs when the owned handle is dropped, even if
+the drop is implicit. A borrowed handle being dropped does *not* trigger the destructor.
+Only the owner's final drop does.
+
+## The handle table
+
+Each component instance maintains its own private **handle table** — a list of all the
+resources it currently holds handles to. Handles are indices into this table. Index `3`
+in component A's table is unrelated to index `3` in component B's table.
+
+**A handle from one component is meaningless to another.** The runtime is the only thing
+that knows the mapping; a component cannot forge a handle or access a resource it was
+never given.
+
+When an owned handle crosses a component boundary, the runtime removes the entry from
+the sender's table and adds a new entry to the receiver's table — the integer index may
+change, but both point at the same underlying resource. For a `borrow<blob>`, the
+original entry stays in the owner's table and the runtime creates a temporary entry in
+the borrower's table that expires when the call returns.
+
+## Sharing resources between components
+
+For a handle to be usable across a component boundary, both components must use the
+resource type from the *same* interface definition. There are two patterns for
+achieving this.
+
+### Pattern 1: upstream component defines the resource
+
+Component A defines and exports the `blob` resource. Component B imports A's interface
+and uses the type. B's WIT requires two things: a `use` statement (type identity) and
+an `import` in the world (runtime instance dependency):
+
+```wit
+package b:comp@0.1.0;
+
+interface transform {
+    use a:comp/blobs@0.1.0.{blob};    // blob here is the same type as blob from A
+
+    process: func(input: borrow<blob>) -> list<u8>;
+}
+
+world b {
+    import a:comp/blobs@0.1.0;        // B must be linked to the same A instance
+    export transform;
+}
+```
+
+`use` tells the WIT compiler that `blob` in `process`'s signature is A's type.
+`import` in the world tells the linker that B requires A's interface to be wired in
+at compose time, and that handle tables for `blob` are shared with that A instance.
+
+**The constraint this creates**: B's WIT hardcodes a dependency on A's specific package.
+B must know at compile time which upstream component it will receive blobs from.
+
+### Pattern 2: shared third component defines the resource
+
+A third "buffer-store" component defines and exports the `blob` resource. Both A and B
+import from the same instance of that component, so a handle produced by A is valid
+when passed to B — they share the same handle table.
+
+```wit
+// buffer-store: defines the shared blob type
+package buffer:store@0.1.0;
+
+interface blobs {
+    resource blob {
+        constructor(data: list<u8>);
+        as-bytes: func() -> list<u8>;
+        size: func() -> u64;
+    }
+}
+
+world buffer-store {
+    export blobs;
+}
+```
+
+```wit
+// Component A: creates blobs
+package a:comp@0.1.0;
+
+interface produce {
+    use buffer:store/blobs@0.1.0.{blob};
+    make: func(data: list<u8>) -> blob;
+}
+
+world a {
+    import buffer:store/blobs@0.1.0;    // same interface as B
+    export produce;
+}
+```
+
+```wit
+// Component B: consumes blobs
+package b:comp@0.1.0;
+
+interface transform {
+    use buffer:store/blobs@0.1.0.{blob};
+    process: func(input: borrow<blob>) -> list<u8>;
+}
+
+world b {
+    import buffer:store/blobs@0.1.0;    // same interface as A
+    export transform;
+}
+```
+
+Both world declarations import `buffer:store/blobs@0.1.0`. When composition wires A and
+B to the **same instance** of the buffer-store, handles A produces are valid in B because
+they reference the same handle table. Sharing the interface name is necessary but not
+sufficient: A and B must be wired to the same actual instance, not two separate
+buffer-stores that happen to export identical interfaces.
+
+If two components each define their own `blob` type independently — even with identical
+WIT text — those types are different, and passing a handle from one to the other fails
+at link time.
+
+## Resources in practice: WASI file descriptors
+
+The most common real-world use of resources is WASI's file descriptor. In
+`wasi:filesystem/types`, a `descriptor` is a resource:
+
+```wit
+resource descriptor {
+    read-via-stream: func(offset: filesize) -> result<input-stream, error-code>;
+    write-via-stream: func(offset: filesize) -> result<output-stream, error-code>;
+    stat: func() -> result<descriptor-stat, error-code>;
+    // ...
+}
+```
+
+The host provides this resource. When a component opens a file, it receives an owned
+`descriptor` handle — a number. The actual file handle lives in the host's table. The
+component can read, write, and stat by calling methods through that number. When the
+component drops the descriptor, the host's destructor runs and the file is closed.
+
+The guest never sees the underlying OS file descriptor. It only sees its own handle
+table index. WASI streams, clocks, sockets, and most other WASI abstractions work
+the same way.
+
+## Lift, lower, and copy counts
+
+"Lifting" is reading a value out of a component's linear memory into a representation the host can work with. "Lowering" is the inverse: writing a value from the host into a component's linear memory. Every cross-component value transfer involves a lift followed by a lower.
+
+For a `list<u8>` payload, the lift materializes the data into a host-side buffer (copy 1: source component → host); the lower writes from that buffer into the destination component's memory (copy 2: host → destination component). In an orchestrated pipeline — where a host program routes data between independently compiled components — this occurs at every step boundary:
+
+```text
+component A runs
+  → lift: A's memory → host          (copy 1)
+orchestrator routes
+  → lower: host → B's memory         (copy 2)
+component B runs
+```
+
+This gives **2 copies per edge** for value types like `list<u8>` in any orchestrated component pipeline.
+
+## Resources do not reduce copies in orchestrated pipelines
+
+**Pattern 1 requires static coupling.** For B to accept a `blob` handle from A, B's WIT must contain `use a:comp/blobs@0.1.0.{blob}` — it must name A's package at compile time. In any pipeline where components are selected or assembled at runtime, no component knows its neighbors when it is compiled. Pattern 1 requires each component to hardcode its upstream neighbor in its WIT, which is incompatible with a generic plugin model.
+
+**Pattern 2 resources give the same copy count as value passing, through a different mechanism.** With `list<u8>`, data surfaces in the host at every step boundary — the orchestrator is in the data path. With Pattern 2 resources, the orchestrator only exchanges 32-bit handles and data never surfaces in the host, but data still crosses two component boundaries per edge: A calls `blob.constructor(data)`, copying its output into the buffer-store (copy 1: A's memory → buffer-store's memory), and B calls `blob.as-bytes()`, copying the data back out (copy 2: buffer-store's memory → B's memory). The buffer-store replaces the host as the intermediate stop, but the boundary count is the same:
+
+| Approach | Copies per edge |
+| --- | --- |
+| `list<u8>` with orchestrator | 2 |
+| Resources, Pattern 2 (shared buffer-store) | 2 |
+| Resources, Pattern 1 (direct composition) | 1 — requires static compile-time coupling |
+
+The only path to 1 copy per edge is direct composition (Pattern 1), which eliminates the intermediate component from the data path but requires static compile-time coupling between components.
diff --git a/pyproject.toml b/pyproject.toml
index 03b2d52..3f5bef1 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -5,12 +5,13 @@ readme = "README.md"
 requires-python = ">=3.13"
 version = "0.0.0"
 dependencies = [
-    "numcodecs",
-    "numpy",
     "oras",
     "wasmtime",
 ]
 
+[project.optional-dependencies]
+native = ["numcodecs", "numpy", "google-crc32c"]
+
 [project.scripts]
 chonkle = "chonkle.cli:main"
 
@@ -53,6 +54,7 @@ select = [
 
 [tool.ruff.lint.per-file-ignores]
 "tests/**" = ["S101"]
+"demo/**" = ["S101", "T20", "E501"]
 
 [tool.pytest.ini_options]
 markers = [
diff --git a/src/chonkle/__init__.py b/src/chonkle/__init__.py
index 3c0a53f..d95c627 100644
--- a/src/chonkle/__init__.py
+++ b/src/chonkle/__init__.py
@@ -1,13 +1 @@
-"""A codec pipeline library for encoding and decoding chunked array data."""
-
-from chonkle.codecs import BytesCodec, Endian, TiffPredictor2
-from chonkle.pipeline import decode, encode, get_codecs
-
-__all__ = [
-    "BytesCodec",
-    "Endian",
-    "TiffPredictor2",
-    "decode",
-    "encode",
-    "get_codecs",
-]
+"""Wasm codec pipeline library."""
diff --git a/src/chonkle/cli.py b/src/chonkle/cli.py
index 938bbd6..749452b 100644
--- a/src/chonkle/cli.py
+++ b/src/chonkle/cli.py
@@ -1,118 +1,252 @@
 """CLI entry point for chonkle."""
 
 import argparse
-import shutil
+import json
 import sys
 from pathlib import Path
+from typing import Any
 
-import numpy as np
-
-from chonkle.pipeline import decode, encode, get_codecs
+from chonkle.executor import run
+from chonkle.pipeline import Direction, prepare
+from chonkle.resolver import CodecInfo, Resolver
+from chonkle.wasm_signature import embed_signature
 
 
 def main() -> None:
     """Entry point for the chonkle CLI."""
     parser = argparse.ArgumentParser(
         prog="chonkle",
-        description="Utilities for encoding and decoding chunks.",
+        description="Execute a Wasm codec pipeline DAG.",
     )
-
     subparsers = parser.add_subparsers(dest="command", required=True)
 
-    # ── decode ──
-    dc_parser = subparsers.add_parser(
-        "decode",
-        help="Decode a chunk and display or save the result.",
+    # --- run command ---
+    run_parser = subparsers.add_parser(
+        "run",
+        help="Execute a pipeline JSON against named input data.",
     )
-    dc_parser.add_argument(
-        "chunk_path",
+    run_parser.add_argument(
+        "pipeline",
         type=Path,
-        help="Path to an encoded chunk file "
-        "(sidecar metadata expected at <path>.json).",
+        help="Path to a pipeline JSON file.",
     )
-    dc_parser.add_argument(
-        "--pipeline",
-        type=Path,
-        default=None,
-        help="Path to a pipeline JSON file. Defaults to <chunk_path>.json sidecar.",
+    run_parser.add_argument(
+        "--input",
+        metavar="NAME=FILE",
+        action="append",
+        default=[],
+        dest="inputs",
+        help="Named input port: --input bytes=chunk.bin (repeatable).",
     )
-    dc_parser.add_argument(
-        "-o",
+    run_parser.add_argument(
         "--output",
+        metavar="NAME=FILE",
+        action="append",
+        default=[],
+        dest="outputs",
+        help="Write named output port to file: --output bytes=result.bin (repeatable).",
+    )
+    run_parser.add_argument(
+        "--direction",
+        choices=["encode", "decode"],
+        default=None,
+        help=(
+            "Direction to run the pipeline "
+            "(default: the pipeline's declared direction). "
+            "Specifying the opposite direction inverts the DAG."
+        ),
+    )
+    run_parser.add_argument(
+        "--codec-store",
         type=Path,
         default=None,
-        help="Save the decoded array to a .npy file instead of printing.",
+        help="Path to the local codec store directory.",
+    )
+    run_parser.add_argument(
+        "--preference",
+        default=None,
+        help="Comma-separated backend preference order (e.g., 'core,component').",
+    )
+    run_parser.add_argument(
+        "--override",
+        metavar="ID=IMPL",
+        action="append",
+        default=[],
+        dest="overrides",
+        help="Override codec implementation: --override zlib=zlib-rs (repeatable).",
+    )
+    run_parser.add_argument(
+        "--source",
+        metavar="ID=URI",
+        action="append",
+        default=[],
+        dest="sources",
+        help=(
+            "Force-resolve a codec from a URI, bypassing the local store: "
+            "--source zlib=https://example.com/zlib.wasm (repeatable)."
+        ),
     )
 
-    # ── encode ──
-    en_parser = subparsers.add_parser(
-        "encode",
-        help="Encode a .npy array through a codec pipeline.",
+    # --- codecs command ---
+    codecs_parser = subparsers.add_parser(
+        "codecs",
+        help="List installed codec implementations in the local store.",
+    )
+    codecs_parser.add_argument(
+        "codec_id",
+        nargs="?",
+        default=None,
+        help="Filter to a specific codec_id.",
     )
-    en_parser.add_argument(
-        "input",
+    codecs_parser.add_argument(
+        "--codec-store",
         type=Path,
-        help="Path to a .npy file containing the array to encode.",
+        default=None,
+        help="Path to the local codec store directory.",
     )
-    en_parser.add_argument(
-        "-o",
-        "--output",
+
+    # --- embed-signature command ---
+    embed_parser = subparsers.add_parser(
+        "embed-signature",
+        help="Embed a signature JSON file into a .wasm binary as a custom section.",
+    )
+    embed_parser.add_argument(
+        "wasm_file",
         type=Path,
-        required=True,
-        help="Path for the encoded output. "
-        "Pipeline is read from <output>.json sidecar by default.",
+        help="Path to the .wasm binary (modified in-place).",
     )
-    en_parser.add_argument(
-        "--pipeline",
+    embed_parser.add_argument(
+        "signature_json",
         type=Path,
-        default=None,
-        help="Path to a pipeline JSON file. Defaults to <output>.json sidecar.",
+        help="Path to the signature JSON file.",
     )
 
     args = parser.parse_args()
-
-    if args.command == "decode":
-        _run_decode(args)
-    elif args.command == "encode":
-        _run_encode(args)
-
-
-def _run_decode(args: argparse.Namespace) -> None:
-    """Decode a chunk and print or save the result."""
-    pipeline_path = args.pipeline or args.chunk_path.parent / (
-        args.chunk_path.name + ".json"
+    if args.command == "run":
+        _run_command(args)
+    elif args.command == "codecs":
+        _codecs_command(args)
+    elif args.command == "embed-signature":
+        _embed_signature_command(args)
+
+
+def _build_resolver(
+    args: argparse.Namespace, pipeline_sources: dict[str, str]
+) -> Resolver:
+    """Build a Resolver from CLI flags and pipeline sources."""
+    preference = args.preference.split(",") if args.preference else None
+
+    overrides: dict[str, str] = {}
+    for spec in args.overrides:
+        if "=" not in spec:
+            sys.stderr.write(f"--override must be ID=IMPL, got {spec!r}\n")
+            sys.exit(1)
+        codec_id, impl = spec.split("=", 1)
+        overrides[codec_id] = impl
+
+    force_sources: dict[str, str] = {}
+    for spec in args.sources:
+        if "=" not in spec:
+            sys.stderr.write(f"--source must be ID=URI, got {spec!r}\n")
+            sys.exit(1)
+        codec_id, uri = spec.split("=", 1)
+        force_sources[codec_id] = uri
+
+    return Resolver(
+        codec_store=args.codec_store,
+        preference=preference,
+        overrides=overrides,
+        force_sources=force_sources,
+        pipeline_sources=pipeline_sources,
     )
-    codec_specs = get_codecs(pipeline_path)
-    data = args.chunk_path.read_bytes()
-    arr = decode(data, codec_specs)
-
-    if args.output is not None:
-        args.output.parent.mkdir(parents=True, exist_ok=True)
-        np.save(args.output, arr)
-        sys.stdout.write(f"Saved {arr.shape} {arr.dtype} array to {args.output}\n")
-    else:
-        sys.stdout.write(f"Shape: {arr.shape}, dtype: {arr.dtype}\n")
-        slices = tuple(slice(min(5, s)) for s in arr.shape)
-        sys.stdout.write(f"First 5x5:\n{arr[slices]}\n")
 
 
-def _run_encode(args: argparse.Namespace) -> None:
-    """Encode a .npy array through a codec pipeline."""
-    arr = np.load(args.input)
-
-    sidecar = args.output.parent / (args.output.name + ".json")
-    pipeline_path = args.pipeline if args.pipeline is not None else sidecar
+def _run_command(args: argparse.Namespace) -> None:
+    """Execute a pipeline and write or report outputs."""
+    with args.pipeline.open() as f:
+        pipeline_json: dict[str, Any] = json.load(f)
+
+    inputs: dict[str, bytes] = {}
+    for spec in args.inputs:
+        if "=" not in spec:
+            sys.stderr.write(f"--input must be NAME=FILE, got {spec!r}\n")
+            sys.exit(1)
+        name, path = spec.split("=", 1)
+        inputs[name] = Path(path).read_bytes()
+
+    raw_direction = args.direction or pipeline_json.get("direction")
+    if raw_direction not in ("encode", "decode"):
+        sys.stderr.write(f"Invalid or missing direction: {raw_direction!r}\n")
+        sys.exit(1)
+    direction: Direction = raw_direction
+    resolver = _build_resolver(args, pipeline_json.get("sources", {}))
+    prepared = prepare(pipeline_json, direction, resolver=resolver)
+    result = run(prepared, inputs)
+
+    requested_outputs: dict[str, str] = {}
+    for spec in args.outputs:
+        if "=" not in spec:
+            sys.stderr.write(f"--output must be NAME=FILE, got {spec!r}\n")
+            sys.exit(1)
+        name, path = spec.split("=", 1)
+        requested_outputs[name] = path
+
+    for port_name, data in result.items():
+        if port_name in requested_outputs:
+            out_path = Path(requested_outputs[port_name])
+            out_path.parent.mkdir(parents=True, exist_ok=True)
+            out_path.write_bytes(data)
+            sys.stdout.write(f"Wrote {len(data)} bytes to {out_path}\n")
+        else:
+            sys.stdout.write(f"Output {port_name!r}: {len(data)} bytes\n")
+
+
+def _print_table(cols: tuple[str, ...], rows: list[CodecInfo]) -> None:
+    """Print a simple aligned table with header and separator."""
+
+    def _cell(entry: CodecInfo, col: str) -> str:
+        return str(getattr(entry, col) or "(unnamed)")
+
+    widths = {c: max(len(c), max(len(_cell(r, c)) for r in rows)) for c in cols}
+    sep = "  "
+    header = sep.join(f"{c:<{widths[c]}s}" for c in cols)
+    rule = sep.join("─" * widths[c] for c in cols)
+    sys.stdout.write(f"{header}\n{rule}\n")
+    for row in rows:
+        line = sep.join(f"{_cell(row, c):<{widths[c]}s}" for c in cols)
+        sys.stdout.write(f"{line}\n")
+
+
+def _codecs_command(args: argparse.Namespace) -> None:
+    """List installed codec implementations."""
+    resolver = Resolver(codec_store=args.codec_store)
+    entries = resolver.list_codecs()
+
+    if args.codec_id:
+        entries = [e for e in entries if e.codec_id == args.codec_id]
+
+    if not entries:
+        if args.codec_id:
+            sys.stdout.write(f"No implementations found for {args.codec_id!r}\n")
+        else:
+            sys.stdout.write(f"No codecs installed in {resolver.store_path}\n")
+        return
+
+    if args.codec_id:
+        cols = ("implementation", "backend", "location")
+    else:
+        cols = ("codec_id", "implementation", "backend")
+    _print_table(cols, entries)
 
-    codec_specs = get_codecs(pipeline_path)
-    encoded = encode(arr, codec_specs)
 
-    args.output.parent.mkdir(parents=True, exist_ok=True)
-    args.output.write_bytes(encoded)
+def _embed_signature_command(args: argparse.Namespace) -> None:
+    """Embed a codec signature into a .wasm binary."""
+    wasm_bytes = args.wasm_file.read_bytes()
+    with args.signature_json.open() as f:
+        signature = json.load(f)
 
-    # Copy pipeline to sidecar if it came from elsewhere.
-    if args.pipeline is not None and args.pipeline.resolve() != sidecar.resolve():
-        shutil.copy2(args.pipeline, sidecar)
+    result = embed_signature(wasm_bytes, signature)
+    args.wasm_file.write_bytes(result)
 
-    sys.stdout.write(
-        f"Encoded {arr.shape} {arr.dtype} → {len(encoded)} bytes → {args.output}\n"
-    )
+    added = len(result) - len(wasm_bytes)
+    sys.stdout.write(f"Embedded {added} byte custom section in {args.wasm_file}\n")
diff --git a/src/chonkle/codecs.py b/src/chonkle/codecs.py
deleted file mode 100644
index c65341f..0000000
--- a/src/chonkle/codecs.py
+++ /dev/null
@@ -1,102 +0,0 @@
-"""Custom numcodecs codecs for chunk decoding."""
-
-from enum import StrEnum
-from typing import Any
-
-import numcodecs
-import numcodecs.abc
-import numpy as np
-
-
-class Endian(StrEnum):
-    """Byte order for multi-byte data types."""
-
-    LITTLE = "little"
-    BIG = "big"
-
-
-_ZARR_TO_NUMPY: dict[str, str] = {
-    "bool": "b1",
-    "int8": "i1",
-    "int16": "i2",
-    "int32": "i4",
-    "int64": "i8",
-    "uint8": "u1",
-    "uint16": "u2",
-    "uint32": "u4",
-    "uint64": "u8",
-    "float16": "f2",
-    "float32": "f4",
-    "float64": "f8",
-}
-
-
-class TiffPredictor2(numcodecs.abc.Codec):
-    """TIFF Predictor 2 (horizontal differencing).
-
-    Encoding stores per-row differences; decoding undoes this
-    via cumulative sum along the last axis.
-    """
-
-    codec_id = "tiff_predictor_2"
-
-    def encode(self, buf: Any) -> np.ndarray:
-        arr = np.asarray(buf)
-        out = np.empty_like(arr)
-        out[..., 0] = arr[..., 0]
-        out[..., 1:] = np.diff(arr, axis=-1)
-        return out
-
-    def decode(self, buf: Any, out: Any = None) -> np.ndarray:
-        arr = np.asarray(buf)
-        result = np.cumsum(arr, axis=-1, dtype=arr.dtype)
-        if out is not None:
-            np.copyto(out, result)
-            return out
-        return result
-
-
-def _numpy_dtype(data_type: str, endian: Endian) -> np.dtype:
-    """Build a numpy dtype from a data_type name and endianness."""
-    try:
-        type_char = _ZARR_TO_NUMPY[data_type]
-    except KeyError:
-        msg = f"Unsupported zarr data_type: {data_type}"
-        raise ValueError(msg) from None
-    byte_order = "<" if endian is Endian.LITTLE else ">"
-    return np.dtype(f"{byte_order}{type_char}")
-
-
-class BytesCodec(numcodecs.abc.Codec):
-    """Bytes codec: converts between raw bytes and typed numpy arrays."""
-
-    codec_id = "bytes"
-
-    def __init__(
-        self,
-        data_type: str,
-        endian: Endian | str,
-        shape: tuple[int, ...] | None = None,
-    ) -> None:
-        self.data_type = data_type
-        self.shape = shape
-        self.endian = Endian(endian)
-
-    def encode(self, buf: Any) -> bytes:
-        arr = np.asarray(buf)
-        dtype = _numpy_dtype(self.data_type, self.endian)
-        return arr.astype(dtype).tobytes()
-
-    def decode(self, buf: Any, out: Any = None) -> np.ndarray:
-        dtype = _numpy_dtype(self.data_type, self.endian)
-        arr = np.frombuffer(buf, dtype=dtype)
-        if self.shape is not None:
-            arr = arr.reshape(self.shape)
-        if out is not None:
-            np.copyto(out, arr)
-            return out
-        return arr.copy()
-
-
-numcodecs.register_codec(TiffPredictor2)
-numcodecs.register_codec(BytesCodec)
diff --git a/src/chonkle/codecs/__init__.py b/src/chonkle/codecs/__init__.py
new file mode 100644
index 0000000..48d5562
--- /dev/null
+++ b/src/chonkle/codecs/__init__.py
@@ -0,0 +1,6 @@
+"""Codec wrapper classes normalizing different backends.
+
+Each wrapper loads its signature at instantiation and exposes a uniform
+call(direction, port_map) interface so the executor does not need to
+know which backend is in use.
+"""
diff --git a/src/chonkle/codecs/_base.py b/src/chonkle/codecs/_base.py
new file mode 100644
index 0000000..f798159
--- /dev/null
+++ b/src/chonkle/codecs/_base.py
@@ -0,0 +1,68 @@
+"""Base types and ABC for codec wrappers."""
+
+from __future__ import annotations
+
+from abc import ABC, abstractmethod
+from enum import StrEnum
+from pathlib import Path
+from typing import TYPE_CHECKING, Any
+
+if TYPE_CHECKING:
+    from chonkle.pipeline import Direction
+    from chonkle.wasm_signature import Signature
+
+SIGNATURES_DIR = Path(__file__).parent.parent / "signatures" / "numcodecs"
+
+CODEC_TRANSFORM_IFACE = "chonkle:codec/transform@0.1.0"
+
+
+class Backend(StrEnum):
+    COMPONENT = "component"
+    CORE = "core"
+    NATIVE = "native"
+
+
+type PortMap = list[tuple[str, bytes]]
+
+
+class Codec(ABC):
+    """Abstract base class for codec wrappers.
+
+    The executor interacts only with this interface.  Concrete subclasses
+    handle Component Model, Core Wasm, and native backends.
+    """
+
+    @property
+    @abstractmethod
+    def codec_type(self) -> Backend:
+        """Backend type: "component", "core", or "native"."""
+        ...
+
+    @property
+    @abstractmethod
+    def codec_id(self) -> str:
+        """The logical codec identifier from the signature."""
+        ...
+
+    @property
+    @abstractmethod
+    def implementation(self) -> str:
+        """The specific implementation identifier from the signature."""
+        ...
+
+    @abstractmethod
+    def signature(self) -> Signature:
+        """Return the codec's signature (loaded at instantiation)."""
+        ...
+
+    @abstractmethod
+    def call(
+        self, direction: Direction, port_map: list[tuple[str, bytes | Any]]
+    ) -> list[tuple[str, bytes | Any]]:
+        """Execute encode or decode and return the output port-map.
+
+        Input and return entries may contain bytes values (all backends) or
+        CoreWasmRef deferred references (core wasm backend only).
+        See OutputPortMap type alias.
+        """
+        ...
diff --git a/src/chonkle/codecs/component.py b/src/chonkle/codecs/component.py
new file mode 100644
index 0000000..b9033db
--- /dev/null
+++ b/src/chonkle/codecs/component.py
@@ -0,0 +1,89 @@
+"""Component Model Wasm codec wrapper."""
+
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+import wasmtime
+import wasmtime.component
+
+from chonkle.codecs._base import CODEC_TRANSFORM_IFACE, Backend, Codec, PortMap
+from chonkle.pipeline import Direction
+from chonkle.wasm_signature import Signature
+
+
+class ComponentCodec(Codec):
+    """Wraps a Component Model Wasm codec.
+
+    Instantiates a Wasmtime Component Model component and calls
+    encode/decode via the WIT chonkle:codec/transform interface.
+    Signature is read from the chonkle:signature custom section at init.
+    """
+
+    def __init__(self, engine: wasmtime.Engine, wasm_path: Path) -> None:
+        self._engine = engine
+        self._wasm_path = wasm_path
+        self._sig = Signature.from_wasm(wasm_path)
+        self._component = wasmtime.component.Component.from_file(engine, str(wasm_path))
+
+    @property
+    def codec_type(self) -> Backend:
+        return Backend.COMPONENT
+
+    @property
+    def codec_id(self) -> str:
+        return self._sig.codec_id
+
+    @property
+    def implementation(self) -> str:
+        return self._sig.implementation
+
+    def signature(self) -> Signature:
+        return self._sig
+
+    def call(self, direction: Direction, port_map: PortMap) -> PortMap:
+        """Call encode or decode on the Component Model codec."""
+        store = wasmtime.Store(self._engine)
+        store.set_wasi(wasmtime.WasiConfig())
+        linker = wasmtime.component.Linker(self._engine)
+        linker.add_wasip2()
+
+        instance = linker.instantiate(store, self._component)
+        fn = self._get_function(instance, store, direction)
+
+        result = fn(store, port_map)
+        fn.post_return(store)
+
+        if isinstance(result, str):
+            msg = f"Codec component {direction} returned error: {result}"
+            raise RuntimeError(msg)
+
+        return [(str(name), bytes(data)) for name, data in result]
+
+    def _get_function(
+        self,
+        instance: wasmtime.component.Instance,
+        store: wasmtime.Store,
+        fn_name: str,
+    ) -> Any:
+        """Return the named function from the codec transform interface export."""
+        comp_exports = self._component.type.exports(self._engine)
+        item = comp_exports.get(CODEC_TRANSFORM_IFACE)
+        if isinstance(item, wasmtime.component.ComponentInstanceType):
+            iface_exports = item.exports(self._engine)
+            if fn_name in iface_exports and isinstance(
+                iface_exports[fn_name], wasmtime.component.FuncType
+            ):
+                iface_idx = instance.get_export_index(store, CODEC_TRANSFORM_IFACE)
+                if iface_idx is not None:
+                    fn_idx = instance.get_export_index(store, fn_name, iface_idx)
+                    if fn_idx is not None:
+                        return instance.get_func(store, fn_idx)
+
+        msg = (
+            f"Component at {self._wasm_path} does not export {fn_name!r} "
+            f"in the {CODEC_TRANSFORM_IFACE!r} interface. "
+            "Components must be compiled against codec.wit."
+        )
+        raise RuntimeError(msg)
diff --git a/src/chonkle/codecs/core.py b/src/chonkle/codecs/core.py
new file mode 100644
index 0000000..e737d5a
--- /dev/null
+++ b/src/chonkle/codecs/core.py
@@ -0,0 +1,281 @@
+"""Core Wasm codec wrapper, CoreWasmRef, and port-map serialization."""
+
+from __future__ import annotations
+
+import ctypes
+import struct
+from collections.abc import Sequence
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, cast
+
+import wasmtime
+
+from chonkle.codecs._base import Backend, Codec, PortMap
+from chonkle.pipeline import Direction
+from chonkle.wasm_signature import Signature
+
+type OutputPortMap = list[tuple[str, bytes | CoreWasmRef]]
+
+
+@dataclass
+class CoreWasmRef:
+    """Deferred reference to data in a core wasm module's linear memory.
+
+    The data remains in linear memory and is not copied to Python until
+    materialize() is called. This enables single-copy transfer between
+    sequential core wasm codec steps via ctypes.memmove.
+    """
+
+    codec: CoreWasmCodec
+    ptr: int
+    length: int
+
+    def materialize(self) -> bytes:
+        """Copy the data from linear memory to a Python bytes object."""
+        return bytes(
+            self.codec.memory.read(self.codec.store, self.ptr, self.ptr + self.length)
+        )
+
+
+def _require_export[T](
+    exports: Any, name: str, expected_type: type[T], wasm_path: Path
+) -> T:
+    """Get a named export and verify its type, raising on mismatch."""
+    val = exports[name]
+    if not isinstance(val, expected_type):
+        msg = (
+            f"Core wasm module at {wasm_path}: export {name!r} "
+            f"is {type(val).__name__}, expected {expected_type.__name__}"
+        )
+        raise RuntimeError(msg)
+    return val
+
+
+class CoreWasmCodec(Codec):
+    """Wraps a Core Wasm codec implementing the core port-map ABI.
+
+    Instantiates a core wasm32-wasi reactor module and calls
+    encode/decode using the binary port-map wire format via
+    Memory.read/Memory.write. The module instance is kept alive
+    for the codec's lifetime so downstream core codecs can single-copy
+    transfer data directly from this module's linear memory.
+
+    Required module exports: memory, alloc, dealloc, encode, decode.
+    """
+
+    def __init__(self, engine: wasmtime.Engine, wasm_path: Path) -> None:
+        self._engine = engine
+        self._wasm_path = wasm_path
+        self._sig = Signature.from_wasm(wasm_path)
+
+        module = wasmtime.Module.from_file(engine, str(wasm_path))
+        self._store = wasmtime.Store(engine)
+        self._store.set_wasi(wasmtime.WasiConfig())
+        linker = wasmtime.Linker(engine)
+        linker.define_wasi()
+        instance = linker.instantiate(self._store, module)
+
+        exports = instance.exports(self._store)
+        self._memory = _require_export(exports, "memory", wasmtime.Memory, wasm_path)
+        self._alloc_fn = _require_export(exports, "alloc", wasmtime.Func, wasm_path)
+        self._dealloc_fn = _require_export(exports, "dealloc", wasmtime.Func, wasm_path)
+        self._encode_fn = _require_export(exports, "encode", wasmtime.Func, wasm_path)
+        self._decode_fn = _require_export(exports, "decode", wasmtime.Func, wasm_path)
+
+    @property
+    def codec_type(self) -> Backend:
+        return Backend.CORE
+
+    @property
+    def codec_id(self) -> str:
+        return self._sig.codec_id
+
+    @property
+    def implementation(self) -> str:
+        return self._sig.implementation
+
+    def signature(self) -> Signature:
+        return self._sig
+
+    @property
+    def memory(self) -> wasmtime.Memory:
+        """The module's linear memory (exposed for single-copy transfer)."""
+        return self._memory
+
+    @property
+    def store(self) -> wasmtime.Store:
+        """The Wasmtime store bound to this instance."""
+        return self._store
+
+    def close(self) -> None:
+        """Release the Wasmtime store and its linear memory."""
+        self._store = None  # type: ignore[assignment]
+        self._memory = None  # type: ignore[assignment]
+
+    def _alloc(self, size: int) -> int:
+        """Allocate bytes in the module's linear memory."""
+        ptr = self._alloc_fn(self._store, size)
+        if ptr == 0:
+            msg = f"Core wasm alloc returned null for {size} bytes"
+            raise RuntimeError(msg)
+        return ptr
+
+    def call(self, direction: Direction, port_map: PortMap) -> OutputPortMap:
+        """Call encode or decode using the core port-map ABI.
+
+        Input port-map entries may be bytes or CoreWasmRef. When a
+        CoreWasmRef from another module is present, data is transferred
+        via single-copy (ctypes.memmove) directly into this module's
+        linear memory.
+
+        Output port-map entries are returned as CoreWasmRef (lazy
+        parsing — bulk data stays in linear memory until materialized or
+        single-copied to a downstream module).
+        """
+        fn = self._encode_fn if direction == "encode" else self._decode_fn
+
+        input_ptr, input_len = self._write_input(port_map)
+        result = fn(self._store, input_ptr, input_len)
+
+        output_ptr = (result >> 32) & 0xFFFFFFFF
+        output_len = result & 0xFFFFFFFF
+
+        if output_ptr == 0 and output_len == 0:
+            msg = f"Core wasm codec {direction} returned error"
+            raise RuntimeError(msg)
+
+        return _deserialize_port_map_lazy(self, output_ptr, output_len)
+
+    def _write_input(
+        self, port_map: Sequence[tuple[str, bytes | CoreWasmRef]]
+    ) -> tuple[int, int]:
+        """Write a port-map into this module's linear memory."""
+        if any(isinstance(v, CoreWasmRef) for _, v in port_map):
+            return self._write_input_with_refs(port_map)
+        data = _serialize_port_map(cast(PortMap, port_map))
+        ptr = self._alloc(len(data))
+        self._memory.write(self._store, data, ptr)
+        return ptr, len(data)
+
+    def _write_input_with_refs(
+        self, port_map: Sequence[tuple[str, bytes | CoreWasmRef]]
+    ) -> tuple[int, int]:
+        """Build a serialized port-map in linear memory with single-copy."""
+        entries: list[tuple[bytes, bytes | CoreWasmRef, int]] = []
+        total = 4  # entry_count
+        for name, value in port_map:
+            name_bytes = name.encode("utf-8")
+            data_len = value.length if isinstance(value, CoreWasmRef) else len(value)
+            total += 4 + len(name_bytes) + 4 + data_len
+            entries.append((name_bytes, value, data_len))
+
+        base = self._alloc(total)
+        offset = base
+
+        self._memory.write(self._store, struct.pack("<I", len(entries)), offset)
+        offset += 4
+
+        for name_bytes, value, data_len in entries:
+            self._memory.write(self._store, struct.pack("<I", len(name_bytes)), offset)
+            offset += 4
+            self._memory.write(self._store, name_bytes, offset)
+            offset += len(name_bytes)
+            self._memory.write(self._store, struct.pack("<I", data_len), offset)
+            offset += 4
+            if isinstance(value, CoreWasmRef):
+                _copy_between_memories(value, self, offset)
+            else:
+                self._memory.write(self._store, value, offset)
+            offset += data_len
+
+        return base, total
+
+
+def _serialize_port_map(port_map: PortMap) -> bytes:
+    """Serialize a port-map to the core ABI wire format."""
+    parts = [struct.pack("<I", len(port_map))]
+    for name, data in port_map:
+        name_bytes = name.encode("utf-8")
+        parts.append(struct.pack("<I", len(name_bytes)))
+        parts.append(name_bytes)
+        parts.append(struct.pack("<I", len(data)))
+        parts.append(data)
+    return b"".join(parts)
+
+
+def _deserialize_port_map(data: bytes) -> PortMap:
+    """Deserialize a port-map from the core ABI wire format."""
+    offset = 0
+    (entry_count,) = struct.unpack_from("<I", data, offset)
+    offset += 4
+    result: PortMap = []
+    for _ in range(entry_count):
+        (name_len,) = struct.unpack_from("<I", data, offset)
+        offset += 4
+        name = data[offset : offset + name_len].decode("utf-8")
+        offset += name_len
+        (data_len,) = struct.unpack_from("<I", data, offset)
+        offset += 4
+        port_data = data[offset : offset + data_len]
+        offset += data_len
+        result.append((name, bytes(port_data)))
+    return result
+
+
+def _deserialize_port_map_lazy(
+    codec: CoreWasmCodec, base_ptr: int, total_len: int
+) -> list[tuple[str, bytes | CoreWasmRef]]:
+    """Parse port-map metadata from linear memory, returning CoreWasmRef entries."""
+    store = codec.store
+    mem = codec.memory
+    pos = base_ptr
+
+    hdr = bytes(mem.read(store, pos, pos + 4))
+    (entry_count,) = struct.unpack("<I", hdr)
+    pos += 4
+
+    result: list[tuple[str, bytes | CoreWasmRef]] = []
+    for _ in range(entry_count):
+        nl = bytes(mem.read(store, pos, pos + 4))
+        (name_len,) = struct.unpack("<I", nl)
+        pos += 4
+
+        name = bytes(mem.read(store, pos, pos + name_len)).decode("utf-8")
+        pos += name_len
+
+        dl = bytes(mem.read(store, pos, pos + 4))
+        (data_len,) = struct.unpack("<I", dl)
+        pos += 4
+
+        result.append((name, CoreWasmRef(codec, pos, data_len)))
+        pos += data_len
+
+    return result
+
+
+def _copy_between_memories(
+    src_ref: CoreWasmRef, dst_codec: CoreWasmCodec, dst_offset: int
+) -> None:
+    """Copy data between two core wasm linear memories via ctypes.memmove."""
+    try:
+        src_base = src_ref.codec.memory.data_ptr(src_ref.codec.store)
+        dst_base = dst_codec.memory.data_ptr(dst_codec.store)
+        src_int = ctypes.cast(src_base, ctypes.c_void_p).value
+        dst_int = ctypes.cast(dst_base, ctypes.c_void_p).value
+        if src_int is None or dst_int is None:
+            msg = "Memory.data_ptr() returned a null pointer"
+            raise TypeError(msg)
+        ctypes.memmove(dst_int + dst_offset, src_int + src_ref.ptr, src_ref.length)
+    except (AttributeError, TypeError, OSError):
+        data = src_ref.materialize()
+        dst_codec.memory.write(dst_codec.store, data, dst_offset)
+
+
+def _single_copy_transfer(
+    src_ref: CoreWasmRef, dst_codec: CoreWasmCodec
+) -> CoreWasmRef:
+    """Transfer data between core wasm linear memories, returning a destination ref."""
+    dst_ptr = dst_codec._alloc(src_ref.length)
+    _copy_between_memories(src_ref, dst_codec, dst_ptr)
+    return CoreWasmRef(dst_codec, dst_ptr, src_ref.length)
diff --git a/src/chonkle/codecs/native.py b/src/chonkle/codecs/native.py
new file mode 100644
index 0000000..16949e1
--- /dev/null
+++ b/src/chonkle/codecs/native.py
@@ -0,0 +1,136 @@
+"""Native (numcodecs) codec wrapper."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+
+from chonkle.codecs._base import SIGNATURES_DIR, Backend, Codec, PortMap
+from chonkle.pipeline import Direction
+from chonkle.wasm_signature import Signature
+
+
+def _native_signature_path(codec_id: str) -> Path:
+    """Return the path to a native codec's bundled signature JSON file."""
+    sig_path = SIGNATURES_DIR / f"{codec_id}.json"
+    if not sig_path.exists():
+        msg = f"No native codec signature for {codec_id!r} (looked in {SIGNATURES_DIR})"
+        raise ValueError(msg)
+    return sig_path
+
+
+class NativeCodec(Codec):
+    """Wraps a numcodecs codec object.
+
+    Instantiated from a codec_id that must have a matching signature
+    file in src/chonkle/signatures/numcodecs/. numcodecs is imported
+    lazily — users who only use wasm codecs do not need it installed.
+
+    The native block in the signature JSON controls the calling
+    convention per direction:
+
+    - mode: "bytes": the bytes port is passed directly to
+      codec.encode() / codec.decode().
+    - mode: "ndarray": the bytes port is interpreted as a raw
+      buffer, converted to a numpy ndarray using the dtype from the
+      port named by dtype_port, processed, and converted back to
+      bytes.
+
+    constructor_ports lists input port names whose JSON-decoded
+    values are passed as kwargs to the numcodecs codec constructor.
+    """
+
+    def __init__(self, codec_id: str) -> None:
+        sig_path = _native_signature_path(codec_id)
+        raw = json.loads(sig_path.read_text())
+        self._sig = Signature.from_dict(raw)
+        if "native" not in raw:
+            msg = f"Signature for {codec_id!r} missing required 'native' block"
+            raise ValueError(msg)
+        self._native: dict[str, Any] = raw["native"]
+        self._numcodecs = _import_numcodecs()
+
+    @property
+    def codec_type(self) -> Backend:
+        return Backend.NATIVE
+
+    @property
+    def codec_id(self) -> str:
+        return self._sig.codec_id
+
+    @property
+    def implementation(self) -> str:
+        return self._sig.implementation
+
+    def signature(self) -> Signature:
+        return self._sig
+
+    def call(self, direction: Direction, port_map: PortMap) -> PortMap:
+        """Execute encode or decode via numcodecs."""
+        pm = dict(port_map)
+        data = pm.pop("bytes")
+
+        all_ports = {k: json.loads(v) for k, v in pm.items()}
+
+        constructor_kwargs = {
+            name: all_ports[name]
+            for name in self._native["constructor_ports"]
+            if name in all_ports
+        }
+        codec_obj = self._build_codec(constructor_kwargs)
+
+        recipe = self._native[direction]
+        if recipe["mode"] == "ndarray":
+            dtype_str = all_ports[recipe["dtype_port"]]
+            return self._call_ndarray(codec_obj, direction, data, dtype_str)
+        return self._call_bytes(codec_obj, direction, data)
+
+    def _build_codec(self, constructor_kwargs: dict[str, Any]) -> Any:
+        """Instantiate the numcodecs codec from constructor kwargs."""
+        return self._numcodecs.get_codec(
+            {"id": self._sig.codec_id, **constructor_kwargs}
+        )
+
+    @staticmethod
+    def _call_bytes(codec: Any, direction: Direction, data: bytes) -> PortMap:
+        result = codec.encode(data) if direction == "encode" else codec.decode(data)
+        return [("bytes", bytes(result))]
+
+    @staticmethod
+    def _call_ndarray(
+        codec: Any, direction: Direction, data: bytes, dtype_str: str
+    ) -> PortMap:
+        np = _import_numpy()
+        dtype = np.dtype(dtype_str)
+        arr = np.frombuffer(data, dtype=dtype)
+        result = codec.encode(arr) if direction == "encode" else codec.decode(arr)
+        if isinstance(result, np.ndarray):
+            return [("bytes", result.tobytes())]
+        return [("bytes", bytes(result))]
+
+
+def _import_numcodecs() -> Any:
+    """Lazily import numcodecs, raising a clear error if not installed."""
+    try:
+        import numcodecs
+    except ImportError:
+        msg = (
+            "numcodecs is required for native codec support but is not installed. "
+            "Install it with: pip install numcodecs"
+        )
+        raise ImportError(msg) from None
+    return numcodecs
+
+
+def _import_numpy() -> Any:
+    """Lazily import numpy, raising a clear error if not installed."""
+    try:
+        import numpy
+    except ImportError:
+        msg = (
+            "numpy is required for ndarray-format native codecs but is not installed. "
+            "Install it with: pip install numpy"
+        )
+        raise ImportError(msg) from None
+    return numpy
diff --git a/src/chonkle/executor.py b/src/chonkle/executor.py
new file mode 100644
index 0000000..32bb226
--- /dev/null
+++ b/src/chonkle/executor.py
@@ -0,0 +1,213 @@
+"""Execute a prepared DAG pipeline via codec wrappers."""
+
+from __future__ import annotations
+
+import json
+import logging
+
+from chonkle.codecs.core import CoreWasmCodec, CoreWasmRef, OutputPortMap
+from chonkle.pipeline import (
+    PreparedPipeline,
+    StepSpec,
+)
+
+log = logging.getLogger(__name__)
+
+
+def run(
+    prepared: PreparedPipeline,
+    inputs: dict[str, bytes],
+) -> dict[str, bytes]:
+    """Execute a prepared pipeline and return output port values.
+
+    Args:
+        prepared: A PreparedPipeline from prepare().
+        inputs: Pipeline-level input names mapped to byte values. For
+            forward execution (direction == pipeline.direction) keys
+            correspond to pipeline.inputs names. For inverted
+            execution keys correspond to pipeline.outputs names.
+
+    Returns:
+        Pipeline output names mapped to byte values.
+
+    Raises:
+        ValueError: A required input is missing.
+        RuntimeError: A codec call returns an error.
+    """
+    pipeline = prepared.pipeline
+    direction = prepared.direction
+    inverted = direction != pipeline.direction
+
+    if not inverted:
+        required = [
+            name
+            for name, desc in pipeline.inputs.items()
+            if (direction == "encode" or not desc.encode_only)
+            and (direction == "decode" or not desc.decode_only)
+        ]
+        for name in required:
+            if name not in inputs:
+                msg = f"Missing pipeline input: {name!r}"
+                raise ValueError(msg)
+    else:
+        for name in pipeline.outputs:
+            if name not in inputs:
+                msg = f"Missing pipeline input: {name!r}"
+                raise ValueError(msg)
+
+    if inverted:
+        return _execute_inverted(prepared, inputs)
+    return _execute_forward(prepared, inputs)
+
+
+def _execute_forward(
+    prepared: PreparedPipeline,
+    inputs: dict[str, bytes],
+) -> dict[str, bytes]:
+    """Execute steps in topological order, calling the direction function."""
+    pipeline = prepared.pipeline
+    direction = prepared.direction
+    value_store: dict[str, bytes | CoreWasmRef] = {}
+
+    for name, descriptor in pipeline.constants.items():
+        value_store[f"constant.{name}"] = json.dumps(descriptor.value).encode()
+
+    active_inputs = [
+        name
+        for name, desc in pipeline.inputs.items()
+        if (direction == "encode" or not desc.encode_only)
+        and (direction == "decode" or not desc.decode_only)
+    ]
+    for name in active_inputs:
+        value_store[f"input.{name}"] = inputs[name]
+
+    for step_name, step in pipeline.steps.items():
+        codec = prepared.codecs[step_name]
+        encode_only = prepared.encode_only_inputs[step_name]
+        decode_only = prepared.decode_only_inputs[step_name]
+        port_map = _forward_port_map(
+            step, value_store, direction, encode_only, decode_only, codec
+        )
+        log.debug(
+            "step %r: calling %s with %d ports", step_name, direction, len(port_map)
+        )
+        output_map = codec.call(direction, port_map)
+        for port_name, value in output_map:
+            value_store[f"{step_name}.{port_name}"] = value
+
+    return {
+        out_name: _materialize(value_store[str(ref)])
+        for out_name, ref in pipeline.outputs.items()
+    }
+
+
+def _execute_inverted(
+    prepared: PreparedPipeline,
+    inputs: dict[str, bytes],
+) -> dict[str, bytes]:
+    """Execute steps in reverse topological order, routing results backward."""
+    pipeline = prepared.pipeline
+    direction = prepared.direction
+    value_store: dict[str, bytes | CoreWasmRef] = {}
+
+    for name, descriptor in pipeline.constants.items():
+        value_store[f"constant.{name}"] = json.dumps(descriptor.value).encode()
+
+    for out_name, ref in pipeline.outputs.items():
+        value_store[str(ref)] = inputs[out_name]
+
+    for step_name, step in reversed(pipeline.steps.items()):
+        codec = prepared.codecs[step_name]
+        encode_only = prepared.encode_only_inputs[step_name]
+        decode_only = prepared.decode_only_inputs[step_name]
+        output_ports = prepared.output_ports[step_name]
+        port_map = _inverted_port_map(
+            step_name,
+            step,
+            value_store,
+            direction,
+            output_ports,
+            encode_only,
+            decode_only,
+            codec,
+        )
+        log.debug(
+            "step %r: calling %s with %d ports", step_name, direction, len(port_map)
+        )
+        output_map = codec.call(direction, port_map)
+        for port_name, value in output_map:
+            if port_name in step.inputs and port_name not in encode_only | decode_only:
+                value_store[str(step.inputs[port_name])] = value
+
+    return {
+        name: _materialize(value_store[f"input.{name}"])
+        for name, desc in pipeline.inputs.items()
+        if (direction == "encode" or not desc.encode_only)
+        and (direction == "decode" or not desc.decode_only)
+    }
+
+
+def _forward_port_map(
+    step: StepSpec,
+    value_store: dict[str, bytes | CoreWasmRef],
+    direction: str,
+    encode_only_inputs: frozenset[str],
+    decode_only_inputs: frozenset[str],
+    codec: object,
+) -> OutputPortMap:
+    """Build the port-map for a step in forward execution."""
+    is_core = isinstance(codec, CoreWasmCodec)
+    port_map: OutputPortMap = []
+    for port_name, ref in step.inputs.items():
+        if direction == "decode" and port_name in encode_only_inputs:
+            continue
+        if direction == "encode" and port_name in decode_only_inputs:
+            continue
+        value = value_store[str(ref)]
+        if isinstance(value, CoreWasmRef) and not is_core:
+            value = value.materialize()
+        port_map.append((port_name, value))
+    return port_map
+
+
+def _inverted_port_map(
+    step_name: str,
+    step: StepSpec,
+    value_store: dict[str, bytes | CoreWasmRef],
+    direction: str,
+    output_ports: tuple[str, ...],
+    encode_only_inputs: frozenset[str],
+    decode_only_inputs: frozenset[str],
+    codec: object,
+) -> OutputPortMap:
+    """Build the port-map for a step in inverted execution."""
+    is_core = isinstance(codec, CoreWasmCodec)
+    port_map: OutputPortMap = []
+    for port_name in output_ports:
+        val = value_store.get(f"{step_name}.{port_name}")
+        if val is not None:
+            if isinstance(val, CoreWasmRef) and not is_core:
+                val = val.materialize()
+            port_map.append((port_name, val))
+    direction_only = encode_only_inputs | decode_only_inputs
+    for port_name, ref in step.inputs.items():
+        if ref.kind == "constant" and port_name not in direction_only:
+            port_map.append((port_name, value_store[str(ref)]))
+    active_dir_ports = (
+        encode_only_inputs
+        if direction == "encode"
+        else decode_only_inputs
+        if direction == "decode"
+        else frozenset()
+    )
+    for port_name in active_dir_ports:
+        if port_name in step.inputs:
+            port_map.append((port_name, value_store[str(step.inputs[port_name])]))
+    return port_map
+
+
+def _materialize(value: bytes | CoreWasmRef) -> bytes:
+    """Resolve a value to bytes, materializing CoreWasmRef if needed."""
+    if isinstance(value, CoreWasmRef):
+        return value.materialize()
+    return value
diff --git a/src/chonkle/pipeline.py b/src/chonkle/pipeline.py
index 9f408fb..665e4c3 100644
--- a/src/chonkle/pipeline.py
+++ b/src/chonkle/pipeline.py
@@ -1,132 +1,487 @@
-"""Encode and decode raw chunks using metadata-driven codec pipelines."""
+"""DAG pipeline: parse, validate, and prepare pipelines for execution."""
+
+from __future__ import annotations
 
 import json
+from collections import defaultdict, deque
+from collections.abc import Mapping
+from dataclasses import dataclass
 from pathlib import Path
-from typing import Any, Literal
-
-import numcodecs
-import numpy as np
+from typing import TYPE_CHECKING, Any, Literal
 
-import chonkle.codecs  # noqa: F401 — register custom codecs
-from chonkle.wasm_runner import resolve_wasm_uri, wasm_decode, wasm_encode
+if TYPE_CHECKING:
+    from chonkle.codecs._base import Codec
+    from chonkle.resolver import Resolver
+    from chonkle.wasm_signature import PortDescriptor, Signature
 
 Direction = Literal["encode", "decode"]
 
 
-def get_codecs(source: Path | dict[str, Any]) -> list[dict[str, Any]]:
-    """Extract codec specs from a pipeline definition.
-
-    'source' may be a Path to a JSON file or a dict, both expected to
-    contain a "codecs" key.
-    """
-    if isinstance(source, Path):
-        with source.open() as f:
-            metadata: dict[str, Any] = json.load(f)
-    else:
-        metadata = source
-
-    if "codecs" not in metadata:
-        msg = f"Missing required key 'codecs' in {metadata!r}"
-        raise KeyError(msg)
-
-    return metadata["codecs"]
-
-
-def decode(data: bytes, codecs: list[dict[str, Any]]) -> np.ndarray:
-    """Decode raw bytes through a codec pipeline, returning an ndarray.
+@dataclass(frozen=True)
+class InputDescriptor:
+    """Descriptor for a pipeline-level input port."""
 
-    Codecs are applied in reverse order, unwinding the encoding.
-    """
-    result: bytes | np.ndarray = data
+    type: str = ""
+    encode_only: bool = False
+    decode_only: bool = False
 
-    for codec_spec in reversed(codecs):
-        result = _apply_step(codec_spec, result, "decode")
 
-    if not isinstance(result, np.ndarray):
-        msg = "Codec pipeline did not produce a numpy array"
-        raise TypeError(msg)
+@dataclass(frozen=True)
+class ConstantDescriptor:
+    """Descriptor for a pipeline-level constant."""
 
-    return result
+    type: str = ""
+    value: Any = None
 
 
-def encode(data: np.ndarray, codecs: list[dict[str, Any]]) -> bytes:
-    """Encode an ndarray through a codec pipeline, returning raw bytes.
+@dataclass
+class WiringRef:
+    """A parsed wiring reference from the pipeline JSON."""
 
-    Codecs are applied in forward order.
+    kind: Literal["input", "constant", "step"]
+    source: str  # "input", "constant", or step name
+    port: str  # port name within the source
+
+    @classmethod
+    def parse(cls, ref_str: str) -> WiringRef:
+        """Parse a dot-notation wiring reference string into a WiringRef.
+
+        Accepts the three wiring reference forms used in pipeline JSON:
+        "input.<port>", "constant.<port>", and "<step_name>.<port>".
+
+        Args:
+            ref_str: A dot-separated wiring reference of the form
+                <source>.<port>. The source is either the literal
+                string "input", the literal string "constant",
+                or the name of a pipeline step.
+
+        Returns:
+            A WiringRef with kind set to "input", "constant", or "step"
+            depending on the source token.
+
+        Raises:
+            ValueError: If ref_str does not contain a dot separator.
+        """
+        parts = ref_str.split(".", 1)
+        if len(parts) != 2:
+            msg = f"Invalid wiring reference {ref_str!r}: expected <source>.<port>"
+            raise ValueError(msg)
+
+        source, port = parts
+        if source == "input":
+            kind: Literal["input", "constant", "step"] = "input"
+        elif source == "constant":
+            kind = "constant"
+        else:
+            kind = "step"
+
+        return cls(kind=kind, source=source, port=port)
+
+    def __str__(self) -> str:
+        """Reconstruct the original dot-notation wiring reference string."""
+        return f"{self.source}.{self.port}"
+
+
+@dataclass
+class StepSpec:
+    """Specification for a single pipeline step."""
+
+    name: str  # unique DAG node identifier; used in wiring references
+    codec_id: str  # logical codec identifier; may repeat across steps
+    inputs: dict[str, WiringRef]  # port_name -> parsed wiring reference
+
+
+@dataclass
+class Pipeline:
+    """A parsed and validated DAG pipeline."""
+
+    codec_id: str
+    direction: Direction
+    inputs: dict[str, InputDescriptor]
+    constants: dict[str, ConstantDescriptor]
+    outputs: dict[str, WiringRef]  # pipeline_output_name -> parsed wiring ref
+    sources: dict[str, str]  # codec_id -> URI (advisory fetch hints)
+    steps: dict[str, StepSpec]  # keys in topological execution order
+
+    @classmethod
+    def parse(cls, pipeline: Path | dict[str, Any]) -> Pipeline:
+        """Parse a pipeline JSON document into a validated Pipeline.
+
+        Reads and deserializes the JSON if pipeline is a Path, then
+        constructs a Pipeline, runs wiring validation, and computes a
+        topological execution order for the steps.
+
+        Args:
+            pipeline: Either a Path to a JSON file on disk or a
+                pre-parsed dict representing the pipeline document.
+
+        Returns:
+            A fully validated Pipeline with steps ordered topologically.
+
+        Raises:
+            ValueError: If direction is missing or not "encode"/"decode",
+                if codec_id is not a string, if any wiring reference is
+                unresolvable or references a non-existent step, or if the
+                step dependency graph contains a cycle.
+        """
+        if isinstance(pipeline, Path):
+            with pipeline.open() as f:
+                data: dict[str, Any] = json.load(f)
+        else:
+            data = pipeline
+
+        direction = data.get("direction")
+        if direction not in ("encode", "decode"):
+            msg = f"'direction' must be 'encode' or 'decode', got {direction!r}"
+            raise ValueError(msg)
+
+        codec_id = data.get("codec_id")
+        if not isinstance(codec_id, str):
+            msg = f"'codec_id' must be a string, got {codec_id!r}"
+            raise ValueError(msg)
+        inputs = {
+            k: InputDescriptor(
+                type=v.get("type", ""),
+                encode_only=v.get("encode_only", False),
+                decode_only=v.get("decode_only", False),
+            )
+            for k, v in data.get("inputs", {}).items()
+        }
+        constants = {
+            k: ConstantDescriptor(type=v.get("type", ""), value=v.get("value"))
+            for k, v in data.get("constants", {}).items()
+        }
+        raw_outputs: dict[str, str] = dict(data.get("outputs", {}))
+        sources = dict(data.get("sources", {}))
+
+        steps: dict[str, StepSpec] = {}
+        for step_name, step_data in data.get("steps", {}).items():
+            steps[step_name] = StepSpec(
+                name=step_name,
+                codec_id=step_data["codec_id"],
+                inputs={
+                    k: WiringRef.parse(v)
+                    for k, v in step_data.get("inputs", {}).items()
+                },
+            )
+
+        outputs = {k: WiringRef.parse(v) for k, v in raw_outputs.items()}
+
+        _validate_wiring(steps, inputs, constants, outputs)
+        topo_order = _topological_sort(steps)
+        steps = {name: steps[name] for name in topo_order}
+
+        return cls(
+            codec_id=codec_id,
+            direction=direction,
+            inputs=inputs,
+            constants=constants,
+            outputs=outputs,
+            sources=sources,
+            steps=steps,
+        )
+
+
+@dataclass
+class PreparedPipeline:
+    """A pipeline that has been validated and is ready for execution.
+
+    Created by prepare(). Pass to run() to execute.
     """
-    result: bytes | np.ndarray = data
-
-    for codec_spec in codecs:
-        result = _apply_step(codec_spec, result, "encode")
-
-    if not isinstance(result, bytes):
-        msg = "Codec pipeline did not produce bytes"
-        raise TypeError(msg)
 
-    return result
+    pipeline: Pipeline
+    direction: Direction
+    codecs: Mapping[str, Codec]
+    encode_only_inputs: Mapping[str, frozenset[str]]  # step_name -> encode_only ports
+    decode_only_inputs: Mapping[str, frozenset[str]]  # step_name -> decode_only ports
+    output_ports: Mapping[str, tuple[str, ...]]  # step_name -> output port names
 
 
-# ── codec step dispatch ──
-
-
-def _apply_step(
-    codec_spec: dict[str, Any],
-    data: bytes | np.ndarray,
+def prepare(
+    pipeline: Path | dict[str, Any],
     direction: Direction,
-) -> bytes | np.ndarray:
-    """Apply one encode or decode step of the codec pipeline."""
-    codec_type = codec_spec["type"]
+    *,
+    resolver: Resolver | None = None,
+) -> PreparedPipeline:
+    """Parse, validate, and prepare a pipeline for execution.
+
+    Parses the pipeline JSON, resolves codec_ids via the resolver,
+    validates wiring against codec signatures, and validates all step
+    signatures. If this returns successfully, the pipeline is guaranteed
+    executable for the given direction.
+
+    Args:
+        pipeline: Either a Path to a pipeline JSON file or a pre-parsed
+            dict representing the pipeline document.
+        direction: Direction to execute ("encode" or "decode").
+        resolver: Codec resolver. If None, a default resolver is
+            created using the pipeline's sources field.
+
+    Returns:
+        A PreparedPipeline ready for executor.run().
+
+    Raises:
+        ValueError: Parsing fails, signature validation fails, a codec
+            cannot be resolved, or wiring references invalid output ports.
+    """
+    from chonkle.resolver import Resolver
+
+    parsed = Pipeline.parse(pipeline)
+
+    if resolver is None:
+        resolver = Resolver(pipeline_sources=parsed.sources)
+
+    codecs: dict[str, Codec] = {}
+    for step_name, step in parsed.steps.items():
+        codecs[step_name] = resolver.resolve(step.codec_id)
+
+    _validate_codec_signatures(parsed, codecs, direction)
+
+    encode_only_inputs: dict[str, frozenset[str]] = {}
+    decode_only_inputs: dict[str, frozenset[str]] = {}
+    output_ports: dict[str, tuple[str, ...]] = {}
+    for step_name, codec in codecs.items():
+        sig = codec.signature()
+        encode_only_inputs[step_name] = frozenset(sig.encode_only_inputs())
+        decode_only_inputs[step_name] = frozenset(sig.decode_only_inputs())
+        output_ports[step_name] = tuple(sig.outputs.keys())
+
+    return PreparedPipeline(
+        pipeline=parsed,
+        direction=direction,
+        codecs=codecs,
+        encode_only_inputs=encode_only_inputs,
+        decode_only_inputs=decode_only_inputs,
+        output_ports=output_ports,
+    )
+
+
+def _validate_wiring(
+    steps: dict[str, StepSpec],
+    inputs: dict[str, InputDescriptor],
+    constants: dict[str, ConstantDescriptor],
+    outputs: dict[str, WiringRef],
+) -> None:
+    """Validate step declarations and all wiring references."""
+    step_names = set(steps.keys())
+
+    for step in steps.values():
+        for port_name, ref in step.inputs.items():
+            _validate_wiring_ref(
+                inputs,
+                constants,
+                ref,
+                f"step {step.name!r} input {port_name!r}",
+                step_names,
+            )
+
+    for out_name, ref in outputs.items():
+        _validate_wiring_ref(
+            inputs,
+            constants,
+            ref,
+            f"pipeline output {out_name!r}",
+            step_names,
+        )
+
+
+def _validate_wiring_ref(
+    inputs: dict[str, InputDescriptor],
+    constants: dict[str, ConstantDescriptor],
+    ref: WiringRef,
+    context: str,
+    step_names: set[str],
+) -> None:
+    """Validate a single wiring reference against pipeline declarations."""
+    if ref.kind == "input":
+        if ref.port not in inputs:
+            msg = (
+                f"{context}: input {ref.port!r} is not declared in pipeline inputs"
+                f" (declared: {inputs})"
+            )
+            raise ValueError(msg)
+    elif ref.kind == "constant":
+        if ref.port not in constants:
+            msg = (
+                f"{context}: constant {ref.port!r} is not declared"
+                f" in pipeline constants (declared: {sorted(constants)})"
+            )
+            raise ValueError(msg)
+    else:
+        if ref.source not in step_names:
+            msg = f"{context}: step {ref.source!r} does not exist"
+            raise ValueError(msg)
+
+
+def _topological_sort(steps: dict[str, StepSpec]) -> list[str]:
+    """Return step names in a valid execution order using Kahn's algorithm."""
+    in_degree: dict[str, int] = dict.fromkeys(steps, 0)
+    dependents: dict[str, list[str]] = defaultdict(list)
+
+    for step_name, step in steps.items():
+        deps: set[str] = set()
+        for ref in step.inputs.values():
+            if ref.kind == "step":
+                deps.add(ref.source)
+        for dep in deps:
+            dependents[dep].append(step_name)
+            in_degree[step_name] += 1
+
+    queue: deque[str] = deque(name for name, deg in in_degree.items() if deg == 0)
+    order: list[str] = []
+
+    while queue:
+        name = queue.popleft()
+        order.append(name)
+        for dependent in dependents[name]:
+            in_degree[dependent] -= 1
+            if in_degree[dependent] == 0:
+                queue.append(dependent)
+
+    if len(order) != len(steps):
+        remaining = [n for n in in_degree if n not in set(order)]
+        msg = f"Pipeline contains a cycle involving steps: {remaining}"
+        raise ValueError(msg)
 
-    if codec_type == "numcodecs":
-        return _apply_numcodecs(codec_spec, data, direction)
-    if codec_type == "wasm":
-        return _apply_wasm(codec_spec, data, direction)
+    return order
 
-    msg = f"Unknown codec type: {codec_type!r}"
-    raise ValueError(msg)
 
+@dataclass
+class _ValidationContext:
+    """Accumulated cross-step state for signature validation."""
 
-def _apply_numcodecs(
-    codec_spec: dict[str, Any],
-    data: bytes | np.ndarray,
-    direction: Direction,
-) -> bytes | np.ndarray:
-    """Apply one step using a numcodecs codec."""
-    name = codec_spec["name"]
-    config = codec_spec.get("configuration", {})
-    codec_id = name.removeprefix("numcodecs.")
-
-    codec = numcodecs.get_codec({"id": codec_id, **config})
-    result = getattr(codec, direction)(data)
-    if result is None:
-        msg = f"Codec '{codec_id}' returned None"
-        raise ValueError(msg)
-    return result
+    pipeline: Pipeline
+    direction: Direction
+    sig_outputs: dict[str, set[str]]
+    step_output_types: dict[str, dict[str, str]]
 
 
-def _apply_wasm(
-    codec_spec: dict[str, Any],
-    data: bytes | np.ndarray,
+def _validate_codec_signatures(
+    pipeline: Pipeline,
+    codecs: Mapping[str, Codec],
     direction: Direction,
-) -> bytes | np.ndarray:
-    """Apply one step using a Wasm codec module."""
-    wasm_path = resolve_wasm_uri(codec_spec["uri"])
-    config = codec_spec.get("configuration", {})
-
-    # Wasm operates on raw bytes.  If the pipeline handed us an ndarray,
-    # convert to bytes and reconstruct the array from the output.
-    dtype: np.dtype | None = None
-    shape: tuple[int, ...] | None = None
-
-    if isinstance(data, np.ndarray):
-        dtype = data.dtype
-        shape = data.shape
-        data = data.tobytes(order="C")
-
-    wasm_fn = wasm_decode if direction == "decode" else wasm_encode
-    result_bytes = wasm_fn(wasm_path, data, config)
+) -> None:
+    """Validate all step signatures and collect errors into a single ValueError."""
+    errors: list[str] = []
+    ctx = _ValidationContext(
+        pipeline=pipeline,
+        direction=direction,
+        sig_outputs={},
+        step_output_types={},
+    )
+    for step_name, step in pipeline.steps.items():
+        sig = codecs[step_name].signature()
+        ctx.sig_outputs[step_name] = set(sig.outputs.keys())
+        ctx.step_output_types[step_name] = sig.output_types()
+        try:
+            _validate_step_signature(step, sig, ctx)
+        except ValueError as exc:
+            errors.append(str(exc))
+
+    for out_name, ref in pipeline.outputs.items():
+        if ref.kind == "step":
+            available = ctx.sig_outputs.get(ref.source, set())
+            if ref.port not in available:
+                errors.append(
+                    f"Pipeline output {out_name!r}: "
+                    f"step {ref.source!r} does not declare output port "
+                    f"{ref.port!r} (signature outputs: {sorted(available)})"
+                )
+
+    if errors:
+        raise ValueError("Pipeline signature validation failed:\n" + "\n".join(errors))
+
+
+def _validate_step_signature(
+    step: StepSpec,
+    signature: Signature,
+    ctx: _ValidationContext,
+) -> None:
+    """Verify a step's wiring and port declarations against its codec signature."""
+    errors: list[str] = []
+
+    for port_name, ref in step.inputs.items():
+        if ref.kind == "step":
+            available = ctx.sig_outputs.get(ref.source, set())
+            if ref.port not in available:
+                errors.append(
+                    f"Step {step.name!r} input {port_name!r}: "
+                    f"step {ref.source!r} does not declare output port "
+                    f"{ref.port!r} (signature outputs: {sorted(available)})"
+                )
+
+    sig_inputs = signature.inputs
+    if sig_inputs:
+        errors.extend(
+            f"input port {p!r} is missing required 'type' field"
+            for p, d in sig_inputs.items()
+            if not d.type
+        )
+
+        encode_only_ports = signature.encode_only_inputs()
+        decode_only_ports = signature.decode_only_inputs()
+
+        if ctx.direction == "decode":
+            valid_inputs = {
+                name for name, desc in sig_inputs.items() if not desc.encode_only
+            }
+        elif ctx.direction == "encode":
+            valid_inputs = {
+                name for name, desc in sig_inputs.items() if not desc.decode_only
+            }
+        else:
+            valid_inputs = set(sig_inputs.keys())
+
+        active_inputs = set(step.inputs.keys()) - encode_only_ports - decode_only_ports
+        unknown = active_inputs - valid_inputs
+        if unknown:
+            errors.append(
+                f"inputs {sorted(unknown)} are not valid signature "
+                f"{ctx.direction} inputs {sorted(valid_inputs)}"
+            )
+
+        errors.extend(_check_input_types(step, sig_inputs, active_inputs, ctx))
+
+    output_ports = signature.outputs
+    if output_ports:
+        errors.extend(
+            f"output port {p!r} is missing required 'type' field"
+            for p, d in output_ports.items()
+            if not d.type
+        )
+
+    if errors:
+        joined = "; ".join(errors)
+        msg = f"Step {step.name!r}: {joined}"
+        raise ValueError(msg)
 
-    if dtype is not None and shape is not None:
-        return np.frombuffer(result_bytes, dtype=dtype).reshape(shape, order="C")
 
-    return result_bytes
+def _check_input_types(
+    step: StepSpec,
+    sig_inputs: dict[str, PortDescriptor],
+    active_inputs: set[str],
+    ctx: _ValidationContext,
+) -> list[str]:
+    """Return type-mismatch error strings for each active wired input."""
+    errors: list[str] = []
+    for port_name in active_inputs:
+        if port_name not in sig_inputs:
+            continue
+        expected_type = sig_inputs[port_name].type
+        if not expected_type:
+            continue
+        ref = step.inputs[port_name]
+        if ref.kind == "input":
+            desc = ctx.pipeline.inputs.get(ref.port)
+            actual_type = desc.type if desc and desc.type else None
+        elif ref.kind == "constant":
+            cdesc = ctx.pipeline.constants.get(ref.port)
+            actual_type = cdesc.type if cdesc and cdesc.type else None
+        else:
+            actual_type = ctx.step_output_types.get(ref.source, {}).get(ref.port)
+        if actual_type is not None and actual_type != expected_type:
+            errors.append(
+                f"input {port_name!r}: wiring {str(ref)!r} provides type"
+                f" {actual_type!r} but codec expects {expected_type!r}"
+            )
+    return errors
diff --git a/src/chonkle/resolver.py b/src/chonkle/resolver.py
new file mode 100644
index 0000000..794b6e1
--- /dev/null
+++ b/src/chonkle/resolver.py
@@ -0,0 +1,316 @@
+"""Codec resolution: maps codec_ids to Codec instances.
+
+The Resolver handles finding, downloading, and instantiating codec
+implementations for a pipeline. It maintains a local codec store
+indexed by codec_id and implementation name. Set
+CHONKLE_FORCE_INSTALL=1 to overwrite an existing store entry when
+re-installing a codec from a source URI.
+"""
+
+from __future__ import annotations
+
+import os
+import re
+import shutil
+from collections.abc import Sequence
+from dataclasses import dataclass
+from pathlib import Path
+
+import wasmtime
+
+from chonkle.codecs._base import SIGNATURES_DIR, Backend, Codec
+from chonkle.codecs.component import ComponentCodec
+from chonkle.codecs.core import CoreWasmCodec
+from chonkle.codecs.native import NativeCodec
+from chonkle.wasm_signature import Signature, detect_codec_type
+
+
+def _default_store_path() -> Path:
+    """Return the default local codec store path."""
+    override = os.environ.get("CHONKLE_CODEC_STORE", "")
+    if override:
+        return Path(override)
+    return Path.home() / ".chonkle" / "codecs"
+
+
+@dataclass(frozen=True)
+class CodecInfo:
+    """Metadata about a codec implementation."""
+
+    codec_id: str
+    implementation: str
+    backend: Backend
+    location: Path  # .wasm for wasm entries, .json for native
+
+
+class CodecStore:
+    """Local codec store: scan, install, and list wasm codec entries."""
+
+    def __init__(self, store_path: Path) -> None:
+        self._store_path = store_path
+        self._index: dict[str, dict[str, CodecInfo]] | None = None
+
+    @property
+    def path(self) -> Path:
+        return self._store_path
+
+    def get_index(self) -> dict[str, dict[str, CodecInfo]]:
+        """Return the codec index, scanning the store on first access."""
+        if self._index is None:
+            self._index = _scan_store(self._store_path)
+        return self._index
+
+    def install(self, wasm_path: Path, *, force_install: bool = False) -> CodecInfo:
+        """Install a wasm file into the store and return its entry."""
+        sig = Signature.from_wasm(wasm_path)
+        backend = detect_codec_type(wasm_path)
+
+        impl_name = _sanitize_filename(sig.implementation or wasm_path.stem)
+        dest_dir = self._store_path / sig.codec_id
+        dest_dir.mkdir(parents=True, exist_ok=True)
+        dest = dest_dir / f"{impl_name}.wasm"
+
+        if not dest.exists() or force_install or _should_force_install():
+            shutil.copy2(wasm_path, dest)
+
+        entry = CodecInfo(
+            codec_id=sig.codec_id,
+            implementation=sig.implementation,
+            backend=backend,
+            location=dest,
+        )
+        self._index = None  # invalidate so next access re-scans
+        return entry
+
+    def list_entries(self) -> list[CodecInfo]:
+        """Return all wasm entries in the store."""
+        result: list[CodecInfo] = []
+        for entries in self.get_index().values():
+            result.extend(entries.values())
+        return result
+
+
+class Resolver:
+    """Maps codec_ids to Codec instances using a resolution chain.
+
+    Resolution order:
+
+    1. Explicit paths (if provided via paths)
+    2. Force sources (download from URI regardless of store)
+    3. Per-codec overrides (specific implementation name)
+    4. Local store and native (selected by backend preference)
+    5. Pipeline sources (download and install)
+
+    Args:
+        codec_store: Path to the local codec store directory.
+            Defaults to CHONKLE_CODEC_STORE env var or
+            ~/.chonkle/codecs.
+        preference: Ordered list of backend types to prefer when
+            multiple implementations exist for a codec_id.
+            Supported values: "core", "component", "native".
+        overrides: Maps codec_id to a specific implementation name,
+            bypassing the preference system.
+        force_sources: Maps codec_id to a download URI. Unlike
+            pipeline_sources, these are fetched unconditionally,
+            bypassing the local store. The fetched binary is
+            installed into the store (overwriting any existing
+            entry with the same implementation name).
+        pipeline_sources: Maps codec_id to a download URI (from the
+            pipeline's sources field).
+        paths: Maps codec_id to a specific .wasm file path.
+            Useful for testing.
+        engine: Shared Wasmtime engine for codec instantiation.
+            If not provided, a new engine with compilation caching
+            is created.
+    """
+
+    def __init__(
+        self,
+        *,
+        codec_store: Path | None = None,
+        preference: Sequence[str] | None = None,
+        overrides: dict[str, str] | None = None,
+        force_sources: dict[str, str] | None = None,
+        pipeline_sources: dict[str, str] | None = None,
+        paths: dict[str, Path] | None = None,
+        engine: wasmtime.Engine | None = None,
+    ) -> None:
+        self._store = CodecStore(codec_store or _default_store_path())
+        self._preference = list(
+            preference or (Backend.NATIVE, Backend.CORE, Backend.COMPONENT)
+        )
+        self._overrides = dict(overrides or {})
+        self._force_sources = dict(force_sources or {})
+        self._pipeline_sources = dict(pipeline_sources or {})
+        self._paths = dict(paths or {})
+        if engine is not None:
+            self._engine = engine
+        else:
+            config = wasmtime.Config()
+            config.cache = True
+            self._engine = wasmtime.Engine(config)
+
+    @property
+    def store_path(self) -> Path:
+        """Path to the local codec store directory."""
+        return self._store.path
+
+    def resolve(self, codec_id: str) -> Codec:
+        """Resolve a codec_id to a Codec instance.
+
+        Args:
+            codec_id: The logical codec identifier to resolve.
+
+        Returns:
+            A ready-to-use Codec instance.
+
+        Raises:
+            ValueError: If no implementation is found or the resolved
+                binary uses an unsupported backend.
+        """
+        # 0. Explicit paths (testing)
+        if codec_id in self._paths:
+            return self._instantiate(self._paths[codec_id])
+
+        # 1. Force sources (download from URI regardless of store)
+        if codec_id in self._force_sources:
+            return self._resolve_from_source(
+                self._force_sources[codec_id], force_install=True
+            )
+
+        # 2. Per-codec override
+        if codec_id in self._overrides:
+            return self._resolve_override(codec_id, self._overrides[codec_id])
+
+        # 3. Local store + native: collect candidates and pick by preference
+        entries = self._store.get_index().get(codec_id, {})
+        has_native = _has_native_signature(codec_id)
+        if entries or has_native:
+            return self._resolve_by_preference(codec_id, entries, has_native)
+
+        # 4. Pipeline sources
+        if codec_id in self._pipeline_sources:
+            return self._resolve_from_source(self._pipeline_sources[codec_id])
+
+        msg = f"No codec implementation found for {codec_id!r}"
+        raise ValueError(msg)
+
+    def list_codecs(self) -> list[CodecInfo]:
+        """List all codec implementations (local store and native)."""
+        result = list(self._store.list_entries())
+        result.extend(_list_native_codecs())
+        return result
+
+    def _resolve_by_preference(
+        self,
+        codec_id: str,
+        entries: dict[str, CodecInfo],
+        has_native: bool,
+    ) -> Codec:
+        """Select the best implementation for a codec_id using the preference list."""
+        for pref in self._preference:
+            if pref == Backend.NATIVE and has_native:
+                return NativeCodec(codec_id)
+            for entry in entries.values():
+                if entry.backend == pref:
+                    return self._instantiate(entry.location)
+        available = [e.backend for e in entries.values()]
+        if has_native:
+            available.append(Backend.NATIVE)
+        msg = (
+            f"No implementation for {codec_id!r} matches preference "
+            f"{self._preference}. Available backends: {available}"
+        )
+        raise ValueError(msg)
+
+    def _resolve_override(self, codec_id: str, impl_name: str) -> Codec:
+        """Resolve a codec_id using a specific implementation name override."""
+        entries = self._store.get_index().get(codec_id, {})
+        if impl_name in entries:
+            return self._instantiate(entries[impl_name].location)
+        # Check native
+        available = list(entries)
+        sig_path = SIGNATURES_DIR / f"{codec_id}.json"
+        if sig_path.exists():
+            sig = Signature.from_json(sig_path)
+            if sig.implementation == impl_name:
+                return NativeCodec(codec_id)
+            available.append(sig.implementation)
+        msg = (
+            f"Override implementation {impl_name!r} for codec {codec_id!r} "
+            f"not found. Available: {available}"
+        )
+        raise ValueError(msg)
+
+    def _instantiate(self, wasm_path: Path) -> Codec:
+        """Create a Codec from a wasm path."""
+        codec_type = detect_codec_type(wasm_path)
+        if codec_type == Backend.COMPONENT:
+            return ComponentCodec(self._engine, wasm_path)
+        return CoreWasmCodec(self._engine, wasm_path)
+
+    def _resolve_from_source(self, uri: str, *, force_install: bool = False) -> Codec:
+        from chonkle.wasm_download import resolve_uri
+
+        with resolve_uri(uri) as wasm_path:
+            entry = self._store.install(wasm_path, force_install=force_install)
+            return self._instantiate(entry.location)
+
+
+def _should_force_install() -> bool:
+    return os.environ.get("CHONKLE_FORCE_INSTALL", "") == "1"
+
+
+_UNSAFE_CHARS = re.compile(r'[/\\:*?"<>|\x00-\x1f]')
+
+
+def _sanitize_filename(name: str) -> str:
+    """Replace filesystem-unsafe characters in *name* with underscores."""
+    return _UNSAFE_CHARS.sub("_", name)
+
+
+def _has_native_signature(codec_id: str) -> bool:
+    """Check whether a native codec signature file exists for *codec_id*."""
+    return (SIGNATURES_DIR / f"{codec_id}.json").exists()
+
+
+def _list_native_codecs() -> list[CodecInfo]:
+    """Return CodecInfo for all native codecs with bundled signatures."""
+    if not SIGNATURES_DIR.exists():
+        return []
+    result: list[CodecInfo] = []
+    for sig_file in sorted(SIGNATURES_DIR.glob("*.json")):
+        sig = Signature.from_json(sig_file)
+        result.append(
+            CodecInfo(
+                codec_id=sig.codec_id,
+                implementation=sig.implementation,
+                backend=Backend.NATIVE,
+                location=sig_file,
+            )
+        )
+    return result
+
+
+def _scan_store(store_path: Path) -> dict[str, dict[str, CodecInfo]]:
+    """Scan the local codec store and return entries indexed by codec_id."""
+    index: dict[str, dict[str, CodecInfo]] = {}
+    if not store_path.exists():
+        return index
+    for codec_dir in sorted(store_path.iterdir()):
+        if not codec_dir.is_dir():
+            continue
+        for wasm_file in sorted(codec_dir.glob("*.wasm")):
+            try:
+                sig = Signature.from_wasm(wasm_file)
+                backend = detect_codec_type(wasm_file)
+            except (ValueError, OSError):
+                continue
+            entry = CodecInfo(
+                codec_id=sig.codec_id,
+                implementation=sig.implementation,
+                backend=backend,
+                location=wasm_file,
+            )
+            index.setdefault(entry.codec_id, {})[entry.implementation] = entry
+    return index
diff --git a/src/chonkle/signatures/numcodecs/adler32.json b/src/chonkle/signatures/numcodecs/adler32.json
new file mode 100644
index 0000000..31b9008
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/adler32.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "adler32",
+  "implementation": "numcodecs.adler32",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "location": {"type": "string", "required": false}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["location"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/astype.json b/src/chonkle/signatures/numcodecs/astype.json
new file mode 100644
index 0000000..a34e976
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/astype.json
@@ -0,0 +1,17 @@
+{
+  "codec_id": "astype",
+  "implementation": "numcodecs.astype",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "encode_dtype": {"type": "string", "required": true},
+    "decode_dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["encode_dtype", "decode_dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "decode_dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "encode_dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/base64.json b/src/chonkle/signatures/numcodecs/base64.json
new file mode 100644
index 0000000..15e413d
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/base64.json
@@ -0,0 +1,15 @@
+{
+  "codec_id": "base64",
+  "implementation": "numcodecs.base64",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": [],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/bitround.json b/src/chonkle/signatures/numcodecs/bitround.json
new file mode 100644
index 0000000..db15aac
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/bitround.json
@@ -0,0 +1,17 @@
+{
+  "codec_id": "bitround",
+  "implementation": "numcodecs.bitround",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "keepbits": {"type": "int", "required": true, "encode_only": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["keepbits"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/blosc.json b/src/chonkle/signatures/numcodecs/blosc.json
new file mode 100644
index 0000000..e62e779
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/blosc.json
@@ -0,0 +1,18 @@
+{
+  "codec_id": "blosc",
+  "implementation": "numcodecs.blosc",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "cname": {"type": "string", "required": false, "default": "lz4", "encode_only": true},
+    "clevel": {"type": "int", "required": false, "default": 5, "encode_only": true},
+    "shuffle": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["cname", "clevel", "shuffle"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/bz2.json b/src/chonkle/signatures/numcodecs/bz2.json
new file mode 100644
index 0000000..c0ef4f6
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/bz2.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "bz2",
+  "implementation": "numcodecs.bz2",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 9, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["level"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/categorize.json b/src/chonkle/signatures/numcodecs/categorize.json
new file mode 100644
index 0000000..cd3ebd2
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/categorize.json
@@ -0,0 +1,18 @@
+{
+  "codec_id": "categorize",
+  "implementation": "numcodecs.categorize",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "labels": {"type": "list", "required": true},
+    "dtype": {"type": "string", "required": true},
+    "astype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["labels", "dtype", "astype"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "astype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/crc32.json b/src/chonkle/signatures/numcodecs/crc32.json
new file mode 100644
index 0000000..4b96ae9
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/crc32.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "crc32",
+  "implementation": "numcodecs.crc32",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "location": {"type": "string", "required": false}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["location"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/crc32c.json b/src/chonkle/signatures/numcodecs/crc32c.json
new file mode 100644
index 0000000..220b837
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/crc32c.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "crc32c",
+  "implementation": "numcodecs.crc32c",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "location": {"type": "string", "required": false}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["location"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/delta.json b/src/chonkle/signatures/numcodecs/delta.json
new file mode 100644
index 0000000..3001bfe
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/delta.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "delta",
+  "implementation": "numcodecs.delta",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/fixedscaleoffset.json b/src/chonkle/signatures/numcodecs/fixedscaleoffset.json
new file mode 100644
index 0000000..00c7942
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/fixedscaleoffset.json
@@ -0,0 +1,18 @@
+{
+  "codec_id": "fixedscaleoffset",
+  "implementation": "numcodecs.fixedscaleoffset",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "offset": {"type": "number", "required": true},
+    "scale": {"type": "number", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["offset", "scale", "dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/fletcher32.json b/src/chonkle/signatures/numcodecs/fletcher32.json
new file mode 100644
index 0000000..9263f0a
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/fletcher32.json
@@ -0,0 +1,15 @@
+{
+  "codec_id": "fletcher32",
+  "implementation": "numcodecs.fletcher32",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": [],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/gzip.json b/src/chonkle/signatures/numcodecs/gzip.json
new file mode 100644
index 0000000..afb5863
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/gzip.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "gzip",
+  "implementation": "numcodecs.gzip",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["level"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/jenkins_lookup3.json b/src/chonkle/signatures/numcodecs/jenkins_lookup3.json
new file mode 100644
index 0000000..9e48a8f
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/jenkins_lookup3.json
@@ -0,0 +1,17 @@
+{
+  "codec_id": "jenkins_lookup3",
+  "implementation": "numcodecs.jenkins_lookup3",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "initval": {"type": "int", "required": false, "default": 0},
+    "prefix": {"type": "string", "required": false}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["initval", "prefix"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/json2.json b/src/chonkle/signatures/numcodecs/json2.json
new file mode 100644
index 0000000..d426c5d
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/json2.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "json2",
+  "implementation": "numcodecs.json2",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": [],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/lz4.json b/src/chonkle/signatures/numcodecs/lz4.json
new file mode 100644
index 0000000..78f678e
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/lz4.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "lz4",
+  "implementation": "numcodecs.lz4",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "acceleration": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["acceleration"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/lzma.json b/src/chonkle/signatures/numcodecs/lzma.json
new file mode 100644
index 0000000..d893f00
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/lzma.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "lzma",
+  "implementation": "numcodecs.lzma",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "preset": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["preset"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/packbits.json b/src/chonkle/signatures/numcodecs/packbits.json
new file mode 100644
index 0000000..3804861
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/packbits.json
@@ -0,0 +1,17 @@
+{
+  "codec_id": "packbits",
+  "implementation": "numcodecs.packbits",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "encode_dtype": {"type": "string", "required": false, "default": "bool", "encode_only": true},
+    "decode_dtype": {"type": "string", "required": false, "default": "uint8", "decode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": [],
+    "encode": {"mode": "ndarray", "dtype_port": "encode_dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "decode_dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/pickle.json b/src/chonkle/signatures/numcodecs/pickle.json
new file mode 100644
index 0000000..ebf73c0
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/pickle.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "pickle",
+  "implementation": "numcodecs.pickle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "protocol": {"type": "int", "required": false, "default": 5, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["protocol"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/quantize.json b/src/chonkle/signatures/numcodecs/quantize.json
new file mode 100644
index 0000000..fc12942
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/quantize.json
@@ -0,0 +1,17 @@
+{
+  "codec_id": "quantize",
+  "implementation": "numcodecs.quantize",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "digits": {"type": "int", "required": true, "encode_only": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["digits", "dtype"],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/shuffle.json b/src/chonkle/signatures/numcodecs/shuffle.json
new file mode 100644
index 0000000..c70c641
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/shuffle.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "shuffle",
+  "implementation": "numcodecs.shuffle",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "dtype": {"type": "string", "required": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": [],
+    "encode": {"mode": "ndarray", "dtype_port": "dtype"},
+    "decode": {"mode": "ndarray", "dtype_port": "dtype"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/zlib.json b/src/chonkle/signatures/numcodecs/zlib.json
new file mode 100644
index 0000000..822792a
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/zlib.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "zlib",
+  "implementation": "numcodecs.zlib",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["level"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/signatures/numcodecs/zstd.json b/src/chonkle/signatures/numcodecs/zstd.json
new file mode 100644
index 0000000..f1c554e
--- /dev/null
+++ b/src/chonkle/signatures/numcodecs/zstd.json
@@ -0,0 +1,16 @@
+{
+  "codec_id": "zstd",
+  "implementation": "numcodecs.zstd",
+  "inputs": {
+    "bytes": {"type": "bytes", "required": true},
+    "level": {"type": "int", "required": false, "default": 1, "encode_only": true}
+  },
+  "outputs": {
+    "bytes": {"type": "bytes"}
+  },
+  "native": {
+    "constructor_ports": ["level"],
+    "encode": {"mode": "bytes"},
+    "decode": {"mode": "bytes"}
+  }
+}
diff --git a/src/chonkle/wasm_download.py b/src/chonkle/wasm_download.py
index 886d886..1f433d8 100644
--- a/src/chonkle/wasm_download.py
+++ b/src/chonkle/wasm_download.py
@@ -1,58 +1,68 @@
-"""Download Wasm codec modules from HTTPS URLs and OCI registries."""
+"""Resolve codec URIs to local paths; remote downloads use a temporary directory."""
 
-import hashlib
 import logging
 import os
 import shutil
 import tempfile
 import urllib.parse
 import urllib.request
+from collections.abc import Iterator
+from contextlib import contextmanager
 from pathlib import Path, PurePosixPath
 
 import oras.client
 
 
-def get_cache_dir() -> Path:
-    """Return the directory for cached .wasm downloads.
+@contextmanager
+def resolve_uri(uri: str) -> Iterator[Path]:
+    """Resolve a codec URI, yielding an absolute local file path.
 
-    Uses CHONKLE_CACHE_DIR if set, otherwise falls back to a
-    chonkle/wasm subdirectory inside the OS temporary directory.
-    """
-    override = os.environ.get("CHONKLE_CACHE_DIR", "")
-    if override:
-        base = Path(override)
-    else:
-        base = Path(tempfile.gettempdir()) / "chonkle" / "wasm"
-    base.mkdir(parents=True, exist_ok=True)
-    return base
+    Dispatches based on URI scheme:
 
+    - file:// — yields the path directly; no temp dir created
+    - https:// — downloads to a temp dir, yields the path, cleans up on exit
+    - oci:// — pulls to a temp dir, yields the path, cleans up on exit
+    - http:// — rejected; use HTTPS
 
-def _should_force() -> bool:
-    """Check whether CHONKLE_FORCE_DOWNLOAD is set."""
-    return os.environ.get("CHONKLE_FORCE_DOWNLOAD", "") == "1"
+    The caller is responsible for installing the yielded file into the codec
+    store before the context exits if durable storage is needed.
 
+    Args:
+        uri: The codec URI to resolve.
 
-def download_https(
-    url: str, *, cache_dir: Path | None = None, force: bool = False
-) -> Path:
-    """Download a .wasm file from an HTTPS URL, returning the cached path.
+    Yields:
+        The absolute path to a local .wasm file.
 
-    The file is cached by a SHA-256 hash of the URL. Pass 'force' (or set
-    CHONKLE_FORCE_DOWNLOAD=1) to re-download even when a cached copy
-    exists.
+    Raises:
+        ValueError: If the scheme is http://, unsupported, or absent.
     """
-    if cache_dir is None:
-        cache_dir = get_cache_dir()
+    parsed = urllib.parse.urlparse(uri)
 
-    url_hash = hashlib.sha256(url.encode()).hexdigest()
-    filename = PurePosixPath(urllib.parse.urlparse(url).path).name or "module.wasm"
-    dest_dir = cache_dir / "https" / url_hash
-    dest = dest_dir / filename
+    if parsed.scheme == "file":
+        yield Path(parsed.path)
+        return
 
-    if dest.exists() and not force and not _should_force():
-        return dest
+    if parsed.scheme == "http":
+        msg = "HTTP is not supported for Wasm downloads; use HTTPS instead"
+        raise ValueError(msg)
 
-    dest_dir.mkdir(parents=True, exist_ok=True)
+    with tempfile.TemporaryDirectory() as tmp:
+        tmp_dir = Path(tmp)
+        if parsed.scheme == "https":
+            yield _download_https(uri, tmp_dir)
+        elif parsed.scheme == "oci":
+            yield _download_oci(uri, tmp_dir)
+        else:
+            msg = (
+                f"Unsupported URI scheme: {parsed.scheme!r}"
+                if parsed.scheme
+                else f"URI must include a scheme (file://, https://, oci://): {uri!r}"
+            )
+            raise ValueError(msg)
+
+
+def _download_url_to(url: str, dest: Path, dest_dir: Path) -> None:
+    """Download a URL to dest atomically via a temp file."""
     request = urllib.request.Request(url)  # noqa: S310
     tmp_fd, tmp_path_str = tempfile.mkstemp(dir=dest_dir)
     tmp = Path(tmp_path_str)
@@ -64,33 +74,18 @@ def download_https(
         tmp.unlink()
         raise
 
-    return dest
-
-
-def download_oci(
-    uri: str, *, cache_dir: Path | None = None, force: bool = False
-) -> Path:
-    """Download a .wasm file from an OCI registry, returning the cached path.
 
-    'uri' should include the oci:// scheme prefix, e.g.
-    oci://ghcr.io/cylf-dev/tiff-predictor-2-c:v0.1.0.
+def _download_https(url: str, dest_dir: Path) -> Path:
+    """Download a .wasm file from HTTPS into dest_dir."""
+    filename = PurePosixPath(urllib.parse.urlparse(url).path).name or "module.wasm"
+    dest = dest_dir / filename
+    _download_url_to(url, dest, dest_dir)
+    return dest
 
-    The file is cached by the OCI reference. Pass 'force' (or set
-    CHONKLE_FORCE_DOWNLOAD=1) to re-download even when a cached copy
-    exists.
-    """
-    if cache_dir is None:
-        cache_dir = get_cache_dir()
 
+def _download_oci(uri: str, dest_dir: Path) -> Path:
+    """Pull a .wasm codec from an OCI registry into dest_dir."""
     ref = uri.removeprefix("oci://")
-    ref_dir = cache_dir / "oci" / ref.replace(":", "/")
-
-    if not force and not _should_force() and ref_dir.exists():
-        wasm_files = list(ref_dir.glob("*.wasm"))
-        if wasm_files:
-            return wasm_files[0]
-
-    ref_dir.mkdir(parents=True, exist_ok=True)
     client = oras.client.OrasClient()
 
     # The oras library reads ~/.docker/config.json and tries credential
@@ -101,7 +96,7 @@ def download_oci(
     prev_level = oras_logger.level
     oras_logger.setLevel(logging.ERROR)
     try:
-        files = client.pull(target=ref, outdir=str(ref_dir))
+        files = client.pull(target=ref, outdir=str(dest_dir))
     finally:
         oras_logger.setLevel(prev_level)
 
diff --git a/src/chonkle/wasm_runner.py b/src/chonkle/wasm_runner.py
deleted file mode 100644
index c9c0b9c..0000000
--- a/src/chonkle/wasm_runner.py
+++ /dev/null
@@ -1,250 +0,0 @@
-"""Load and run Wasm codec modules via the generic codec ABI."""
-
-import json
-from pathlib import Path
-from typing import cast
-from urllib.parse import urlparse
-
-import wasmtime
-import wasmtime.component
-
-from chonkle.wasm_download import download_https, download_oci
-
-# Wasm binary format: 4-byte magic followed by 4-byte version field.
-_WASM_MAGIC = b"\x00asm"
-_CORE_VERSION = b"\x01\x00\x00\x00"
-_COMPONENT_VERSION = b"\x0d\x00\x01\x00"
-
-
-def resolve_wasm_uri(uri: str, *, force_download: bool = False) -> Path:
-    """Resolve a Wasm URI to an absolute file path.
-
-    Supported URI forms:
-
-    - file:///path/to/codec.wasm (local file)
-    - https://… (HTTPS download, cached)
-    - oci://ghcr.io/org/repo:tag (OCI registry pull, cached)
-    """
-    parsed = urlparse(uri)
-
-    if parsed.scheme == "file":
-        return Path(parsed.path)
-
-    if parsed.scheme == "http":
-        msg = "HTTP is not supported for Wasm downloads; use HTTPS instead"
-        raise ValueError(msg)
-
-    if parsed.scheme == "https":
-        return download_https(uri, force=force_download)
-
-    if parsed.scheme == "oci":
-        return download_oci(uri, force=force_download)
-
-    msg = (
-        f"Unsupported URI scheme: {parsed.scheme!r}"
-        if parsed.scheme
-        else f"URI must include a scheme (e.g. file://, https://, oci://): {uri!r}"
-    )
-    raise NotImplementedError(msg)
-
-
-def _is_wasm_component(path: Path) -> bool:
-    """Return True if the file is a Wasm Component, False if a Core module."""
-    with path.open("rb") as f:
-        header = f.read(8)
-    if len(header) < 8 or header[:4] != _WASM_MAGIC:
-        msg = f"Not a valid Wasm binary: {path}"
-        raise ValueError(msg)
-    version = header[4:]
-    if version == _CORE_VERSION:
-        return False
-    if version == _COMPONENT_VERSION:
-        return True
-    msg = f"Unknown Wasm format (version bytes {version.hex()!r}): {path}"
-    raise ValueError(msg)
-
-
-def _wasm_core_call(
-    wasm_path: Path,
-    export_name: str,
-    data: bytes,
-    configuration: dict,
-) -> bytes:
-    """Call a Core Wasm codec export and return the result bytes.
-
-    The module must export the generic Wasm codec ABI:
-
-    - alloc(size: i32) -> i32
-    - dealloc(ptr: i32, size: i32)
-    - <export_name>(input_ptr, input_len, config_ptr, config_len) -> i64
-    - memory
-
-    Both freestanding (no imports) and WASI modules are supported.
-    """
-    resolved = wasm_path.resolve()
-    config = wasmtime.Config()
-    config.cache = True
-    engine = wasmtime.Engine(config)
-    module = wasmtime.Module.from_file(engine, str(resolved))
-    store = wasmtime.Store(engine)
-
-    if module.imports:
-        wasi_config = wasmtime.WasiConfig()
-        store.set_wasi(wasi_config)
-        linker = wasmtime.Linker(engine)
-        linker.define_wasi()
-        instance = linker.instantiate(store, module)
-    else:
-        instance = wasmtime.Instance(store, module, [])
-
-    # Get exported functions and memory.
-    exports = instance.exports(store)
-
-    # Validate required ABI exports before accessing them.
-    required = ["alloc", "dealloc", "memory", export_name]
-    missing = [name for name in required if exports.get(name) is None]
-    if missing:
-        msg = (
-            f"Wasm module is missing required ABI export(s): {', '.join(missing)}. "
-            f"Module at {resolved}"
-        )
-        raise RuntimeError(msg)
-
-    # WASI reactors export _initialize for runtime setup.
-    initialize_fn = exports.get("_initialize")
-    if initialize_fn is not None:
-        cast(wasmtime.Func, initialize_fn)(store)
-    alloc_fn = cast(wasmtime.Func, exports["alloc"])
-    dealloc_fn = cast(wasmtime.Func, exports["dealloc"])
-    codec_fn = cast(wasmtime.Func, exports[export_name])
-    memory = cast(wasmtime.Memory, exports["memory"])
-
-    # Write input data into Wasm linear memory.
-    input_len = len(data)
-    input_ptr = alloc_fn(store, input_len)
-    memory.write(store, data, input_ptr)
-
-    # Write configuration JSON into Wasm linear memory.
-    config_bytes = json.dumps(configuration).encode()
-    config_len = len(config_bytes)
-    config_ptr = alloc_fn(store, config_len)
-    memory.write(store, config_bytes, config_ptr)
-
-    # Call the codec function.
-    result_i64 = codec_fn(store, input_ptr, input_len, config_ptr, config_len)
-
-    # Unpack (output_ptr << 32) | output_len.
-    out_ptr = (result_i64 >> 32) & 0xFFFFFFFF
-    out_len = result_i64 & 0xFFFFFFFF
-
-    if out_ptr == 0 and out_len == 0:
-        msg = f"Wasm {export_name} returned null (bad configuration?)"
-        raise RuntimeError(msg)
-
-    # Read output from Wasm memory.
-    output = memory.read(store, out_ptr, out_ptr + out_len)
-
-    # Free all Wasm-side allocations.
-    dealloc_fn(store, input_ptr, input_len)
-    dealloc_fn(store, config_ptr, config_len)
-    dealloc_fn(store, out_ptr, out_len)
-
-    return bytes(output)
-
-
-def _wasm_component_call(
-    wasm_path: Path,
-    export_name: str,
-    data: bytes,
-    configuration: dict,
-) -> bytes:
-    """Call a Wasm Component Model codec export and return the result bytes.
-
-    The component must export a function named 'encode' or 'decode' (either
-    directly at the world level or within an interface) with signature:
-
-        func(data: list<u8>, config: string) -> result<list<u8>, string>
-
-    componentize-py components built against WASIp2 are supported.
-    """
-    resolved = wasm_path.resolve()
-    config = wasmtime.Config()
-    config.cache = True
-    engine = wasmtime.Engine(config)
-
-    component = wasmtime.component.Component.from_file(engine, str(resolved))
-
-    store = wasmtime.Store(engine)
-    # WASIp2 components require both set_wasi and add_wasip2.
-    store.set_wasi(wasmtime.WasiConfig())
-    linker = wasmtime.component.Linker(engine)
-    linker.add_wasip2()
-
-    instance = linker.instantiate(store, component)
-
-    # Discover the export: check world-level functions first, then interfaces.
-    comp_exports = component.type.exports(engine)
-    fn_idx = None
-
-    for iface_name, item in comp_exports.items():
-        if isinstance(item, wasmtime.component.FuncType) and iface_name == export_name:
-            fn_idx = instance.get_export_index(store, iface_name)
-            break
-
-    if fn_idx is None:
-        for iface_name, item in comp_exports.items():
-            if isinstance(item, wasmtime.component.ComponentInstanceType):
-                iface_exports = item.exports(engine)
-                if export_name in iface_exports and isinstance(
-                    iface_exports[export_name], wasmtime.component.FuncType
-                ):
-                    iface_idx = instance.get_export_index(store, iface_name)
-                    fn_idx = instance.get_export_index(store, export_name, iface_idx)
-                    break
-
-    if fn_idx is None:
-        msg = (
-            f"Wasm component is missing required export '{export_name}'. "
-            f"Component at {resolved}"
-        )
-        raise RuntimeError(msg)
-
-    fn = instance.get_func(store, fn_idx)
-    if fn is None:
-        msg = (
-            f"Wasm component export '{export_name}' is not a function. "
-            f"Component at {resolved}"
-        )
-        raise RuntimeError(msg)
-    config_str = json.dumps(configuration)
-    result = fn(store, data, config_str)
-    fn.post_return(store)
-
-    # result<list<u8>, string> returns bytes on Ok or str on Err.
-    if isinstance(result, str):
-        msg = f"Wasm component {export_name} returned error: {result}"
-        raise RuntimeError(msg)
-
-    return result
-
-
-def _wasm_call(
-    wasm_path: Path,
-    export_name: str,
-    data: bytes,
-    configuration: dict,
-) -> bytes:
-    """Dispatch to the Core or Component call path based on the binary header."""
-    if _is_wasm_component(wasm_path):
-        return _wasm_component_call(wasm_path, export_name, data, configuration)
-    return _wasm_core_call(wasm_path, export_name, data, configuration)
-
-
-def wasm_decode(wasm_path: Path, data: bytes, configuration: dict) -> bytes:
-    """Decode 'data' using the Wasm codec module at 'wasm_path'."""
-    return _wasm_call(wasm_path, "decode", data, configuration)
-
-
-def wasm_encode(wasm_path: Path, data: bytes, configuration: dict) -> bytes:
-    """Encode 'data' using the Wasm codec module at 'wasm_path'."""
-    return _wasm_call(wasm_path, "encode", data, configuration)
diff --git a/src/chonkle/wasm_signature.py b/src/chonkle/wasm_signature.py
new file mode 100644
index 0000000..f2ca8fc
--- /dev/null
+++ b/src/chonkle/wasm_signature.py
@@ -0,0 +1,276 @@
+"""Codec signatures: types, detection, and Wasm custom-section I/O.
+
+Pure-Python implementation — no wasmtime or instantiation needed. Works with
+both core Wasm (version 01 00 00 00) and Component Model (version
+0d 00 01 00) binaries.
+"""
+
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any
+
+from chonkle.codecs._base import Backend
+
+SECTION_NAME = "chonkle:signature"
+
+# Wasm binary header constants.
+_WASM_MAGIC = b"\x00asm"
+_HEADER_SIZE = 8  # magic (4) + version (4)
+_CORE_WASM_VERSION = b"\x01\x00\x00\x00"
+_COMPONENT_VERSION = b"\x0d\x00\x01\x00"
+
+
+@dataclass(frozen=True)
+class PortDescriptor:
+    """Descriptor for a single input or output port in a codec signature."""
+
+    type: str
+    required: bool = True
+    default: Any = None
+    encode_only: bool = False
+    decode_only: bool = False
+
+
+@dataclass(frozen=True)
+class Signature:
+    """Typed representation of a codec's input and output port declarations."""
+
+    codec_id: str
+    implementation: str
+    inputs: dict[str, PortDescriptor] = field(default_factory=dict)
+    outputs: dict[str, PortDescriptor] = field(default_factory=dict)
+
+    @classmethod
+    def from_wasm(cls, wasm_path: str | Path) -> Signature:
+        """Construct a Signature from a .wasm file's embedded custom section."""
+        return cls.from_dict(read_signature(wasm_path))
+
+    @classmethod
+    def from_json(cls, json_path: str | Path) -> Signature:
+        """Construct a Signature from a JSON signature file."""
+        return cls.from_dict(json.loads(Path(json_path).read_text()))
+
+    @classmethod
+    def from_dict(cls, d: dict[str, Any]) -> Signature:
+        """Construct a Signature from a raw JSON-derived dict."""
+        inputs: dict[str, PortDescriptor] = {}
+        for name, desc in d.get("inputs", {}).items():
+            inputs[name] = PortDescriptor(
+                type=desc.get("type", ""),
+                required=desc.get("required", True),
+                default=desc.get("default"),
+                encode_only=desc.get("encode_only", False),
+                decode_only=desc.get("decode_only", False),
+            )
+        outputs: dict[str, PortDescriptor] = {}
+        for name, desc in d.get("outputs", {}).items():
+            outputs[name] = PortDescriptor(
+                type=desc.get("type", ""),
+                required=desc.get("required", True),
+                default=desc.get("default"),
+            )
+        codec_id = d.get("codec_id")
+        if not codec_id:
+            msg = "Signature missing required field 'codec_id'"
+            raise ValueError(msg)
+        return cls(
+            codec_id=codec_id,
+            implementation=d.get("implementation", ""),
+            inputs=inputs,
+            outputs=outputs,
+        )
+
+    def encode_only_inputs(self) -> set[str]:
+        """Port names where encode_only is True."""
+        return {n for n, p in self.inputs.items() if p.encode_only}
+
+    def decode_only_inputs(self) -> set[str]:
+        """Port names where decode_only is True."""
+        return {n for n, p in self.inputs.items() if p.decode_only}
+
+    def output_types(self) -> dict[str, str]:
+        """Map of output port name to type string."""
+        return {n: p.type for n, p in self.outputs.items()}
+
+
+def _read_leb128(data: bytes, offset: int) -> tuple[int, int]:
+    """Decode an unsigned LEB128 value, returning (value, new_offset)."""
+    result = 0
+    shift = 0
+    while True:
+        byte = data[offset]
+        offset += 1
+        result |= (byte & 0x7F) << shift
+        if (byte & 0x80) == 0:
+            break
+        shift += 7
+    return result, offset
+
+
+def _encode_leb128(value: int) -> bytes:
+    """Encode *value* as unsigned LEB128."""
+    buf = bytearray()
+    while True:
+        byte = value & 0x7F
+        value >>= 7
+        if value != 0:
+            byte |= 0x80
+        buf.append(byte)
+        if value == 0:
+            break
+    return bytes(buf)
+
+
+def read_signature(wasm_path: str | Path) -> dict[str, Any]:
+    """Read the chonkle:signature custom section from a .wasm file.
+
+    Args:
+        wasm_path: Path to the .wasm binary.
+
+    Returns:
+        The parsed signature dict.
+
+    Raises:
+        FileNotFoundError: The .wasm file does not exist.
+        ValueError: The file is not a valid Wasm binary or contains no
+            chonkle:signature section.
+    """
+    data = Path(wasm_path).read_bytes()
+    return read_signature_bytes(data, context=str(wasm_path))
+
+
+def read_signature_bytes(data: bytes, *, context: str = "<bytes>") -> dict[str, Any]:
+    """Read the chonkle:signature custom section from in-memory Wasm bytes.
+
+    Args:
+        data: Raw Wasm binary content.
+        context: Label used in error messages (typically a file path).
+
+    Returns:
+        The parsed signature dict.
+
+    Raises:
+        ValueError: Not a valid Wasm binary or no chonkle:signature section.
+    """
+    if len(data) < _HEADER_SIZE or data[:4] != _WASM_MAGIC:
+        msg = f"Not a valid Wasm binary: {context}"
+        raise ValueError(msg)
+
+    offset = _HEADER_SIZE
+
+    while offset < len(data):
+        section_id = data[offset]
+        offset += 1
+        section_size, offset = _read_leb128(data, offset)
+        section_end = offset + section_size
+
+        if section_id == 0:  # custom section
+            name_len, name_start = _read_leb128(data, offset)
+            name = data[name_start : name_start + name_len].decode("utf-8")
+            payload_start = name_start + name_len
+
+            if name == SECTION_NAME:
+                payload = data[payload_start:section_end]
+                return json.loads(payload)
+
+        offset = section_end
+
+    msg = f"No {SECTION_NAME!r} custom section in {context}"
+    raise ValueError(msg)
+
+
+def embed_signature(wasm_bytes: bytes, signature: dict[str, Any]) -> bytes:
+    """Append a chonkle:signature custom section to Wasm bytes.
+
+    If a chonkle:signature section already exists, it is replaced.
+
+    Args:
+        wasm_bytes: Raw Wasm binary content.
+        signature: The signature dict to embed as JSON.
+
+    Returns:
+        New Wasm bytes with the custom section appended.
+
+    Raises:
+        ValueError: wasm_bytes is not a valid Wasm binary.
+    """
+    if len(wasm_bytes) < _HEADER_SIZE or wasm_bytes[:4] != _WASM_MAGIC:
+        msg = "Not a valid Wasm binary"
+        raise ValueError(msg)
+
+    stripped = _strip_section(wasm_bytes, SECTION_NAME)
+    payload = json.dumps(signature, separators=(",", ":")).encode("utf-8")
+    section = _build_custom_section(SECTION_NAME, payload)
+    return stripped + section
+
+
+def _strip_section(data: bytes, section_name: str) -> bytes:
+    """Return *data* with any custom section named *section_name* removed."""
+    result = bytearray(data[:_HEADER_SIZE])
+    offset = _HEADER_SIZE
+
+    while offset < len(data):
+        section_start = offset
+        section_id = data[offset]
+        offset += 1
+        section_size, offset = _read_leb128(data, offset)
+        section_end = offset + section_size
+
+        keep = True
+        if section_id == 0:
+            name_len, name_start = _read_leb128(data, offset)
+            name = data[name_start : name_start + name_len].decode("utf-8")
+            if name == section_name:
+                keep = False
+
+        if keep:
+            result.extend(data[section_start:section_end])
+
+        offset = section_end
+
+    return bytes(result)
+
+
+def _build_custom_section(name: str, payload: bytes) -> bytes:
+    """Build a Wasm custom section (id=0) with *name* and *payload*."""
+    name_bytes = name.encode("utf-8")
+    name_field = _encode_leb128(len(name_bytes)) + name_bytes
+    section_body = name_field + payload
+    return b"\x00" + _encode_leb128(len(section_body)) + section_body
+
+
+def detect_codec_type(wasm_path: Path) -> Backend:
+    """Detect whether a .wasm file is core or component model.
+
+    Reads the 8-byte header (magic + version) to distinguish:
+
+    - 0d 00 01 00 → "component" (Component Model)
+    - 01 00 00 00 → "core" (Core Wasm)
+
+    Args:
+        wasm_path: Path to the .wasm binary.
+
+    Returns:
+        "component" or "core".
+
+    Raises:
+        ValueError: Not a valid Wasm binary or unrecognized version.
+    """
+    with Path(wasm_path).open("rb") as f:
+        header = f.read(8)
+
+    if len(header) < 8 or header[:4] != _WASM_MAGIC:
+        msg = f"Not a valid Wasm binary: {wasm_path}"
+        raise ValueError(msg)
+
+    version = header[4:8]
+    if version == _COMPONENT_VERSION:
+        return Backend.COMPONENT
+    if version == _CORE_WASM_VERSION:
+        return Backend.CORE
+
+    msg = f"Unknown Wasm version {version.hex()}: {wasm_path}"
+    raise ValueError(msg)
diff --git a/tests/conftest.py b/tests/conftest.py
index bb691eb..3c6d7b7 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1,51 +1,18 @@
-import shutil
+"""pytest configuration and shared fixtures for chonkle tests."""
+
+import json
 from pathlib import Path
 
 import pytest
 
-from chonkle.wasm_download import download_https
-
-FIXTURES_DIR = Path(__file__).parent / "fixtures" / "chunks"
-
-_COG_WASM_DIR = FIXTURES_DIR / "cog_wasm"
-
-# Registry of .wasm fixtures not committed to git.
-# To add a new fixture: add an entry here, add a named fixture below,
-# and add a CI cache step in .github/workflows/ci.yml keyed on the version tag.
-_WASM_DOWNLOADS: dict[str, str] = {
-    "tiff-predictor-2-c.wasm": (
-        "https://github.com/cylf-dev/tiff-predictor-2-c"
-        "/releases/download/v0.2.0/tiff-predictor-2-c.wasm"
-    ),
-    "tiff-predictor-2-python.wasm": (
-        "https://github.com/cylf-dev/tiff-predictor-2-python"
-        "/releases/download/v0.1.0/tiff-predictor-2-python.wasm"
-    ),
-}
-
-
-def _ensure_wasm_fixture(filename: str) -> Path:
-    dest = _COG_WASM_DIR / filename
-    if not dest.exists():
-        shutil.copy2(download_https(_WASM_DOWNLOADS[filename]), dest)
-    return dest
-
+from chonkle.resolver import Resolver
 
-@pytest.fixture(scope="session")
-def ensure_core_wasm() -> Path:
-    return _ensure_wasm_fixture("tiff-predictor-2-c.wasm")
+FIXTURES_DIR = Path(__file__).parent / "fixtures"
+PIPELINES_DIR = FIXTURES_DIR / "pipelines"
+CHUNKS_DIR = FIXTURES_DIR / "chunks"
 
-
-@pytest.fixture(scope="session")
-def ensure_component_wasm() -> Path:
-    return _ensure_wasm_fixture("tiff-predictor-2-python.wasm")
-
-
-CHUNK_PATHS = [
-    FIXTURES_DIR / "cog" / "0",
-    FIXTURES_DIR / "zarr_zlib" / "0",
-    FIXTURES_DIR / "zarr_zstd" / "0",
-]
+REPO_ROOT = Path(__file__).parent.parent
+CODEC_DIR = REPO_ROOT / "codec"
 
 
 def pytest_addoption(parser: pytest.Parser) -> None:
@@ -66,3 +33,68 @@ def pytest_collection_modifyitems(
     for item in items:
         if "network" in item.keywords:
             item.add_marker(skip_network)
+
+
+@pytest.fixture
+def raw_chunk() -> bytes:
+    """Raw bytes of known content suitable for codec round-trip testing.
+
+    Patterned bytes compress and decompress deterministically.
+    """
+    return bytes(range(256)) * 64
+
+
+@pytest.fixture
+def page_split_input() -> tuple[bytes, int, int]:
+    """Input bytes and split offsets for page-split testing.
+
+    Returns (data, rep_length, def_length) such that rep_length + def_length
+    is less than len(data), producing three non-empty segments.
+    """
+    data = bytes(range(256)) * 16  # 4096 bytes
+    rep_length = 128
+    def_length = 256
+    return data, rep_length, def_length
+
+
+@pytest.fixture
+def cog_chunk() -> bytes:
+    """A real COG tile compressed with zlib (level 9) + tiff-predictor-2.
+
+    1024x1024 uint16 Sentinel-2 band tile. Decodes to 2,097,152 bytes.
+    """
+    return (CHUNKS_DIR / "cog-chunk-0").read_bytes()
+
+
+@pytest.fixture
+def cog_decode_pipeline_json() -> dict:
+    """Pipeline dict for the COG zlib+tiff-predictor-2 decode chain."""
+    with (PIPELINES_DIR / "cog-decode-pipeline.json").open() as f:
+        return json.load(f)
+
+
+@pytest.fixture
+def cog_encode_pipeline_json() -> dict:
+    """Pipeline dict for the COG tiff-predictor-2+zlib encode chain."""
+    with (PIPELINES_DIR / "cog-encode-pipeline.json").open() as f:
+        return json.load(f)
+
+
+@pytest.fixture
+def cog_codec_resolver() -> Resolver:
+    """Resolver with explicit paths for the COG codec wasm files."""
+    return Resolver(
+        paths={
+            "zlib": CODEC_DIR / "zlib-rs" / "zlib.wasm",
+            "tiff-predictor-2": (
+                CODEC_DIR / "tiff-predictor-2-c" / "tiff-predictor-2.wasm"
+            ),
+        }
+    )
+
+
+@pytest.fixture
+def page_split_pipeline_json() -> dict:
+    """Parsed pipeline JSON for the page-split-dag fixture."""
+    with (PIPELINES_DIR / "page-split-dag.json").open() as f:
+        return json.load(f)
diff --git a/tests/fixtures/chunks/cog/0 b/tests/fixtures/chunks/cog-chunk-0
similarity index 100%
rename from tests/fixtures/chunks/cog/0
rename to tests/fixtures/chunks/cog-chunk-0
diff --git a/tests/fixtures/chunks/cog/0.json b/tests/fixtures/chunks/cog/0.json
deleted file mode 100644
index 497acd6..0000000
--- a/tests/fixtures/chunks/cog/0.json
+++ /dev/null
@@ -1,28 +0,0 @@
-{
-  "codecs": [
-    {
-      "name": "tiff_predictor_2",
-      "type": "numcodecs",
-      "configuration": {}
-    },
-    {
-      "name": "bytes",
-      "type": "numcodecs",
-      "configuration": {
-        "endian": "little",
-        "data_type": "uint16",
-        "shape": [
-          1024,
-          1024
-        ]
-      }
-    },
-    {
-      "name": "zlib",
-      "type": "numcodecs",
-      "configuration": {
-        "level": 9
-      }
-    }
-  ]
-}
diff --git a/tests/fixtures/chunks/cog_wasm/0 b/tests/fixtures/chunks/cog_wasm/0
deleted file mode 100644
index 6b724d1..0000000
Binary files a/tests/fixtures/chunks/cog_wasm/0 and /dev/null differ
diff --git a/tests/fixtures/chunks/cog_wasm/0.json b/tests/fixtures/chunks/cog_wasm/0.json
deleted file mode 100644
index 7baabd7..0000000
--- a/tests/fixtures/chunks/cog_wasm/0.json
+++ /dev/null
@@ -1,32 +0,0 @@
-{
-  "codecs": [
-    {
-      "name": "bytes",
-      "type": "numcodecs",
-      "configuration": {
-        "endian": "little",
-        "data_type": "uint16",
-        "shape": [
-          1024,
-          1024
-        ]
-      }
-    },
-    {
-      "name": "tiff_predictor_2",
-      "type": "wasm",
-      "uri": "file:///absolute/path/to/tiff-predictor-2-c.wasm",
-      "configuration": {
-        "bytes_per_sample": 2,
-        "width": 1024
-      }
-    },
-    {
-      "name": "zlib",
-      "type": "numcodecs",
-      "configuration": {
-        "level": 9
-      }
-    }
-  ]
-}
diff --git a/tests/fixtures/chunks/zarr_zlib/0 b/tests/fixtures/chunks/zarr_zlib/0
deleted file mode 100644
index ff6d3c9..0000000
Binary files a/tests/fixtures/chunks/zarr_zlib/0 and /dev/null differ
diff --git a/tests/fixtures/chunks/zarr_zlib/0.json b/tests/fixtures/chunks/zarr_zlib/0.json
deleted file mode 100644
index 66d650c..0000000
--- a/tests/fixtures/chunks/zarr_zlib/0.json
+++ /dev/null
@@ -1,23 +0,0 @@
-{
-  "codecs": [
-    {
-      "name": "bytes",
-      "type": "numcodecs",
-      "configuration": {
-        "endian": "little",
-        "data_type": "uint16",
-        "shape": [
-          1024,
-          1024
-        ]
-      }
-    },
-    {
-      "name": "numcodecs.zlib",
-      "type": "numcodecs",
-      "configuration": {
-        "level": 9
-      }
-    }
-  ]
-}
diff --git a/tests/fixtures/chunks/zarr_zstd/0 b/tests/fixtures/chunks/zarr_zstd/0
deleted file mode 100644
index b4db3e3..0000000
Binary files a/tests/fixtures/chunks/zarr_zstd/0 and /dev/null differ
diff --git a/tests/fixtures/chunks/zarr_zstd/0.json b/tests/fixtures/chunks/zarr_zstd/0.json
deleted file mode 100644
index 0292aa8..0000000
--- a/tests/fixtures/chunks/zarr_zstd/0.json
+++ /dev/null
@@ -1,24 +0,0 @@
-{
-  "codecs": [
-    {
-      "name": "bytes",
-      "type": "numcodecs",
-      "configuration": {
-        "endian": "little",
-        "data_type": "uint16",
-        "shape": [
-          1024,
-          1024
-        ]
-      }
-    },
-    {
-      "name": "zstd",
-      "type": "numcodecs",
-      "configuration": {
-        "level": 3,
-        "checksum": false
-      }
-    }
-  ]
-}
diff --git a/tests/fixtures/pipelines/cog-decode-pipeline.json b/tests/fixtures/pipelines/cog-decode-pipeline.json
new file mode 100644
index 0000000..678fb1c
--- /dev/null
+++ b/tests/fixtures/pipelines/cog-decode-pipeline.json
@@ -0,0 +1,28 @@
+{
+  "codec_id": "cog-zlib-predictor2",
+  "direction": "decode",
+  "inputs": {"bytes": {"type": "bytes"}},
+  "constants": {
+    "bytes_per_sample": {"type": "int", "value": 2},
+    "width":            {"type": "int", "value": 1024},
+    "level":            {"type": "int", "value": 9}
+  },
+  "outputs": {"bytes": "predictor2.bytes"},
+  "steps": {
+    "zlib": {
+      "codec_id": "zlib",
+      "inputs": {
+        "bytes": "input.bytes",
+        "level": "constant.level"
+      }
+    },
+    "predictor2": {
+      "codec_id": "tiff-predictor-2",
+      "inputs": {
+        "bytes":            "zlib.bytes",
+        "bytes_per_sample": "constant.bytes_per_sample",
+        "width":            "constant.width"
+      }
+    }
+  }
+}
diff --git a/tests/fixtures/pipelines/cog-encode-pipeline.json b/tests/fixtures/pipelines/cog-encode-pipeline.json
new file mode 100644
index 0000000..864d1e5
--- /dev/null
+++ b/tests/fixtures/pipelines/cog-encode-pipeline.json
@@ -0,0 +1,25 @@
+{
+  "codec_id": "cog-predictor2-zlib",
+  "direction": "encode",
+  "inputs": {"bytes": {"type": "bytes"}},
+  "constants": {
+    "bytes_per_sample": {"type": "int", "value": 2},
+    "width":            {"type": "int", "value": 1024},
+    "level":            {"type": "int", "value": 9}
+  },
+  "outputs": {"bytes": "zlib.bytes"},
+  "steps": {
+    "predictor2": {
+      "codec_id": "tiff-predictor-2",
+      "inputs": {
+        "bytes":            "input.bytes",
+        "bytes_per_sample": "constant.bytes_per_sample",
+        "width":            "constant.width"
+      }
+    },
+    "zlib": {
+      "codec_id": "zlib",
+      "inputs": {"bytes": "predictor2.bytes", "level": "constant.level"}
+    }
+  }
+}
diff --git a/tests/fixtures/pipelines/page-split-dag.json b/tests/fixtures/pipelines/page-split-dag.json
new file mode 100644
index 0000000..0d56698
--- /dev/null
+++ b/tests/fixtures/pipelines/page-split-dag.json
@@ -0,0 +1,42 @@
+{
+  "codec_id": "page-split-dag",
+  "direction": "encode",
+  "inputs": {"bytes": {"type": "bytes"}},
+  "constants": {
+    "rep_length": {"type": "uint", "value": 128},
+    "def_length": {"type": "uint", "value": 256}
+  },
+  "outputs": {
+    "rep_levels": "identity_rep.bytes",
+    "def_levels": "identity_def.bytes",
+    "data": "identity_data.bytes"
+  },
+  "steps": {
+    "page_split": {
+      "codec_id": "page-split",
+      "inputs": {
+        "bytes": "input.bytes",
+        "rep_length": "constant.rep_length",
+        "def_length": "constant.def_length"
+      }
+    },
+    "identity_rep": {
+      "codec_id": "identity",
+      "inputs": {
+        "bytes": "page_split.rep_levels"
+      }
+    },
+    "identity_def": {
+      "codec_id": "identity",
+      "inputs": {
+        "bytes": "page_split.def_levels"
+      }
+    },
+    "identity_data": {
+      "codec_id": "identity",
+      "inputs": {
+        "bytes": "page_split.data"
+      }
+    }
+  }
+}
diff --git a/tests/test_codecs.py b/tests/test_codecs.py
index 5bbe548..136ec8f 100644
--- a/tests/test_codecs.py
+++ b/tests/test_codecs.py
@@ -1,93 +1,82 @@
-import numpy as np
+"""Tests for codec wrapper classes and binary detection."""
+
+import struct
+from pathlib import Path
+
 import pytest
 
-from chonkle.codecs import (
-    BytesCodec,
-    Endian,
-    TiffPredictor2,
-    _numpy_dtype,
-)
-
-
-class TestTiffPredictor2:
-    def test_roundtrip(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([[10, 12, 15, 11], [100, 98, 95, 99]], dtype=np.uint16)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-    def test_encode_values(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([[10, 12, 15, 11]], dtype=np.int32)
-        encoded = codec.encode(arr)
-        expected = np.array([[10, 2, 3, -4]], dtype=np.int32)
-        np.testing.assert_array_equal(encoded, expected)
-
-    def test_roundtrip_1d(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([5, 10, 20, 15], dtype=np.float32)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-    def test_roundtrip_3d(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.arange(24, dtype=np.uint8).reshape(2, 3, 4)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-    def test_decode_with_out_param(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([[1, 3, 6, 10]], dtype=np.uint16)
-        encoded = codec.encode(arr)
-        out = np.empty_like(arr)
-        result = codec.decode(encoded, out=out)
-        np.testing.assert_array_equal(result, arr)
-        assert result is out
-
-
-class TestBytesCodec:
-    def test_roundtrip(self) -> None:
-        codec = BytesCodec(data_type="uint16", endian=Endian.LITTLE, shape=(2, 3))
-        arr = np.array([[1, 2, 3], [4, 5, 6]], dtype=np.uint16)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-    def test_roundtrip_big_endian(self) -> None:
-        codec = BytesCodec(data_type="uint16", endian=Endian.BIG, shape=(2, 2))
-        arr = np.array([[256, 512], [1024, 2048]], dtype=np.uint16)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-    def test_roundtrip_float32(self) -> None:
-        codec = BytesCodec(data_type="float32", endian=Endian.LITTLE, shape=(3,))
-        arr = np.array([1.5, 2.5, 3.5], dtype=np.float32)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-    def test_roundtrip_no_shape(self) -> None:
-        codec = BytesCodec(data_type="int32", endian=Endian.LITTLE)
-        arr = np.array([10, 20, 30], dtype=np.int32)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-        assert result.ndim == 1
-
-    def test_decode_with_out_param(self) -> None:
-        codec = BytesCodec(data_type="uint8", endian=Endian.LITTLE, shape=(4,))
-        arr = np.array([1, 2, 3, 4], dtype=np.uint8)
-        encoded = codec.encode(arr)
-        out = np.empty(4, dtype=np.uint8)
-        result = codec.decode(encoded, out=out)
-        np.testing.assert_array_equal(result, arr)
-        assert result is out
-
-    def test_string_endian(self) -> None:
-        codec = BytesCodec(data_type="uint16", endian="little", shape=(2,))
-        arr = np.array([100, 200], dtype=np.uint16)
-        result = codec.decode(codec.encode(arr))
-        np.testing.assert_array_equal(result, arr)
-
-
-class TestNumpyDtype:
-    def test_unsupported_type_raises(self) -> None:
-        with pytest.raises(ValueError, match="Unsupported zarr data_type"):
-            _numpy_dtype("complex128", Endian.LITTLE)
+from chonkle.codecs.core import _deserialize_port_map, _serialize_port_map
+from chonkle.wasm_signature import detect_codec_type
+
+# Wasm magic + version headers.
+_CORE_HEADER = b"\x00asm\x01\x00\x00\x00"
+_COMPONENT_HEADER = b"\x00asm\x0d\x00\x01\x00"
+
+
+class TestDetectCodecType:
+    def test_component_model(self, tmp_path: Path) -> None:
+        wasm = tmp_path / "codec.wasm"
+        wasm.write_bytes(_COMPONENT_HEADER)
+        assert detect_codec_type(wasm) == "component"
+
+    def test_core_wasm(self, tmp_path: Path) -> None:
+        wasm = tmp_path / "codec.wasm"
+        wasm.write_bytes(_CORE_HEADER)
+        assert detect_codec_type(wasm) == "core"
+
+    def test_not_wasm_raises(self, tmp_path: Path) -> None:
+        bad = tmp_path / "not.wasm"
+        bad.write_bytes(b"not a wasm file")
+        with pytest.raises(ValueError, match="Not a valid Wasm binary"):
+            detect_codec_type(bad)
+
+    def test_too_short_raises(self, tmp_path: Path) -> None:
+        short = tmp_path / "short.wasm"
+        short.write_bytes(b"\x00asm")
+        with pytest.raises(ValueError, match="Not a valid Wasm binary"):
+            detect_codec_type(short)
+
+    def test_unknown_version_raises(self, tmp_path: Path) -> None:
+        wasm = tmp_path / "future.wasm"
+        wasm.write_bytes(b"\x00asm\xff\x00\x00\x00")
+        with pytest.raises(ValueError, match="Unknown Wasm version"):
+            detect_codec_type(wasm)
+
+
+class TestPortMapSerialization:
+    """Tests for the core ABI port-map wire format serialization."""
+
+    def test_roundtrip_single_port(self) -> None:
+        port_map = [("bytes", b"\xde\xad\xbe\xef")]
+        assert _deserialize_port_map(_serialize_port_map(port_map)) == port_map
+
+    def test_roundtrip_multiple_ports(self) -> None:
+        port_map = [("bytes", b"hello"), ("level", b"3")]
+        assert _deserialize_port_map(_serialize_port_map(port_map)) == port_map
+
+    def test_roundtrip_empty_port_map(self) -> None:
+        port_map: list[tuple[str, bytes]] = []
+        assert _deserialize_port_map(_serialize_port_map(port_map)) == port_map
+
+    def test_roundtrip_empty_data(self) -> None:
+        port_map = [("empty", b"")]
+        assert _deserialize_port_map(_serialize_port_map(port_map)) == port_map
+
+    def test_wire_format_structure(self) -> None:
+        """Verify the wire format matches the spec (little-endian u32 lengths)."""
+        port_map = [("bytes", b"\xde\xad")]
+        raw = _serialize_port_map(port_map)
+        # u32 entry_count=1, u32 name_len=5, "bytes", u32 data_len=2, 0xDE 0xAD
+        assert struct.unpack_from("<I", raw, 0)[0] == 1  # entry_count
+        assert struct.unpack_from("<I", raw, 4)[0] == 5  # name_len
+        assert raw[8:13] == b"bytes"
+        assert struct.unpack_from("<I", raw, 13)[0] == 2  # data_len
+        assert raw[17:19] == b"\xde\xad"
+        assert len(raw) == 19
+
+    def test_roundtrip_large_data(self) -> None:
+        data = bytes(range(256)) * 1024  # 256 KB
+        port_map = [("bytes", data)]
+        result = _deserialize_port_map(_serialize_port_map(port_map))
+        assert result[0][0] == "bytes"
+        assert result[0][1] == data
diff --git a/tests/test_executor.py b/tests/test_executor.py
new file mode 100644
index 0000000..19478ba
--- /dev/null
+++ b/tests/test_executor.py
@@ -0,0 +1,1755 @@
+"""Tests for DAG executor via codec wrappers."""
+
+from collections.abc import Mapping
+from pathlib import Path
+from typing import Any
+
+import pytest
+
+from chonkle.codecs._base import Backend, Codec, PortMap
+from chonkle.codecs.core import CoreWasmCodec, CoreWasmRef
+from chonkle.executor import run
+from chonkle.pipeline import Direction, Pipeline, PreparedPipeline, prepare
+from chonkle.resolver import CodecStore, Resolver
+from chonkle.wasm_signature import Signature, embed_signature
+
+_REPO_ROOT = Path(__file__).parent.parent
+_CODEC_DIR = _REPO_ROOT / "codec"
+
+# Minimal valid Component Model header (magic + version, no sections).
+_CM_HEADER = b"\x00asm\x0d\x00\x01\x00"
+
+
+def _wasm_with_signature(signature: dict) -> bytes:
+    """Return minimal .wasm bytes with an embedded chonkle:signature section."""
+    sig = {"codec_id": "test-codec", **signature}
+    return embed_signature(_CM_HEADER, sig)
+
+
+# ---------- Fake codec for wiring tests ----------
+
+
+_DEFAULT_SIG = Signature.from_dict(
+    {
+        "codec_id": "fake",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "outputs": {"bytes": {"type": "bytes"}},
+    }
+)
+
+
+class _FakeCodec(Codec):
+    """Test double that records calls and returns canned outputs."""
+
+    def __init__(
+        self,
+        call_fn: Any = None,
+        sig: Signature | None = None,
+    ) -> None:
+        self._call_fn = call_fn or (lambda d, pm: [("bytes", b"output")])
+        self._sig = sig or _DEFAULT_SIG
+
+    @property
+    def codec_type(self) -> Backend:
+        return Backend.COMPONENT
+
+    @property
+    def codec_id(self) -> str:
+        return self._sig.codec_id or "fake"
+
+    @property
+    def implementation(self) -> str:
+        return self._sig.implementation or "fake"
+
+    def signature(self) -> Signature:
+        return self._sig
+
+    def call(self, direction: Direction, port_map: PortMap) -> PortMap:
+        return self._call_fn(direction, port_map)
+
+
+def _prepared(
+    pipeline: Pipeline,
+    direction: Direction | None = None,
+    codecs: Mapping[str, Codec] | None = None,
+) -> PreparedPipeline:
+    """Build a PreparedPipeline with fake codecs for wiring tests."""
+    if direction is None:
+        direction = pipeline.direction
+    if codecs is None:
+        codecs = {name: _FakeCodec() for name in pipeline.steps}
+    encode_only_inputs = {
+        name: frozenset(codecs[name].signature().encode_only_inputs())
+        for name in pipeline.steps
+    }
+    decode_only_inputs = {
+        name: frozenset(codecs[name].signature().decode_only_inputs())
+        for name in pipeline.steps
+    }
+    output_ports = {
+        name: tuple(codecs[name].signature().outputs.keys()) for name in pipeline.steps
+    }
+    return PreparedPipeline(
+        pipeline=pipeline,
+        direction=direction,
+        codecs=codecs,
+        encode_only_inputs=encode_only_inputs,
+        decode_only_inputs=decode_only_inputs,
+        output_ports=output_ports,
+    )
+
+
+# Tests below this marker require compiled codec .wasm files.
+CODEC_REQUIRED = pytest.mark.skipif(
+    True,
+    reason="requires compiled codec .wasm files (built in a separate session)",
+)
+
+_CORE_IDENTITY_WASM = _CODEC_DIR / "identity-core-c" / "zig-out" / "identity-core.wasm"
+CORE_CODEC_REQUIRED = pytest.mark.skipif(
+    not _CORE_IDENTITY_WASM.exists(),
+    reason=(
+        "Core identity codec .wasm not present — build codec/identity-core-c first"
+    ),
+)
+
+_COG_WASM_ZLIB = _CODEC_DIR / "zlib-rs" / "zlib.wasm"
+_COG_WASM_PREDICTOR2 = _CODEC_DIR / "tiff-predictor-2-c" / "tiff-predictor-2.wasm"
+COG_CODECS_REQUIRED = pytest.mark.skipif(
+    not (_COG_WASM_ZLIB.exists() and _COG_WASM_PREDICTOR2.exists()),
+    reason=(
+        "COG codec .wasm files not present"
+        " — build codec/zlib-rs and codec/tiff-predictor-2-c first"
+    ),
+)
+
+
+def _make_simple_pipeline() -> dict:
+    """Return a single-step encode pipeline dict."""
+    return {
+        "codec_id": "test",
+        "direction": "encode",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "constants": {"level": {"type": "int", "value": 3}},
+        "outputs": {"bytes": "codec.bytes"},
+        "steps": {
+            "codec": {
+                "codec_id": "codec",
+                "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+            }
+        },
+    }
+
+
+def _make_decode_pipeline() -> dict:
+    """Return a single-step decode pipeline dict (decode-declared)."""
+    return {
+        "codec_id": "test",
+        "direction": "decode",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "constants": {"level": {"type": "int", "value": 3}},
+        "outputs": {"bytes": "codec.bytes"},
+        "steps": {
+            "codec": {
+                "codec_id": "codec",
+                "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+            }
+        },
+    }
+
+
+# Signature with level marked as encode_only.
+_SIG_WITH_ENCODE_ONLY = Signature.from_dict(
+    {
+        "codec_id": "fake",
+        "inputs": {
+            "bytes": {"type": "bytes"},
+            "level": {"type": "int", "encode_only": True},
+        },
+        "outputs": {"bytes": {"type": "bytes"}},
+    }
+)
+
+
+class TestRunWiring:
+    """Verify executor wiring logic using fake codec instances."""
+
+    def test_passes_input_bytes_to_step(self, raw_chunk: bytes) -> None:
+        pipeline = Pipeline.parse(_make_simple_pipeline())
+        received: list = []
+
+        def fake_call(direction, port_map):
+            received.append((direction, list(port_map)))
+            return [("bytes", b"compressed")]
+
+        codecs = {"codec": _FakeCodec(call_fn=fake_call, sig=_SIG_WITH_ENCODE_ONLY)}
+        prepared = _prepared(pipeline, codecs=codecs)
+        result = run(prepared, {"bytes": raw_chunk})
+
+        assert result == {"bytes": b"compressed"}
+        assert len(received) == 1
+        direction, port_map = received[0]
+        assert direction == "encode"
+        assert ("bytes", raw_chunk) in port_map
+
+    def test_routes_step_output_to_next_step_input(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        call_log: list = []
+
+        def fake_a(direction, port_map):
+            call_log.append(("a", list(port_map)))
+            return [("bytes", b"after_a")]
+
+        def fake_b(direction, port_map):
+            call_log.append(("b", list(port_map)))
+            return [("bytes", b"after_b")]
+
+        codecs = {
+            "step_a": _FakeCodec(call_fn=fake_a),
+            "step_b": _FakeCodec(call_fn=fake_b),
+        }
+        prepared = _prepared(pipeline, codecs=codecs)
+        result = run(prepared, {"bytes": b"original"})
+
+        assert result == {"bytes": b"after_b"}
+        _, first_pm = call_log[0]
+        _, second_pm = call_log[1]
+        assert first_pm == [("bytes", b"original")]
+        assert second_pm == [("bytes", b"after_a")]
+
+    def test_constants_passed_as_json_bytes(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"level": {"type": "int", "value": 5}},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {
+                        "bytes": "input.bytes",
+                        "level": "constant.level",
+                    },
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        received_pm: list = []
+
+        def fake_call(direction, port_map):
+            received_pm.extend(port_map)
+            return [("bytes", b"output")]
+
+        codecs = {"s": _FakeCodec(call_fn=fake_call)}
+        prepared = _prepared(pipeline, codecs=codecs)
+        run(prepared, {"bytes": b"data"})
+
+        pm_dict = dict(received_pm)
+        assert pm_dict["level"] == b"5"
+
+    def test_encode_only_inputs_skipped_during_decode(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"sym_table": {"type": "string", "value": "abc"}},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {
+                        "bytes": "input.bytes",
+                        "sym_table": "constant.sym_table",
+                    },
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        received_pm: list = []
+
+        def fake_call(direction, port_map):
+            received_pm.extend(port_map)
+            return [("bytes", b"decoded")]
+
+        sig = Signature.from_dict(
+            {
+                "codec_id": "fake",
+                "inputs": {
+                    "bytes": {"type": "bytes"},
+                    "sym_table": {"type": "string", "encode_only": True},
+                },
+                "outputs": {"bytes": {"type": "bytes"}},
+            }
+        )
+        codecs = {"s": _FakeCodec(call_fn=fake_call, sig=sig)}
+        prepared = _prepared(pipeline, codecs=codecs)
+        run(prepared, {"bytes": b"encoded"})
+
+        port_names = [name for name, _ in received_pm]
+        assert "sym_table" not in port_names
+        assert "bytes" in port_names
+
+    def test_encode_only_inputs_included_during_encode(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"sym_table": {"type": "string", "value": "abc"}},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {
+                        "bytes": "input.bytes",
+                        "sym_table": "constant.sym_table",
+                    },
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        received_pm: list = []
+
+        def fake_call(direction, port_map):
+            received_pm.extend(port_map)
+            return [("bytes", b"encoded")]
+
+        sig = Signature.from_dict(
+            {
+                "codec_id": "fake",
+                "inputs": {
+                    "bytes": {"type": "bytes"},
+                    "sym_table": {"type": "string", "encode_only": True},
+                },
+                "outputs": {"bytes": {"type": "bytes"}},
+            }
+        )
+        codecs = {"s": _FakeCodec(call_fn=fake_call, sig=sig)}
+        prepared = _prepared(pipeline, codecs=codecs)
+        run(prepared, {"bytes": b"data"})
+
+        port_names = [name for name, _ in received_pm]
+        assert "sym_table" in port_names
+
+    def test_encode_only_pipeline_input_not_required_during_decode(self) -> None:
+        """encode_only pipeline inputs are not required when running decode."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {
+                "bytes": {"type": "bytes"},
+                "dtype": {"type": "string", "encode_only": True},
+            },
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        codecs = {"s": _FakeCodec(call_fn=lambda d, pm: [("bytes", b"decoded")])}
+        prepared = _prepared(pipeline, codecs=codecs)
+        result = run(prepared, {"bytes": b"encoded"})
+        assert result == {"bytes": b"decoded"}
+
+    def test_encode_only_pipeline_input_required_during_encode(self) -> None:
+        """encode_only pipeline inputs ARE required when running encode."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {
+                "bytes": {"type": "bytes"},
+                "dtype": {"type": "string", "encode_only": True},
+            },
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        prepared = _prepared(pipeline)
+        with pytest.raises(ValueError, match="Missing pipeline input"):
+            run(prepared, {"bytes": b"data"})
+
+    def test_missing_input_raises(self) -> None:
+        pipeline = Pipeline.parse(_make_simple_pipeline())
+        prepared = _prepared(pipeline)
+        with pytest.raises(ValueError, match="Missing pipeline input"):
+            run(prepared, {})
+
+    def test_fan_out_routing(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {
+                "a_out": "proc_a.bytes",
+                "b_out": "proc_b.bytes",
+            },
+            "steps": {
+                "split": {
+                    "codec_id": "page-split",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "proc_a": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "split.a"},
+                },
+                "proc_b": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "split.b"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+
+        def fake_split(direction, port_map):
+            data_bytes = port_map[0][1]
+            mid = len(data_bytes) // 2
+            return [("a", data_bytes[:mid]), ("b", data_bytes[mid:])]
+
+        split_sig = Signature.from_dict(
+            {
+                "codec_id": "fake-split",
+                "inputs": {"bytes": {"type": "bytes"}},
+                "outputs": {"a": {"type": "bytes"}, "b": {"type": "bytes"}},
+            }
+        )
+        codecs = {
+            "split": _FakeCodec(call_fn=fake_split, sig=split_sig),
+            "proc_a": _FakeCodec(call_fn=lambda d, pm: pm),
+            "proc_b": _FakeCodec(call_fn=lambda d, pm: pm),
+        }
+        prepared = _prepared(pipeline, codecs=codecs)
+        result = run(prepared, {"bytes": b"abcdefgh"})
+
+        assert result["a_out"] == b"abcd"
+        assert result["b_out"] == b"efgh"
+
+
+class TestResolverIntegration:
+    """Verify that prepare() uses the Resolver to find codecs."""
+
+    def test_resolver_resolves_codec_id(self, tmp_path: Path) -> None:
+        """The resolver maps codec_id to a wasm file."""
+        wasm_file = tmp_path / "codec.wasm"
+        sig = {
+            "codec_id": "some-codec",
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(sig))
+
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+
+        resolver = Resolver(paths={"some-codec": wasm_file})
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_unresolvable_codec_raises(self) -> None:
+        """A codec_id with no available implementation raises ValueError."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "nonexistent-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+
+        resolver = Resolver(codec_store=Path("/nonexistent"))
+        with pytest.raises(ValueError, match="No codec implementation found"):
+            prepare(data, "encode", resolver=resolver)
+
+
+class TestForceSourceResolution:
+    """Verify force_sources bypasses the store and overwrites on install."""
+
+    def _make_wasm(self, codec_id: str, impl: str = "impl-a") -> bytes:
+        return _wasm_with_signature(
+            {
+                "codec_id": codec_id,
+                "implementation": impl,
+                "inputs": {"bytes": {"type": "bytes"}},
+                "outputs": {"bytes": {"type": "bytes"}},
+            }
+        )
+
+    def test_force_install_overwrites_store_entry(self, tmp_path: Path) -> None:
+        """force_install=True causes install() to overwrite an existing file."""
+        store = CodecStore(tmp_path)
+
+        wasm_v1 = tmp_path / "v1.wasm"
+        wasm_v1.write_bytes(self._make_wasm("fc", "same-impl"))
+        entry1 = store.install(wasm_v1)
+        mtime_after_v1 = entry1.location.stat().st_mtime
+
+        # Build a distinct v2 binary (different implementation tag to get
+        # different bytes, but re-use the same implementation *name* so
+        # install targets the same store path).
+        wasm_v2 = tmp_path / "v2.wasm"
+        v2_bytes = self._make_wasm("fc", "same-impl")
+        wasm_v2.write_bytes(v2_bytes)
+
+        # Without force_install, existing file is kept (mtime unchanged)
+        store.install(wasm_v2)
+        assert entry1.location.stat().st_mtime == mtime_after_v1
+
+        # With force_install, file is overwritten
+        entry2 = store.install(wasm_v2, force_install=True)
+        assert entry2.location == entry1.location
+        assert entry2.location.stat().st_mtime >= mtime_after_v1
+
+    def test_force_source_bypasses_store(self, tmp_path: Path) -> None:
+        """force_sources is consulted before the local store."""
+        store_dir = tmp_path / "store"
+        store_dir.mkdir()
+
+        # Pre-populate store with impl-a
+        store_codec_dir = store_dir / "fc"
+        store_codec_dir.mkdir()
+        (store_codec_dir / "impl-a.wasm").write_bytes(self._make_wasm("fc", "impl-a"))
+
+        # Put a different impl in a file:// source
+        source_wasm = tmp_path / "source.wasm"
+        source_wasm.write_bytes(self._make_wasm("fc", "impl-b"))
+
+        resolver = Resolver(
+            codec_store=store_dir,
+            force_sources={"fc": f"file://{source_wasm}"},
+        )
+
+        # Before resolve, only impl-a is in the store
+        assert "impl-b" not in CodecStore(store_dir).get_index().get("fc", {})
+
+        codec = resolver.resolve("fc")
+        assert codec.signature().implementation == "impl-b"
+
+        # After resolve, impl-b is installed in the store
+        assert "impl-b" in CodecStore(store_dir).get_index().get("fc", {})
+
+    def test_explicit_path_takes_priority_over_force_source(
+        self, tmp_path: Path
+    ) -> None:
+        """paths (step 0) beats force_sources (step 1)."""
+        wasm_file = tmp_path / "local.wasm"
+        wasm_file.write_bytes(self._make_wasm("fc", "local-impl"))
+
+        resolver = Resolver(
+            codec_store=Path("/nonexistent"),
+            paths={"fc": wasm_file},
+            # If force_sources were consulted, this would try to download
+            # from a non-existent URL and fail.
+            force_sources={"fc": "https://should-not-be-called.example.com/x.wasm"},
+        )
+        codec = resolver.resolve("fc")
+        assert codec.signature().implementation == "local-impl"
+
+    def test_force_source_only_affects_specified_codec(self, tmp_path: Path) -> None:
+        """Codecs not in force_sources resolve normally."""
+        from chonkle.codecs.native import NativeCodec
+
+        resolver = Resolver(
+            codec_store=Path("/nonexistent"),
+            force_sources={"other-codec": "file:///doesnt-matter.wasm"},
+        )
+        codec = resolver.resolve("zlib")
+        assert isinstance(codec, NativeCodec)
+
+
+class TestSignatureValidation:
+    """Signature format: protospec codec signature — inputs/outputs are dicts."""
+
+    def _make_pipeline(
+        self,
+        *,
+        codec_id: str = "some-codec",
+        direction: str = "encode",
+        step_inputs: dict | None = None,
+        pipeline_output_ref: str = "s.bytes",
+    ) -> dict:
+        """Build a single-step pipeline dict for signature validation tests."""
+        if step_inputs is None:
+            step_inputs = {"bytes": "input.bytes"}
+        pipeline_input_names = list(
+            {
+                ref.split(".")[1]
+                for ref in step_inputs.values()
+                if ref.startswith("input.")
+            }
+        )
+        pipeline_inputs = {
+            name: {"type": "bytes"} for name in (pipeline_input_names or ["bytes"])
+        }
+        return {
+            "codec_id": "test",
+            "direction": direction,
+            "inputs": pipeline_inputs,
+            "constants": {},
+            "outputs": {"bytes": pipeline_output_ref},
+            "steps": {
+                "s": {
+                    "codec_id": codec_id,
+                    "inputs": step_inputs,
+                }
+            },
+        }
+
+    def _resolver(self, codec_id: str, wasm_file: Path) -> Resolver:
+        return Resolver(paths={codec_id: wasm_file})
+
+    def test_signature_mismatch_raises(self, tmp_path: Path) -> None:
+        """A signature with an unknown output port name raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"wrong_port": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="does not declare output port"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_matching_signature_does_not_raise(self, tmp_path: Path) -> None:
+        """A signature matching declared ports does not raise."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {
+                "bytes": {"type": "bytes", "required": True},
+                "level": {"type": "int", "required": False, "encode_only": True},
+            },
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline(
+            step_inputs={"bytes": "input.bytes"},
+        )
+        resolver = self._resolver("some-codec", wasm_file)
+
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_signature_unknown_input_raises(self, tmp_path: Path) -> None:
+        """A step wiring a port not in the signature inputs raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline(
+            step_inputs={"wrong": "input.bytes"},
+        )
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="not valid signature encode inputs"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_signature_encode_only_excluded_during_decode(self, tmp_path: Path) -> None:
+        """encode_only inputs are excluded from valid decode inputs; no raise."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {
+                "bytes": {"type": "bytes", "required": True},
+                "level": {"type": "int", "required": False, "encode_only": True},
+            },
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline(
+            direction="decode",
+            step_inputs={"bytes": "input.bytes", "level": "input.bytes"},
+        )
+        resolver = self._resolver("some-codec", wasm_file)
+
+        prepared = prepare(data, "decode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_signature_subset_output_passes(self, tmp_path: Path) -> None:
+        """Declaring fewer outputs than the signature advertises is valid."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}, "checksum": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_signature_outputs_key_absent_skips_check(self, tmp_path: Path) -> None:
+        """A signature with no 'outputs' key skips output validation."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {"inputs": {"bytes": {"type": "bytes", "required": True}}}
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        resolver = self._resolver("some-codec", wasm_file)
+
+        # Wiring validation will skip step output port check when
+        # the signature has no outputs key (empty set, not missing step).
+        # The pipeline output ref "s.bytes" won't match but the wiring
+        # validation catches it. Use an input passthrough instead.
+        data = self._make_pipeline(pipeline_output_ref="input.bytes")
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_signature_inputs_key_absent_skips_check(self, tmp_path: Path) -> None:
+        """A signature with no 'inputs' key skips input validation."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {"outputs": {"bytes": {"type": "bytes"}}}
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_signature_both_wrong_reports_both(self, tmp_path: Path) -> None:
+        """When inputs are wrong, the error reports the input issue."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline(
+            step_inputs={"bad_in": "input.bytes"},
+        )
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="not valid signature encode inputs"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_missing_signature_raises(self, tmp_path: Path) -> None:
+        """A .wasm with no embedded chonkle:signature section raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        wasm_file.write_bytes(_CM_HEADER)
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="chonkle:signature"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_two_steps_both_wrong_reports_both(self, tmp_path: Path) -> None:
+        """Signature errors from multiple steps are all reported in one exception."""
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_a = tmp_path / "codec_a.wasm"
+        wasm_b = tmp_path / "codec_b.wasm"
+        wasm_a.write_bytes(_wasm_with_signature(signature))
+        wasm_b.write_bytes(_wasm_with_signature(signature))
+
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "codec-a",
+                    "inputs": {"bad_a": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "codec-b",
+                    "inputs": {"bad_b": "step_a.bytes"},
+                },
+            },
+        }
+
+        resolver = Resolver(
+            paths={
+                "codec-a": wasm_a,
+                "codec-b": wasm_b,
+            }
+        )
+
+        with pytest.raises(ValueError, match="step_a") as exc_info:
+            prepare(data, "encode", resolver=resolver)
+
+        assert "step_b" in str(exc_info.value)
+
+    def test_signature_input_port_missing_type_raises(self, tmp_path: Path) -> None:
+        """A signature input port without a 'type' field raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"required": True}},  # no "type"
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="missing required 'type' field"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_signature_output_port_missing_type_raises(self, tmp_path: Path) -> None:
+        """A signature output port without a 'type' field raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {}},  # no "type"
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="missing required 'type' field"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_type_mismatch_pipeline_input_raises(self, tmp_path: Path) -> None:
+        """Pipeline input type mismatch against codec signature raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"data": {"type": "bytes", "required": True}},
+            "outputs": {"data": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"data": {"type": "int"}},
+            "constants": {},
+            "outputs": {"data": "s.data"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"data": "input.data"},
+                }
+            },
+        }
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="provides type"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_type_mismatch_constant_raises(self, tmp_path: Path) -> None:
+        """Constant type mismatch against codec signature raises ValueError."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {
+                "bytes": {"type": "bytes", "required": True},
+                "level": {"type": "bytes", "required": False},
+            },
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"level": {"type": "int", "value": 3}},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+                }
+            },
+        }
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="provides type"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_type_mismatch_step_output_raises(self, tmp_path: Path) -> None:
+        """Step output type mismatch against downstream codec input raises."""
+        sig_a = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        sig_b = {
+            "inputs": {"bytes": {"type": "int", "required": True}},
+            "outputs": {"bytes": {"type": "int"}},
+        }
+        wasm_a = tmp_path / "codec_a.wasm"
+        wasm_b = tmp_path / "codec_b.wasm"
+        wasm_a.write_bytes(_wasm_with_signature(sig_a))
+        wasm_b.write_bytes(_wasm_with_signature(sig_b))
+
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "codec-a",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "codec-b",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+
+        resolver = Resolver(
+            paths={
+                "codec-a": wasm_a,
+                "codec-b": wasm_b,
+            }
+        )
+
+        with pytest.raises(ValueError, match="provides type"):
+            prepare(data, "encode", resolver=resolver)
+
+    def test_type_match_passes(self, tmp_path: Path) -> None:
+        """Matching types across pipeline input and codec signature do not raise."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = self._make_pipeline()
+        resolver = self._resolver("some-codec", wasm_file)
+
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_missing_type_in_source_skips_check(self, tmp_path: Path) -> None:
+        """Pipeline input with no 'type' key skips type check without raising."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        resolver = self._resolver("some-codec", wasm_file)
+
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert "s" in prepared.codecs
+
+    def test_step_output_port_wiring_validated(self, tmp_path: Path) -> None:
+        """Wiring ref to a non-existent output port raises at prepare() time."""
+        wasm_file = tmp_path / "codec.wasm"
+        signature = {
+            "inputs": {"bytes": {"type": "bytes", "required": True}},
+            "outputs": {"bytes": {"type": "bytes"}},
+        }
+        wasm_file.write_bytes(_wasm_with_signature(signature))
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"out": "s.missing_port"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        resolver = self._resolver("some-codec", wasm_file)
+
+        with pytest.raises(ValueError, match="does not declare output port"):
+            prepare(data, "encode", resolver=resolver)
+
+
+class TestInvertedExecution:
+    """Verify inverted DAG execution (direction != pipeline.direction)."""
+
+    def test_inverted_single_step_calls_encode(self) -> None:
+        """A decode-declared pipeline run as encode calls encode on the codec."""
+        pipeline = Pipeline.parse(_make_decode_pipeline())
+        received_directions: list[str] = []
+
+        def fake_call(direction, port_map):
+            received_directions.append(direction)
+            return [("bytes", b"raw")]
+
+        codecs = {"codec": _FakeCodec(call_fn=fake_call, sig=_SIG_WITH_ENCODE_ONLY)}
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        run(prepared, {"bytes": b"compressed"})
+
+        assert received_directions == ["encode"]
+
+    def test_inverted_port_map_built_from_step_outputs(self) -> None:
+        """Inverted encode: port-map is built from signature outputs."""
+        pipeline = Pipeline.parse(_make_decode_pipeline())
+        received_port_maps: list = []
+
+        def fake_call(direction, port_map):
+            received_port_maps.append(list(port_map))
+            return [("bytes", b"raw")]
+
+        codecs = {"codec": _FakeCodec(call_fn=fake_call, sig=_SIG_WITH_ENCODE_ONLY)}
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        run(prepared, {"bytes": b"compressed"})
+
+        assert len(received_port_maps) == 1
+        pm = dict(received_port_maps[0])
+        assert "bytes" in pm
+        assert pm["bytes"] == b"compressed"
+        assert "level" in pm
+        assert pm["level"] == b"3"
+
+    def test_inverted_result_keyed_by_pipeline_inputs(self) -> None:
+        """Inverted encode: result is keyed by pipeline.inputs names."""
+        pipeline = Pipeline.parse(_make_decode_pipeline())
+        codecs = {
+            "codec": _FakeCodec(
+                call_fn=lambda d, pm: [("bytes", b"encoded_result")],
+                sig=_SIG_WITH_ENCODE_ONLY,
+            )
+        }
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        result = run(prepared, {"bytes": b"raw"})
+        assert result == {"bytes": b"encoded_result"}
+
+    def test_inverted_execution_order_reversed(self) -> None:
+        """Inverted encode on a two-step decode pipeline runs in reversed order."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        call_order: list[str] = []
+
+        def fake_a(direction, port_map):
+            call_order.append("a")
+            return [("bytes", b"result")]
+
+        def fake_b(direction, port_map):
+            call_order.append("b")
+            return [("bytes", b"result")]
+
+        codecs = {
+            "step_a": _FakeCodec(call_fn=fake_a),
+            "step_b": _FakeCodec(call_fn=fake_b),
+        }
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        run(prepared, {"bytes": b"raw"})
+
+        assert call_order == ["b", "a"]
+
+    def test_inverted_encode_routes_results_backward(self) -> None:
+        """Inverted encode: step results route via decode-direction input wiring."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        received: dict[str, list] = {}
+
+        def fake_a(direction, port_map):
+            received["a"] = list(port_map)
+            return [("bytes", b"a_encoded")]
+
+        def fake_b(direction, port_map):
+            received["b"] = list(port_map)
+            return [("bytes", b"b_encoded")]
+
+        codecs = {
+            "step_a": _FakeCodec(call_fn=fake_a),
+            "step_b": _FakeCodec(call_fn=fake_b),
+        }
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        result = run(prepared, {"bytes": b"raw_input"})
+
+        assert dict(received["b"]) == {"bytes": b"raw_input"}
+        assert dict(received["a"]) == {"bytes": b"b_encoded"}
+        assert result == {"bytes": b"a_encoded"}
+
+    def test_inverted_fan_in_page_split_pattern(self) -> None:
+        """Inverted encode on a fan-out decode pipeline fans back in correctly."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {
+                "rep": "identity_rep.bytes",
+                "def_": "identity_def.bytes",
+                "data": "identity_data.bytes",
+            },
+            "steps": {
+                "page_split": {
+                    "codec_id": "page-split",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "identity_rep": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "page_split.rep_levels"},
+                },
+                "identity_def": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "page_split.def_levels"},
+                },
+                "identity_data": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "page_split.data"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+
+        def fake_split(direction, port_map):
+            pm = dict(port_map)
+            combined = (
+                pm.get("rep_levels", b"")
+                + pm.get("def_levels", b"")
+                + pm.get("data", b"")
+            )
+            return [("bytes", combined)]
+
+        split_sig = Signature.from_dict(
+            {
+                "codec_id": "fake-split",
+                "inputs": {"bytes": {"type": "bytes"}},
+                "outputs": {
+                    "rep_levels": {"type": "bytes"},
+                    "def_levels": {"type": "bytes"},
+                    "data": {"type": "bytes"},
+                },
+            }
+        )
+        codecs = {
+            "page_split": _FakeCodec(call_fn=fake_split, sig=split_sig),
+            "identity_rep": _FakeCodec(call_fn=lambda d, pm: pm),
+            "identity_def": _FakeCodec(call_fn=lambda d, pm: pm),
+            "identity_data": _FakeCodec(call_fn=lambda d, pm: pm),
+        }
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        result = run(
+            prepared,
+            {"rep": b"RRR", "def_": b"DDD", "data": b"VVV"},
+        )
+
+        assert "bytes" in result
+        assert b"RRR" in result["bytes"]
+        assert b"DDD" in result["bytes"]
+        assert b"VVV" in result["bytes"]
+
+    def test_inverted_missing_input_raises(self) -> None:
+        """Inverted direction with a missing pipeline.outputs key raises ValueError."""
+        pipeline = Pipeline.parse(_make_decode_pipeline())
+        prepared = _prepared(pipeline, direction="encode")
+        with pytest.raises(ValueError, match="Missing pipeline input"):
+            run(prepared, {})
+
+    def test_inverted_encode_only_excluded_from_decode_call(self) -> None:
+        """Encode-declared pipeline inverted to decode: encode_only_inputs absent."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"level": {"type": "int", "value": 9}},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        received_port_maps: list = []
+
+        def fake_call(direction, port_map):
+            received_port_maps.append(list(port_map))
+            return [("bytes", b"decoded")]
+
+        sig = Signature.from_dict(
+            {
+                "codec_id": "fake",
+                "inputs": {
+                    "bytes": {"type": "bytes"},
+                    "level": {"type": "int", "encode_only": True},
+                },
+                "outputs": {"bytes": {"type": "bytes"}},
+            }
+        )
+        codecs = {"s": _FakeCodec(call_fn=fake_call, sig=sig)}
+        prepared = _prepared(pipeline, direction="decode", codecs=codecs)
+        run(prepared, {"bytes": b"encoded"})
+
+        assert len(received_port_maps) == 1
+        port_names = [name for name, _ in received_port_maps[0]]
+        assert "level" not in port_names
+        assert "bytes" in port_names
+
+
+@CODEC_REQUIRED
+class TestPageSplitFanOut:
+    def test_fan_out_produces_three_outputs(
+        self,
+        page_split_input: tuple[bytes, int, int],
+        page_split_pipeline_json: dict,
+    ) -> None:
+        """page-split produces three non-empty output byte buffers."""
+        data, rep_length, def_length = page_split_input
+
+        page_split_pipeline_json["constants"]["rep_length"]["value"] = rep_length
+        page_split_pipeline_json["constants"]["def_length"]["value"] = def_length
+
+        prepared = prepare(
+            page_split_pipeline_json, page_split_pipeline_json["direction"]
+        )
+        result = run(prepared, {"bytes": data})
+
+        assert set(result.keys()) == {"rep_levels", "def_levels", "data"}
+        for port_name, value in result.items():
+            assert len(value) > 0, f"Expected non-empty output for {port_name!r}"
+
+    def test_split_segment_sizes(
+        self,
+        page_split_input: tuple[bytes, int, int],
+        page_split_pipeline_json: dict,
+    ) -> None:
+        """page-split segments have the sizes dictated by rep_length and def_length."""
+        data, rep_length, def_length = page_split_input
+
+        page_split_pipeline_json["constants"]["rep_length"]["value"] = rep_length
+        page_split_pipeline_json["constants"]["def_length"]["value"] = def_length
+
+        prepared = prepare(
+            page_split_pipeline_json, page_split_pipeline_json["direction"]
+        )
+        result = run(prepared, {"bytes": data})
+
+        assert len(result["rep_levels"]) == rep_length
+        assert len(result["def_levels"]) == def_length
+        assert len(result["data"]) == len(data) - rep_length - def_length
+
+
+@CORE_CODEC_REQUIRED
+class TestCoreWasmPipeline:
+    """Tests for core wasm codec steps in DAG pipelines."""
+
+    def test_single_step_core_encode(self, raw_chunk: bytes) -> None:
+        """A single core wasm identity step produces identical output on encode."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "encode", resolver=resolver)
+        assert prepared.codecs["s"].codec_type == "core"
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
+
+    def test_single_step_core_decode(self, raw_chunk: bytes) -> None:
+        """A single core wasm identity step produces identical output on decode."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "decode", resolver=resolver)
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
+
+    def test_two_step_core_pipeline(self, raw_chunk: bytes) -> None:
+        """Two sequential core wasm identity steps route data correctly."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "encode", resolver=resolver)
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
+
+    def test_inverted_core_pipeline(self, raw_chunk: bytes) -> None:
+        """An inverted core wasm pipeline produces identical output."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "encode", resolver=resolver)
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
+
+    def test_core_codec_resolver_detects_type(self) -> None:
+        """The resolver instantiates a CoreWasmCodec for core wasm binaries."""
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        codec = resolver.resolve("identity")
+        assert codec.codec_type == "core"
+        assert codec.codec_id == "identity"
+        assert codec.implementation == "identity-core-c"
+
+
+@COG_CODECS_REQUIRED
+class TestCogChunkPipeline:
+    """Round-trip and decode tests for the COG zlib+tiff-predictor-2 pipeline."""
+
+    _TILE_BYTES = 1024 * 1024 * 2  # 1024x1024 uint16
+
+    def test_decode_declared_forward_decode_output_size(
+        self,
+        cog_chunk: bytes,
+        cog_decode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Decode-declared pipeline forward decode: output is 1024x1024x2 bytes."""
+        prepared = prepare(
+            cog_decode_pipeline_json, "decode", resolver=cog_codec_resolver
+        )
+        result = run(prepared, {"bytes": cog_chunk})
+        assert len(result["bytes"]) == self._TILE_BYTES
+
+    def test_decode_declared_forward_decode_output_is_bytes(
+        self,
+        cog_chunk: bytes,
+        cog_decode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Decode-declared pipeline running forward decode returns a bytes object."""
+        prepared = prepare(
+            cog_decode_pipeline_json, "decode", resolver=cog_codec_resolver
+        )
+        result = run(prepared, {"bytes": cog_chunk})
+        assert isinstance(result["bytes"], bytes)
+
+    def test_encode_declared_inverted_decode_output_size(
+        self,
+        cog_chunk: bytes,
+        cog_encode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Encode-declared pipeline inverted decode: output is 1024x1024x2 bytes."""
+        prepared = prepare(
+            cog_encode_pipeline_json, "decode", resolver=cog_codec_resolver
+        )
+        result = run(prepared, {"bytes": cog_chunk})
+        assert len(result["bytes"]) == self._TILE_BYTES
+
+    def test_encode_declared_inverted_decode_output_is_bytes(
+        self,
+        cog_chunk: bytes,
+        cog_encode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Encode-declared pipeline running inverted decode returns a bytes object."""
+        prepared = prepare(
+            cog_encode_pipeline_json, "decode", resolver=cog_codec_resolver
+        )
+        result = run(prepared, {"bytes": cog_chunk})
+        assert isinstance(result["bytes"], bytes)
+
+    def test_decode_declared_pipeline_roundtrip(
+        self,
+        raw_chunk: bytes,
+        cog_decode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Decode-declared pipeline roundtrip: inverted encode then forward decode."""
+        tile = (raw_chunk * ((self._TILE_BYTES // len(raw_chunk)) + 1))[
+            : self._TILE_BYTES
+        ]
+        enc_prepared = prepare(
+            cog_decode_pipeline_json, "encode", resolver=cog_codec_resolver
+        )
+        encoded = run(enc_prepared, {"bytes": tile})
+        dec_prepared = prepare(
+            cog_decode_pipeline_json, "decode", resolver=cog_codec_resolver
+        )
+        decoded = run(dec_prepared, {"bytes": encoded["bytes"]})
+        assert decoded["bytes"] == tile
+
+    def test_encode_declared_pipeline_roundtrip(
+        self,
+        raw_chunk: bytes,
+        cog_encode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Encode-declared pipeline roundtrip: forward encode then inverted decode."""
+        tile = (raw_chunk * ((self._TILE_BYTES // len(raw_chunk)) + 1))[
+            : self._TILE_BYTES
+        ]
+        enc_prepared = prepare(
+            cog_encode_pipeline_json, "encode", resolver=cog_codec_resolver
+        )
+        encoded = run(enc_prepared, {"bytes": tile})
+        dec_prepared = prepare(
+            cog_encode_pipeline_json, "decode", resolver=cog_codec_resolver
+        )
+        decoded = run(dec_prepared, {"bytes": encoded["bytes"]})
+        assert decoded["bytes"] == tile
+
+    def test_encode_declared_inverted_decode_matches_decode_declared_forward(
+        self,
+        cog_chunk: bytes,
+        cog_decode_pipeline_json: dict,
+        cog_encode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Encode-declared pipeline inverted decode yields the same raw tile
+        as decode-declared pipeline forward decode."""
+        fwd = prepare(cog_decode_pipeline_json, "decode", resolver=cog_codec_resolver)
+        inv = prepare(cog_encode_pipeline_json, "decode", resolver=cog_codec_resolver)
+        forward = run(fwd, {"bytes": cog_chunk})
+        inverted = run(inv, {"bytes": cog_chunk})
+        assert inverted["bytes"] == forward["bytes"]
+
+    def test_decode_declared_inverted_encode_matches_encode_declared_forward(
+        self,
+        raw_chunk: bytes,
+        cog_decode_pipeline_json: dict,
+        cog_encode_pipeline_json: dict,
+        cog_codec_resolver: Resolver,
+    ) -> None:
+        """Decode-declared pipeline inverted encode yields the same compressed bytes
+        as encode-declared pipeline forward encode."""
+        tile = (raw_chunk * ((self._TILE_BYTES // len(raw_chunk)) + 1))[
+            : self._TILE_BYTES
+        ]
+        inv = prepare(cog_decode_pipeline_json, "encode", resolver=cog_codec_resolver)
+        fwd = prepare(cog_encode_pipeline_json, "encode", resolver=cog_codec_resolver)
+        inverted = run(inv, {"bytes": tile})
+        forward = run(fwd, {"bytes": tile})
+        assert inverted["bytes"] == forward["bytes"]
+
+
+class TestSingleCopyWiring:
+    """Verify that CoreWasmRef values flow through the executor correctly.
+
+    Uses _FakeCodec variants to test materialization logic without needing
+    compiled wasm modules.
+    """
+
+    @staticmethod
+    def _make_ref(data: bytes) -> CoreWasmRef:
+        """Create a CoreWasmRef backed by mock memory for wiring tests."""
+
+        class _MockMemory:
+            def read(self, store: Any, start: int, end: int) -> bytes:
+                return data[start:end]
+
+        class _MockCodec:
+            memory = _MockMemory()
+            store = None
+
+        return CoreWasmRef(codec=_MockCodec(), ptr=0, length=len(data))  # type: ignore[arg-type]
+
+    def test_ref_materialized_for_component_codec(self) -> None:
+        """CoreWasmRef from an upstream step is materialized when passed
+        to a downstream component (non-core) codec."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        received_b: list = []
+
+        ref = self._make_ref(b"from_core")
+
+        def fake_a(direction, port_map):
+            return [("bytes", ref)]
+
+        def fake_b(direction, port_map):
+            received_b.extend(port_map)
+            return [("bytes", b"final")]
+
+        codecs: dict[str, Codec] = {
+            "step_a": _FakeCodec(call_fn=fake_a),
+            "step_b": _FakeCodec(call_fn=fake_b),
+        }
+        prepared = _prepared(pipeline, codecs=codecs)
+        result = run(prepared, {"bytes": b"input"})
+
+        assert result == {"bytes": b"final"}
+        # step_b is a component codec, so the ref should be materialized
+        _, value = received_b[0]
+        assert isinstance(value, bytes)
+        assert value == b"from_core"
+
+    def test_ref_materialized_in_final_output(self) -> None:
+        """CoreWasmRef in the final pipeline output is materialized to bytes."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        ref = self._make_ref(b"deferred_output")
+
+        codecs: dict[str, Codec] = {
+            "s": _FakeCodec(call_fn=lambda d, pm: [("bytes", ref)]),
+        }
+        prepared = _prepared(pipeline, codecs=codecs)
+        result = run(prepared, {"bytes": b"input"})
+
+        assert isinstance(result["bytes"], bytes)
+        assert result["bytes"] == b"deferred_output"
+
+    def test_inverted_ref_materialized_in_final_output(self) -> None:
+        """CoreWasmRef is materialized in inverted execution final output."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        ref = self._make_ref(b"inverted_deferred")
+
+        codecs: dict[str, Codec] = {
+            "s": _FakeCodec(call_fn=lambda d, pm: [("bytes", ref)]),
+        }
+        prepared = _prepared(pipeline, direction="encode", codecs=codecs)
+        result = run(prepared, {"bytes": b"input"})
+
+        assert isinstance(result["bytes"], bytes)
+        assert result["bytes"] == b"inverted_deferred"
+
+
+@CORE_CODEC_REQUIRED
+class TestSingleCopyCorePipeline:
+    """Integration tests for single-copy transfer between core wasm steps.
+
+    Requires the identity-core-c codec to be built.
+    """
+
+    def test_two_step_core_returns_correct_output(self, raw_chunk: bytes) -> None:
+        """Two sequential core identity steps produce correct output via
+        deferred CoreWasmRef values in the value store."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "encode", resolver=resolver)
+
+        assert isinstance(prepared.codecs["step_a"], CoreWasmCodec)
+        assert isinstance(prepared.codecs["step_b"], CoreWasmCodec)
+
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
+
+    def test_core_codec_call_returns_core_wasm_ref(self, raw_chunk: bytes) -> None:
+        """CoreWasmCodec.call() returns CoreWasmRef entries (lazy output)."""
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        codec = resolver.resolve("identity")
+        assert isinstance(codec, CoreWasmCodec)
+
+        output = codec.call("encode", [("bytes", raw_chunk)])
+        assert len(output) == 1
+        name, value = output[0]
+        assert name == "bytes"
+        assert isinstance(value, CoreWasmRef)
+        assert value.materialize() == raw_chunk
+
+    def test_three_step_core_chain(self, raw_chunk: bytes) -> None:
+        """Three sequential core identity steps chain correctly."""
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_c.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+                "step_c": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "step_b.bytes"},
+                },
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "encode", resolver=resolver)
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
+
+    def test_inverted_two_step_core_pipeline(self, raw_chunk: bytes) -> None:
+        """Inverted execution through two core identity steps works correctly."""
+        data = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        resolver = Resolver(paths={"identity": _CORE_IDENTITY_WASM})
+        prepared = prepare(data, "encode", resolver=resolver)
+        result = run(prepared, {"bytes": raw_chunk})
+        assert result["bytes"] == raw_chunk
diff --git a/tests/test_native_codec.py b/tests/test_native_codec.py
new file mode 100644
index 0000000..257002d
--- /dev/null
+++ b/tests/test_native_codec.py
@@ -0,0 +1,560 @@
+"""Tests for NativeCodec and native codec integration."""
+
+from pathlib import Path
+from typing import Any
+from unittest.mock import patch
+
+import pytest
+
+from chonkle.codecs._base import Backend, Codec, PortMap
+from chonkle.codecs.native import NativeCodec
+from chonkle.executor import run
+from chonkle.pipeline import Direction, Pipeline, PreparedPipeline
+from chonkle.resolver import Resolver
+from chonkle.wasm_signature import Signature
+
+numcodecs = pytest.importorskip("numcodecs")
+np = pytest.importorskip("numpy")
+
+
+class TestNativeCodecInstantiation:
+    def test_loads_signature_from_bundled_file(self) -> None:
+        codec = NativeCodec("zlib")
+        sig = codec.signature()
+        assert sig.codec_id == "zlib"
+        assert "bytes" in sig.inputs
+
+    def test_codec_type_is_native(self) -> None:
+        codec = NativeCodec("zlib")
+        assert codec.codec_type == "native"
+
+    def test_codec_id_and_implementation(self) -> None:
+        codec = NativeCodec("zlib")
+        assert codec.codec_id == "zlib"
+        assert codec.implementation == "numcodecs.zlib"
+
+    def test_missing_signature_raises(self) -> None:
+        with pytest.raises(ValueError, match="No native codec signature"):
+            NativeCodec("nonexistent-codec-xyz")
+
+    def test_numcodecs_not_installed_raises(self) -> None:
+        with (
+            patch.dict("sys.modules", {"numcodecs": None}),
+            pytest.raises(ImportError, match="numcodecs is required"),
+        ):
+            NativeCodec("zlib")
+
+
+class TestNativeCodecBytesFormat:
+    """Test bytes-to-bytes native codecs (zlib, gzip, bz2, etc.)."""
+
+    def test_zlib_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("zlib")
+        data = bytes(range(256)) * 64
+        encoded = codec.call("encode", [("bytes", data)])
+        assert len(encoded) == 1
+        assert encoded[0][0] == "bytes"
+        assert encoded[0][1] != data  # compressed
+
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_gzip_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("gzip")
+        data = bytes(range(256)) * 64
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_bz2_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("bz2")
+        data = bytes(range(256)) * 64
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_lzma_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("lzma")
+        data = bytes(range(256)) * 64
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_encode_with_level_parameter(self) -> None:
+        codec = NativeCodec("zlib")
+        data = bytes(range(256)) * 64
+        # level is JSON-encoded in the port-map
+        encoded = codec.call("encode", [("bytes", data), ("level", b"9")])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_adler32_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("adler32")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_base64_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("base64")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_crc32_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("crc32")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_crc32c_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("crc32c")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_fletcher32_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("fletcher32")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_jenkins_lookup3_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("jenkins_lookup3")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_jenkins_lookup3_with_initval(self) -> None:
+        codec = NativeCodec("jenkins_lookup3")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data), ("initval", b"42")])
+        decoded = codec.call("decode", [("bytes", encoded[0][1]), ("initval", b"42")])
+        assert decoded[0][1] == data
+
+    def test_pickle_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("pickle")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data)])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+    def test_pickle_with_protocol(self) -> None:
+        codec = NativeCodec("pickle")
+        data = bytes(range(256)) * 4
+        encoded = codec.call("encode", [("bytes", data), ("protocol", b"4")])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        assert decoded[0][1] == data
+
+
+class TestNativeCodecNdarrayFormat:
+    """Test ndarray-format native codecs (delta, shuffle)."""
+
+    def test_delta_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("delta")
+        arr = np.arange(100, dtype="<i4")
+        data = arr.tobytes()
+        encoded = codec.call("encode", [("bytes", data), ("dtype", b'"<i4"')])
+        decoded = codec.call("decode", [("bytes", encoded[0][1]), ("dtype", b'"<i4"')])
+        result = np.frombuffer(decoded[0][1], dtype="<i4")
+        np.testing.assert_array_equal(result, arr)
+
+    def test_shuffle_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("shuffle")
+        arr = np.arange(100, dtype="<f4")
+        data = arr.tobytes()
+        encoded = codec.call("encode", [("bytes", data), ("dtype", b'"<f4"')])
+        decoded = codec.call("decode", [("bytes", encoded[0][1]), ("dtype", b'"<f4"')])
+        result = np.frombuffer(decoded[0][1], dtype="<f4")
+        np.testing.assert_array_equal(result, arr)
+
+    def test_bitround_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("bitround")
+        arr = np.array([1.23456789, 2.3456789, 3.456789], dtype="<f4")
+        data = arr.tobytes()
+        encoded = codec.call(
+            "encode", [("bytes", data), ("keepbits", b"10"), ("dtype", b'"<f4"')]
+        )
+        decoded = codec.call(
+            "decode",
+            [("bytes", encoded[0][1]), ("keepbits", b"10"), ("dtype", b'"<f4"')],
+        )
+        result = np.frombuffer(decoded[0][1], dtype="<f4")
+        np.testing.assert_allclose(result, arr, atol=0.01)
+
+    def test_fixedscaleoffset_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("fixedscaleoffset")
+        arr = np.array([10.0, 20.0, 30.0], dtype="<f4")
+        data = arr.tobytes()
+        encoded = codec.call(
+            "encode",
+            [("bytes", data), ("offset", b"0"), ("scale", b"10"), ("dtype", b'"<f4"')],
+        )
+        decoded = codec.call(
+            "decode",
+            [
+                ("bytes", encoded[0][1]),
+                ("offset", b"0"),
+                ("scale", b"10"),
+                ("dtype", b'"<f4"'),
+            ],
+        )
+        result = np.frombuffer(decoded[0][1], dtype="<f4")
+        np.testing.assert_allclose(result, arr)
+
+    def test_quantize_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("quantize")
+        arr = np.array([1.23456789, 2.3456789, 3.456789], dtype="<f4")
+        data = arr.tobytes()
+        encoded = codec.call(
+            "encode", [("bytes", data), ("digits", b"3"), ("dtype", b'"<f4"')]
+        )
+        decoded = codec.call(
+            "decode", [("bytes", encoded[0][1]), ("digits", b"3"), ("dtype", b'"<f4"')]
+        )
+        result = np.frombuffer(decoded[0][1], dtype="<f4")
+        np.testing.assert_allclose(result, arr, atol=0.01)
+
+    def test_astype_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("astype")
+        arr = np.array([1.5, 2.5, 3.5], dtype="<f8")
+        data = arr.tobytes()
+        # encode_dtype = output of encode (f4), decode_dtype = output of decode (f8)
+        encoded = codec.call(
+            "encode",
+            [("bytes", data), ("encode_dtype", b'"<f4"'), ("decode_dtype", b'"<f8"')],
+        )
+        assert len(encoded[0][1]) == 3 * 4  # 3 float32 values
+        decoded = codec.call(
+            "decode",
+            [
+                ("bytes", encoded[0][1]),
+                ("encode_dtype", b'"<f4"'),
+                ("decode_dtype", b'"<f8"'),
+            ],
+        )
+        result = np.frombuffer(decoded[0][1], dtype="<f8")
+        np.testing.assert_allclose(result, arr, atol=0.01)
+
+    def test_packbits_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("packbits")
+        arr = np.array(
+            [True, False, True, True, False, True, False, False], dtype="bool"
+        )
+        data = arr.tobytes()
+        encoded = codec.call("encode", [("bytes", data), ("encode_dtype", b'"bool"')])
+        decoded = codec.call(
+            "decode", [("bytes", encoded[0][1]), ("decode_dtype", b'"uint8"')]
+        )
+        result = np.frombuffer(decoded[0][1], dtype="bool")
+        np.testing.assert_array_equal(result, arr)
+
+    def test_categorize_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("categorize")
+        arr = np.array(["cat", "dog", "bird", "cat"], dtype="<U4")
+        data = arr.tobytes()
+        labels = b'["cat", "dog", "bird"]'
+        encoded = codec.call(
+            "encode",
+            [
+                ("bytes", data),
+                ("labels", labels),
+                ("dtype", b'"<U4"'),
+                ("astype", b'"uint8"'),
+            ],
+        )
+        decoded = codec.call(
+            "decode",
+            [
+                ("bytes", encoded[0][1]),
+                ("labels", labels),
+                ("dtype", b'"<U4"'),
+                ("astype", b'"uint8"'),
+            ],
+        )
+        result = np.frombuffer(decoded[0][1], dtype="<U4")
+        np.testing.assert_array_equal(result, arr)
+
+    def test_json2_encode_decode_roundtrip(self) -> None:
+        codec = NativeCodec("json2")
+        arr = np.array([1.0, 2.0, 3.0], dtype="<f4")
+        data = arr.tobytes()
+        encoded = codec.call("encode", [("bytes", data), ("dtype", b'"<f4"')])
+        decoded = codec.call("decode", [("bytes", encoded[0][1])])
+        result = np.frombuffer(decoded[0][1], dtype="<f4")
+        np.testing.assert_array_equal(result, arr)
+
+    def test_ndarray_missing_dtype_raises(self) -> None:
+        codec = NativeCodec("delta")
+        data = bytes(range(16))
+        with pytest.raises((ValueError, TypeError)):
+            codec.call("encode", [("bytes", data)])
+
+
+# -- Fake codec for mixed pipeline tests --
+
+
+_DEFAULT_SIG = Signature.from_dict(
+    {
+        "codec_id": "fake",
+        "inputs": {"bytes": {"type": "bytes"}},
+        "outputs": {"bytes": {"type": "bytes"}},
+    }
+)
+
+
+class _FakeCodec(Codec):
+    """Test double for wiring tests."""
+
+    def __init__(self, call_fn: Any = None, sig: Signature | None = None) -> None:
+        self._call_fn = call_fn or (lambda d, pm: [("bytes", b"output")])
+        self._sig = sig or _DEFAULT_SIG
+
+    @property
+    def codec_type(self) -> Backend:
+        return Backend.COMPONENT
+
+    @property
+    def codec_id(self) -> str:
+        return self._sig.codec_id or "fake"
+
+    @property
+    def implementation(self) -> str:
+        return self._sig.implementation or "fake"
+
+    def signature(self) -> Signature:
+        return self._sig
+
+    def call(self, direction: Direction, port_map: PortMap) -> PortMap:
+        return self._call_fn(direction, port_map)
+
+
+def _make_prepared(
+    pipeline: Pipeline,
+    direction: Direction,
+    codecs: dict[str, Codec],
+) -> PreparedPipeline:
+    """Build a PreparedPipeline computing encode/decode_only_inputs and output_ports."""
+    return PreparedPipeline(
+        pipeline=pipeline,
+        direction=direction,
+        codecs=codecs,
+        encode_only_inputs={
+            name: frozenset(codecs[name].signature().encode_only_inputs())
+            for name in pipeline.steps
+        },
+        decode_only_inputs={
+            name: frozenset(codecs[name].signature().decode_only_inputs())
+            for name in pipeline.steps
+        },
+        output_ports={
+            name: tuple(codecs[name].signature().outputs.keys())
+            for name in pipeline.steps
+        },
+    )
+
+
+class TestNativePipeline:
+    """Test native codecs running in a DAG pipeline via the executor."""
+
+    def test_single_native_step_encode_decode(self) -> None:
+        """Single native zlib step round-trips correctly."""
+        pipeline_dict = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "compress.bytes"},
+            "steps": {
+                "compress": {
+                    "codec_id": "zlib",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(pipeline_dict)
+        codec = NativeCodec("zlib")
+        prepared = _make_prepared(pipeline, "encode", {"compress": codec})
+        data = bytes(range(256)) * 64
+        result = run(prepared, {"bytes": data})
+        assert result["bytes"] != data
+
+        # Decode direction
+        pipeline_dec = Pipeline.parse({**pipeline_dict, "direction": "decode"})
+        prepared_dec = _make_prepared(
+            pipeline_dec, "decode", {"compress": NativeCodec("zlib")}
+        )
+        decoded = run(prepared_dec, {"bytes": result["bytes"]})
+        assert decoded["bytes"] == data
+
+    def test_two_native_steps_chained(self) -> None:
+        """Two native steps (gzip then bz2) chain correctly."""
+        pipeline_dict = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "gzip",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "bz2",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(pipeline_dict)
+        prepared = _make_prepared(
+            pipeline,
+            "encode",
+            {"step_a": NativeCodec("gzip"), "step_b": NativeCodec("bz2")},
+        )
+        data = bytes(range(256)) * 64
+        result = run(prepared, {"bytes": data})
+        assert result["bytes"] != data
+
+    def test_native_with_constant_parameter(self) -> None:
+        """Native codec receives constant parameter from pipeline."""
+        pipeline_dict = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"level": {"type": "int", "value": 9}},
+            "outputs": {"bytes": "compress.bytes"},
+            "steps": {
+                "compress": {
+                    "codec_id": "zlib",
+                    "inputs": {
+                        "bytes": "input.bytes",
+                        "level": "constant.level",
+                    },
+                }
+            },
+        }
+        pipeline = Pipeline.parse(pipeline_dict)
+        codec = NativeCodec("zlib")
+        prepared = _make_prepared(pipeline, "encode", {"compress": codec})
+        data = bytes(range(256)) * 64
+        result = run(prepared, {"bytes": data})
+        assert result["bytes"] != data
+
+    def test_mixed_native_and_fake_wasm_pipeline(self) -> None:
+        """Native codec followed by a fake wasm codec in a pipeline."""
+        pipeline_dict = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "step_b.bytes"},
+            "steps": {
+                "step_a": {
+                    "codec_id": "zlib",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "step_b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "step_a.bytes"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(pipeline_dict)
+        codecs: dict[str, Codec] = {
+            "step_a": NativeCodec("zlib"),
+            "step_b": _FakeCodec(call_fn=lambda d, pm: pm),
+        }
+        prepared = _make_prepared(pipeline, "encode", codecs)
+        data = bytes(range(256)) * 64
+        result = run(prepared, {"bytes": data})
+        # step_b is identity (returns input), so output is zlib-compressed
+        assert result["bytes"] != data
+        assert len(result["bytes"]) < len(data)
+
+    def test_encode_only_inputs_skipped_during_decode(self) -> None:
+        """Encode-only parameters (level) are not passed during decode."""
+        pipeline_dict = {
+            "codec_id": "test",
+            "direction": "decode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {"level": {"type": "int", "value": 9}},
+            "outputs": {"bytes": "compress.bytes"},
+            "steps": {
+                "compress": {
+                    "codec_id": "zlib",
+                    "inputs": {
+                        "bytes": "input.bytes",
+                        "level": "constant.level",
+                    },
+                }
+            },
+        }
+        # First encode some data
+        encode_codec = NativeCodec("zlib")
+        data = bytes(range(256)) * 64
+        encoded = encode_codec.call("encode", [("bytes", data)])
+
+        pipeline = Pipeline.parse(pipeline_dict)
+        prepared = _make_prepared(pipeline, "decode", {"compress": NativeCodec("zlib")})
+        result = run(prepared, {"bytes": encoded[0][1]})
+        assert result["bytes"] == data
+
+
+class TestResolverNativeIntegration:
+    """Test that the Resolver discovers and instantiates native codecs."""
+
+    def test_resolver_finds_native_codec(self) -> None:
+        """Resolver returns NativeCodec when preference includes native."""
+        resolver = Resolver(
+            codec_store=Path("/nonexistent"),
+            preference=["native", "core", "component"],
+        )
+        codec = resolver.resolve("zlib")
+        assert isinstance(codec, NativeCodec)
+        assert codec.codec_type == "native"
+
+    def test_resolver_default_preference_selects_native(self) -> None:
+        """With default preference, native is preferred over wasm."""
+        resolver = Resolver(codec_store=Path("/nonexistent"))
+        codec = resolver.resolve("zlib")
+        assert isinstance(codec, NativeCodec)
+
+    def test_resolver_override_selects_native(self) -> None:
+        """Per-codec override can select native implementation."""
+        resolver = Resolver(
+            codec_store=Path("/nonexistent"),
+            overrides={"zlib": "numcodecs.zlib"},
+        )
+        codec = resolver.resolve("zlib")
+        assert isinstance(codec, NativeCodec)
+
+    def test_resolver_list_includes_native(self) -> None:
+        """list_codecs() includes native codec entries."""
+        resolver = Resolver(codec_store=Path("/nonexistent"))
+        entries = resolver.list_codecs()
+        native_entries = [e for e in entries if e.backend == "native"]
+        assert len(native_entries) > 0
+        codec_ids = {e.codec_id for e in native_entries}
+        assert "zlib" in codec_ids
+        assert "gzip" in codec_ids
+
+    def test_preference_mismatch_raises(self) -> None:
+        """Preference list that excludes available backends raises ValueError."""
+        resolver = Resolver(
+            codec_store=Path("/nonexistent"),
+            preference=["component"],
+        )
+        with pytest.raises(ValueError, match=r"No implementation.*matches preference"):
+            resolver.resolve("zlib")
+
+    def test_resolver_unknown_native_raises(self) -> None:
+        """Resolver raises for unknown codec with no wasm and no native."""
+        resolver = Resolver(codec_store=Path("/nonexistent"))
+        with pytest.raises(ValueError, match="No codec implementation found"):
+            resolver.resolve("totally-unknown-codec-xyz")
diff --git a/tests/test_pipeline.py b/tests/test_pipeline.py
index 3a7c17e..8416260 100644
--- a/tests/test_pipeline.py
+++ b/tests/test_pipeline.py
@@ -1,94 +1,373 @@
-import json
-from pathlib import Path
+"""Tests for DAG pipeline parsing and wiring validation."""
 
-import numpy as np
 import pytest
 
-from chonkle.pipeline import decode, encode, get_codecs
+from chonkle.pipeline import ConstantDescriptor, Pipeline, WiringRef
 
-from .conftest import CHUNK_PATHS, FIXTURES_DIR
 
+class TestParseRef:
+    def test_input_ref(self) -> None:
+        ref = WiringRef.parse("input.bytes")
+        assert ref.kind == "input"
+        assert ref.source == "input"
+        assert ref.port == "bytes"
 
-def _decode_fixture(chunk_path: Path) -> np.ndarray:
-    """Load a test fixture: read chunk bytes + sidecar JSON, decode."""
-    codecs = get_codecs(Path(str(chunk_path) + ".json"))
-    return decode(chunk_path.read_bytes(), codecs)
+    def test_constant_ref(self) -> None:
+        ref = WiringRef.parse("constant.level")
+        assert ref.kind == "constant"
+        assert ref.source == "constant"
+        assert ref.port == "level"
 
+    def test_step_ref(self) -> None:
+        ref = WiringRef.parse("my_step.out_bytes")
+        assert ref.kind == "step"
+        assert ref.source == "my_step"
+        assert ref.port == "out_bytes"
 
-class TestGetCodecs:
-    def test_from_path(self) -> None:
-        codecs = get_codecs(FIXTURES_DIR / "cog" / "0.json")
-        assert isinstance(codecs, list)
-        assert len(codecs) > 0
+    def test_step_ref_returns_wiringref(self) -> None:
+        ref = WiringRef.parse("zstd.bytes")
+        assert isinstance(ref, WiringRef)
 
-    def test_from_dict(self) -> None:
-        spec = {"codecs": [{"name": "zlib", "type": "numcodecs"}]}
-        codecs = get_codecs(spec)
-        assert codecs == [{"name": "zlib", "type": "numcodecs"}]
+    def test_missing_dot_raises(self) -> None:
+        with pytest.raises(ValueError, match="Invalid wiring reference"):
+            WiringRef.parse("noport")
 
-    def test_missing_codecs_key_raises(self) -> None:
-        with pytest.raises(KeyError, match="codecs"):
-            get_codecs({"not_codecs": []})
+    def test_empty_string_raises(self) -> None:
+        with pytest.raises(ValueError, match="Invalid wiring reference"):
+            WiringRef.parse("")
 
 
-class TestDecode:
-    def test_all_chunks_decode_identically(self) -> None:
-        arrays = [_decode_fixture(p) for p in CHUNK_PATHS]
-        for arr in arrays[1:]:
-            np.testing.assert_array_equal(arrays[0], arr)
+class TestParsePipeline:
+    def test_linear_pipeline_from_fixture(self, cog_decode_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(cog_decode_pipeline_json)
+        assert isinstance(pipeline, Pipeline)
+        assert pipeline.direction == "decode"
+        assert list(pipeline.inputs.keys()) == ["bytes"]
+        assert len(pipeline.steps) == 2
+        assert list(pipeline.steps) == ["zlib", "predictor2"]
 
+    def test_linear_pipeline_codec_id(self, cog_decode_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(cog_decode_pipeline_json)
+        assert pipeline.codec_id == "cog-zlib-predictor2"
 
-class TestDecodeWasm:
-    pytestmark = pytest.mark.usefixtures("ensure_core_wasm")
+    def test_linear_pipeline_step_fields(self, cog_decode_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(cog_decode_pipeline_json)
+        step = pipeline.steps["zlib"]
+        assert {k: str(v) for k, v in step.inputs.items()} == {
+            "bytes": "input.bytes",
+            "level": "constant.level",
+        }
 
-    def test_wasm_decode_matches_native(self, tmp_path: Path) -> None:
-        """Wasm TiffPredictor2 produces identical output to native."""
-        native = _decode_fixture(FIXTURES_DIR / "cog" / "0")
+    def test_dag_pipeline_from_fixture(self, page_split_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(page_split_pipeline_json)
+        assert pipeline.direction == "encode"
+        assert len(pipeline.steps) == 4  # page_split + 3 identity steps
 
-        # Build a temporary fixture with a fully-qualified file:// URI
-        # since the metadata JSON cannot contain a portable absolute path.
-        src = FIXTURES_DIR / "cog_wasm"
-        chunk = tmp_path / "0"
-        chunk.write_bytes((src / "0").read_bytes())
+    def test_dag_pipeline_shared_codec_id(self, page_split_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(page_split_pipeline_json)
+        identity_steps = [
+            s for s in pipeline.steps.values() if s.codec_id == "identity"
+        ]
+        assert len(identity_steps) == 3
+        assert {s.name for s in identity_steps} == {
+            "identity_rep",
+            "identity_def",
+            "identity_data",
+        }
 
-        wasm_uri = (src / "tiff-predictor-2-c.wasm").resolve().as_uri()
-        metadata = json.loads((src / "0.json").read_text())
-        wasm_codec = next(c for c in metadata["codecs"] if c["type"] == "wasm")
-        wasm_codec["uri"] = wasm_uri
-        (tmp_path / "0.json").write_text(json.dumps(metadata))
+    def test_dag_execution_order(self, page_split_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(page_split_pipeline_json)
+        order = list(pipeline.steps)
+        page_split_idx = order.index("page_split")
+        for name in ("identity_rep", "identity_def", "identity_data"):
+            assert page_split_idx < order.index(name), f"page_split must precede {name}"
 
-        wasm_result = _decode_fixture(chunk)
-        np.testing.assert_array_equal(native, wasm_result)
+    def test_dag_pipeline_constants(self, page_split_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(page_split_pipeline_json)
+        assert pipeline.constants == {
+            "rep_length": ConstantDescriptor(type="uint", value=128),
+            "def_length": ConstantDescriptor(type="uint", value=256),
+        }
 
+    def test_dag_pipeline_outputs(self, page_split_pipeline_json: dict) -> None:
+        pipeline = Pipeline.parse(page_split_pipeline_json)
+        assert set(pipeline.outputs.keys()) == {"rep_levels", "def_levels", "data"}
 
-class TestEncode:
-    def test_encode_then_decode_roundtrip(self) -> None:
-        """Encoding then decoding through the same pipeline returns the original."""
-        chunk_path = FIXTURES_DIR / "cog" / "0"
-        codec_specs = get_codecs(Path(str(chunk_path) + ".json"))
+    def test_sources_parsed(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "sources": {"zlib": "oci://ghcr.io/example/zlib:v1"},
+            "outputs": {},
+            "steps": {},
+        }
+        pipeline = Pipeline.parse(data)
+        assert pipeline.sources == {"zlib": "oci://ghcr.io/example/zlib:v1"}
 
-        original = _decode_fixture(chunk_path)
-        encoded = encode(original, codec_specs)
-        decoded = decode(encoded, codec_specs)
+    def test_sources_default_empty(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {},
+            "outputs": {},
+            "steps": {},
+        }
+        pipeline = Pipeline.parse(data)
+        assert pipeline.sources == {}
 
-        np.testing.assert_array_equal(decoded, original)
+    def test_missing_direction_raises(self) -> None:
+        with pytest.raises(ValueError, match="direction"):
+            Pipeline.parse({"codec_id": "t", "inputs": {}, "steps": {}})
 
-    def test_encode_produces_bytes(self) -> None:
-        """Encoding should produce a bytes object."""
-        chunk_path = FIXTURES_DIR / "zarr_zstd" / "0"
-        codec_specs = get_codecs(Path(str(chunk_path) + ".json"))
+    def test_invalid_direction_raises(self) -> None:
+        with pytest.raises(ValueError, match="direction"):
+            Pipeline.parse(
+                {"codec_id": "t", "direction": "transform", "inputs": {}, "steps": {}}
+            )
 
-        original = _decode_fixture(chunk_path)
-        encoded = encode(original, codec_specs)
+    def test_missing_codec_id_raises(self) -> None:
+        with pytest.raises(ValueError, match="codec_id"):
+            Pipeline.parse(
+                {
+                    "direction": "encode",
+                    "inputs": {"bytes": {"type": "bytes"}},
+                    "constants": {},
+                    "outputs": {"bytes": "s.bytes"},
+                    "steps": {
+                        "s": {
+                            "codec_id": "some-codec",
+                            "inputs": {"bytes": "input.bytes"},
+                        }
+                    },
+                }
+            )
 
-        assert isinstance(encoded, bytes)
-        assert len(encoded) > 0
+    def test_parse_accepts_dict(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"bytes": "s.bytes"},
+            "steps": {
+                "s": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.bytes"},
+                }
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        assert pipeline.direction == "encode"
 
-    def test_encode_pipeline_not_producing_bytes_raises(self) -> None:
-        """A pipeline that doesn't end with bytes should raise TypeError."""
-        arr = np.array([[1, 2], [3, 4]], dtype=np.uint16)
-        codec_specs = [
-            {"name": "tiff_predictor_2", "type": "numcodecs", "configuration": {}},
-        ]
-        with pytest.raises(TypeError, match="did not produce bytes"):
-            encode(arr, codec_specs)
+
+class TestWiringValidation:
+    def _base(self) -> dict:
+        return {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {},
+            "steps": {},
+        }
+
+    def test_valid_input_ref(self) -> None:
+        data = self._base()
+        data["steps"] = {
+            "s": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "input.bytes"},
+            }
+        }
+        data["outputs"] = {"bytes": "s.bytes"}
+        pipeline = Pipeline.parse(data)
+        assert list(pipeline.steps) == ["s"]
+
+    def test_valid_constant_ref(self) -> None:
+        data = self._base()
+        data["constants"] = {"level": {"type": "int", "value": 3}}
+        data["steps"] = {
+            "s": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "input.bytes", "level": "constant.level"},
+            }
+        }
+        pipeline = Pipeline.parse(data)
+        assert len(pipeline.steps) == 1
+
+    def test_valid_step_to_step_ref(self) -> None:
+        data = self._base()
+        data["steps"] = {
+            "a": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "input.bytes"},
+            },
+            "b": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "a.bytes"},
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        assert list(pipeline.steps).index("a") < list(pipeline.steps).index("b")
+
+    def test_same_codec_id_different_names_valid(self) -> None:
+        """The same codec may appear multiple times with distinct step names."""
+        data = self._base()
+        data["steps"] = {
+            "proc_a": {
+                "codec_id": "identity",
+                "inputs": {"bytes": "input.bytes"},
+            },
+            "proc_b": {
+                "codec_id": "identity",
+                "inputs": {"bytes": "input.bytes"},
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        assert set(pipeline.steps) == {"proc_a", "proc_b"}
+        assert all(s.codec_id == "identity" for s in pipeline.steps.values())
+
+    def test_undefined_input_raises(self) -> None:
+        data = self._base()
+        data["steps"] = {
+            "s": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "input.missing"},
+            }
+        }
+        with pytest.raises(ValueError, match="input 'missing' is not declared"):
+            Pipeline.parse(data)
+
+    def test_undefined_constant_raises(self) -> None:
+        data = self._base()
+        data["steps"] = {
+            "s": {
+                "codec_id": "some-codec",
+                "inputs": {"n": "constant.missing"},
+            }
+        }
+        with pytest.raises(ValueError, match="constant 'missing' is not declared"):
+            Pipeline.parse(data)
+
+    def test_undefined_step_raises(self) -> None:
+        data = self._base()
+        data["steps"] = {
+            "s": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "ghost.bytes"},
+            }
+        }
+        with pytest.raises(ValueError, match="step 'ghost' does not exist"):
+            Pipeline.parse(data)
+
+    def test_pipeline_output_input_passthrough(self) -> None:
+        """A pipeline output may reference a pipeline input directly (passthrough)."""
+        data = self._base()
+        data["outputs"] = {"bytes": "input.bytes"}
+        pipeline = Pipeline.parse(data)
+        assert {k: str(v) for k, v in pipeline.outputs.items()} == {
+            "bytes": "input.bytes",
+        }
+
+    def test_pipeline_output_constant_passthrough(self) -> None:
+        """A pipeline output may reference a constant directly."""
+        data = self._base()
+        data["constants"] = {"level": {"type": "int", "value": 3}}
+        data["outputs"] = {"level": "constant.level"}
+        pipeline = Pipeline.parse(data)
+        assert {k: str(v) for k, v in pipeline.outputs.items()} == {
+            "level": "constant.level",
+        }
+
+    def test_cycle_detection(self) -> None:
+        data = self._base()
+        data["steps"] = {
+            "a": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "b.bytes"},
+            },
+            "b": {
+                "codec_id": "some-codec",
+                "inputs": {"bytes": "a.bytes"},
+            },
+        }
+        with pytest.raises(ValueError, match="cycle"):
+            Pipeline.parse(data)
+
+
+class TestTopologicalOrder:
+    def test_linear_order_regardless_of_declaration_order(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"x": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {"x": "c.x"},
+            "steps": {
+                "c": {
+                    "codec_id": "some-codec",
+                    "inputs": {"x": "b.x"},
+                },
+                "b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"x": "a.x"},
+                },
+                "a": {
+                    "codec_id": "some-codec",
+                    "inputs": {"x": "input.x"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        order = list(pipeline.steps)
+        assert order.index("a") < order.index("b")
+        assert order.index("b") < order.index("c")
+
+    def test_fan_out_order(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"bytes": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {},
+            "steps": {
+                "split": {
+                    "codec_id": "page-split",
+                    "inputs": {"bytes": "input.bytes"},
+                },
+                "proc_a": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "split.a"},
+                },
+                "proc_b": {
+                    "codec_id": "identity",
+                    "inputs": {"bytes": "split.b"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        order = list(pipeline.steps)
+        assert order.index("split") < order.index("proc_a")
+        assert order.index("split") < order.index("proc_b")
+
+    def test_independent_steps_all_appear(self) -> None:
+        data = {
+            "codec_id": "test",
+            "direction": "encode",
+            "inputs": {"a": {"type": "bytes"}, "b": {"type": "bytes"}},
+            "constants": {},
+            "outputs": {},
+            "steps": {
+                "step_a": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.a"},
+                },
+                "step_b": {
+                    "codec_id": "some-codec",
+                    "inputs": {"bytes": "input.b"},
+                },
+            },
+        }
+        pipeline = Pipeline.parse(data)
+        assert set(pipeline.steps) == {"step_a", "step_b"}
diff --git a/tests/test_wasm_download.py b/tests/test_wasm_download.py
index 3a12214..fea6ccb 100644
--- a/tests/test_wasm_download.py
+++ b/tests/test_wasm_download.py
@@ -1,15 +1,10 @@
 import io
-import tempfile
 from pathlib import Path
 from unittest.mock import MagicMock, patch
 
 import pytest
 
-from chonkle.wasm_download import (
-    download_https,
-    download_oci,
-    get_cache_dir,
-)
+from chonkle.wasm_download import resolve_uri
 
 URLOPEN = "chonkle.wasm_download.urllib.request.urlopen"
 ORAS_CLIENT = "oras.client.OrasClient"
@@ -23,65 +18,57 @@ def _fake_response(data: bytes) -> MagicMock:
     return resp
 
 
-class TestGetCacheDir:
-    def test_default_uses_tempdir(
-        self, monkeypatch: pytest.MonkeyPatch, tmp_path: Path
-    ) -> None:
-        monkeypatch.delenv("CHONKLE_CACHE_DIR", raising=False)
-        monkeypatch.setenv("TMPDIR", str(tmp_path))
-        monkeypatch.setattr(tempfile, "tempdir", None)
-        result = get_cache_dir()
-        assert result == tmp_path / "chonkle" / "wasm"
-        assert result.exists()
+class TestResolveUriFile:
+    def test_file_uri_yields_path(self, tmp_path: Path) -> None:
+        wasm = tmp_path / "codec.wasm"
+        wasm.write_bytes(b"\x00asm\x01\x00\x00\x00")
+        with resolve_uri(f"file://{wasm}") as path:
+            assert path == wasm
 
-    def test_env_var_override(
-        self, monkeypatch: pytest.MonkeyPatch, tmp_path: Path
-    ) -> None:
-        monkeypatch.setenv("CHONKLE_CACHE_DIR", str(tmp_path / "custom"))
-        result = get_cache_dir()
-        assert result == tmp_path / "custom"
-        assert result.exists()
+    def test_http_raises(self) -> None:
+        with (
+            pytest.raises(ValueError, match="HTTP is not supported"),
+            resolve_uri("http://example.com/codec.wasm"),
+        ):
+            pass
+
+    def test_unsupported_scheme_raises(self) -> None:
+        with (
+            pytest.raises(ValueError, match="Unsupported URI scheme"),
+            resolve_uri("ftp://example.com/codec.wasm"),
+        ):
+            pass
 
 
 class TestDownloadHttps:
-    def test_downloads_and_caches(self, tmp_path: Path) -> None:
+    def test_downloads_wasm(self) -> None:
         wasm_bytes = b"\x00asm\x01\x00\x00\x00"
         resp = _fake_response(wasm_bytes)
 
-        with patch(URLOPEN, return_value=resp):
-            path = download_https(
-                "https://example.com/releases/v1/codec.wasm",
-                cache_dir=tmp_path,
-            )
-
-        assert path.exists()
-        assert path.name == "codec.wasm"
-        assert path.read_bytes() == wasm_bytes
-
-    def test_cache_hit_skips_download(self, tmp_path: Path) -> None:
-        url = "https://example.com/releases/v1/codec.wasm"
-        resp = _fake_response(b"\x00asm\x01\x00\x00\x00")
-
-        with patch(URLOPEN, return_value=resp) as mock:
-            path1 = download_https(url, cache_dir=tmp_path)
-            path2 = download_https(url, cache_dir=tmp_path)
-
-        assert path1 == path2
-        mock.assert_called_once()
+        with (
+            patch(URLOPEN, return_value=resp),
+            resolve_uri("https://example.com/releases/v1/codec.wasm") as path,
+        ):
+            assert path.exists()
+            assert path.name == "codec.wasm"
+            assert path.read_bytes() == wasm_bytes
 
-    def test_force_redownloads(self, tmp_path: Path) -> None:
-        url = "https://example.com/releases/v1/codec.wasm"
-        resp = _fake_response(b"\x00asm\x01\x00\x00\x00")
+    def test_temp_dir_cleaned_up_after_exit(self) -> None:
+        wasm_bytes = b"\x00asm\x01\x00\x00\x00"
+        resp = _fake_response(wasm_bytes)
+        captured: list[Path] = []
 
-        with patch(URLOPEN, return_value=resp) as mock:
-            download_https(url, cache_dir=tmp_path)
-            download_https(url, cache_dir=tmp_path, force=True)
+        with (
+            patch(URLOPEN, return_value=resp),
+            resolve_uri("https://example.com/releases/v1/codec.wasm") as path,
+        ):
+            captured.append(path)
 
-        assert mock.call_count == 2
+        assert not captured[0].exists()
 
 
 class TestDownloadOci:
-    def test_full_flow(self, tmp_path: Path) -> None:
+    def test_full_flow(self) -> None:
         wasm_bytes = b"\x00asm\x01\x00\x00\x00"
 
         def fake_pull(*, target, outdir):
@@ -92,66 +79,27 @@ def fake_pull(*, target, outdir):
         mock_client = MagicMock()
         mock_client.pull.side_effect = fake_pull
 
-        with patch(ORAS_CLIENT, return_value=mock_client):
-            path = download_oci(
-                "oci://ghcr.io/org/repo:v1.0",
-                cache_dir=tmp_path,
-            )
-
-        assert path.exists()
-        assert path.read_bytes() == wasm_bytes
-        mock_client.pull.assert_called_once()
-
-    def test_cache_hit_skips_pull(self, tmp_path: Path) -> None:
-        # Pre-populate cache directory with a .wasm file.
-        ref_dir = tmp_path / "oci" / "ghcr.io" / "org" / "repo" / "v1.0"
-        ref_dir.mkdir(parents=True)
-        cached = ref_dir / "codec.wasm"
-        cached.write_bytes(b"\x00asm\x01\x00\x00\x00")
-
-        mock_client = MagicMock()
-
-        with patch(ORAS_CLIENT, return_value=mock_client):
-            path = download_oci(
-                "oci://ghcr.io/org/repo:v1.0",
-                cache_dir=tmp_path,
-            )
-
-        assert path == cached
-        mock_client.pull.assert_not_called()
-
-    def test_force_bypasses_cache(self, tmp_path: Path) -> None:
-        ref_dir = tmp_path / "oci" / "ghcr.io" / "org" / "repo" / "v1.0"
-        ref_dir.mkdir(parents=True)
-        cached = ref_dir / "codec.wasm"
-        cached.write_bytes(b"\x00asm\x01\x00\x00\x00")
-
-        mock_client = MagicMock()
-        mock_client.pull.return_value = [str(cached)]
-
-        with patch(ORAS_CLIENT, return_value=mock_client):
-            download_oci(
-                "oci://ghcr.io/org/repo:v1.0",
-                cache_dir=tmp_path,
-                force=True,
-            )
+        with (
+            patch(ORAS_CLIENT, return_value=mock_client),
+            resolve_uri("oci://ghcr.io/org/repo:v1.0") as path,
+        ):
+            assert path.exists()
+            assert path.read_bytes() == wasm_bytes
 
         mock_client.pull.assert_called_once()
 
-    def test_no_wasm_file_raises(self, tmp_path: Path) -> None:
-        txt_file = tmp_path / "oci" / "ghcr.io" / "org" / "repo" / "v1.0" / "readme.txt"
-        txt_file.parent.mkdir(parents=True)
-        txt_file.write_bytes(b"hello")
+    def test_no_wasm_file_raises(self) -> None:
+        def fake_pull(*, target, outdir):
+            txt = Path(outdir) / "readme.txt"
+            txt.write_bytes(b"hello")
+            return [str(txt)]
 
         mock_client = MagicMock()
-        mock_client.pull.return_value = [str(txt_file)]
+        mock_client.pull.side_effect = fake_pull
 
         with (
             patch(ORAS_CLIENT, return_value=mock_client),
             pytest.raises(ValueError, match=r"No \.wasm file found"),
+            resolve_uri("oci://ghcr.io/org/repo:v1.0"),
         ):
-            download_oci(
-                "oci://ghcr.io/org/repo:v1.0",
-                cache_dir=tmp_path,
-                force=True,
-            )
+            pass
diff --git a/tests/test_wasm_download_integration.py b/tests/test_wasm_download_integration.py
deleted file mode 100644
index 37da3ff..0000000
--- a/tests/test_wasm_download_integration.py
+++ /dev/null
@@ -1,62 +0,0 @@
-import numpy as np
-import pytest
-
-from chonkle.codecs import TiffPredictor2
-from chonkle.wasm_download import download_https, download_oci
-from chonkle.wasm_runner import resolve_wasm_uri, wasm_decode
-
-HTTPS_URL = (
-    "https://github.com/cylf-dev/tiff-predictor-2-c/"
-    "releases/download/v0.1.0/tiff-predictor-2.wasm"
-)
-OCI_URI = "oci://ghcr.io/cylf-dev/tiff-predictor-2-c:v0.1.0"
-
-WASM_MAGIC = b"\x00asm"
-
-
-@pytest.mark.network
-class TestHttpsDownloadIntegration:
-    def test_download_from_github_releases(self, tmp_path):
-        path = download_https(HTTPS_URL, cache_dir=tmp_path)
-        assert path.exists()
-        assert path.stat().st_size > 0
-        assert path.read_bytes()[:4] == WASM_MAGIC
-
-    def test_cache_hit_on_second_download(self, tmp_path):
-        path1 = download_https(HTTPS_URL, cache_dir=tmp_path)
-        path2 = download_https(HTTPS_URL, cache_dir=tmp_path)
-        assert path1 == path2
-
-
-@pytest.mark.network
-class TestOciDownloadIntegration:
-    def test_download_from_ghcr(self, tmp_path):
-        path = download_oci(OCI_URI, cache_dir=tmp_path)
-        assert path.exists()
-        assert path.stat().st_size > 0
-        assert path.read_bytes()[:4] == WASM_MAGIC
-
-
-@pytest.mark.network
-class TestResolveWasmUriIntegration:
-    def test_https_resolve_and_run(self, tmp_path, monkeypatch):
-        monkeypatch.setenv("CHONKLE_CACHE_DIR", str(tmp_path))
-        path = resolve_wasm_uri(HTTPS_URL)
-
-        codec = TiffPredictor2()
-        arr = np.array([[10, 12, 15, 11]], dtype=np.uint16)
-        encoded = codec.encode(arr).tobytes()
-        result_bytes = wasm_decode(path, encoded, {"bytes_per_sample": 2, "width": 4})
-        result = np.frombuffer(result_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_oci_resolve_and_run(self, tmp_path, monkeypatch):
-        monkeypatch.setenv("CHONKLE_CACHE_DIR", str(tmp_path))
-        path = resolve_wasm_uri(OCI_URI)
-
-        codec = TiffPredictor2()
-        arr = np.array([[10, 12, 15, 11]], dtype=np.uint16)
-        encoded = codec.encode(arr).tobytes()
-        result_bytes = wasm_decode(path, encoded, {"bytes_per_sample": 2, "width": 4})
-        result = np.frombuffer(result_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
diff --git a/tests/test_wasm_runner.py b/tests/test_wasm_runner.py
deleted file mode 100644
index bdafa67..0000000
--- a/tests/test_wasm_runner.py
+++ /dev/null
@@ -1,291 +0,0 @@
-from pathlib import Path
-
-import numpy as np
-import pytest
-
-from chonkle.codecs import TiffPredictor2
-from chonkle.wasm_runner import resolve_wasm_uri, wasm_decode, wasm_encode
-
-from .conftest import FIXTURES_DIR
-
-CORE_WASM_PATH = FIXTURES_DIR / "cog_wasm" / "tiff-predictor-2-c.wasm"
-COMPONENT_WASM_PATH = FIXTURES_DIR / "cog_wasm" / "tiff-predictor-2-python.wasm"
-
-
-class TestResolveWasmUri:
-    def test_file_uri(self, tmp_path: Path) -> None:
-        wasm = tmp_path / "codec.wasm"
-        wasm.write_bytes(b"\x00")
-        result = resolve_wasm_uri(wasm.as_uri())
-        assert result == wasm.resolve()
-
-    def test_no_scheme_raises(self) -> None:
-        with pytest.raises(NotImplementedError, match="must include a scheme"):
-            resolve_wasm_uri("/absolute/path.wasm")
-
-    def test_unsupported_scheme_raises(self) -> None:
-        with pytest.raises(NotImplementedError, match="Unsupported URI scheme"):
-            resolve_wasm_uri("ftp://registry/codec:v1")
-
-    def test_https_uri_delegates_to_download(
-        self, monkeypatch: pytest.MonkeyPatch, tmp_path: Path
-    ) -> None:
-        wasm = tmp_path / "downloaded.wasm"
-        wasm.write_bytes(b"\x00")
-        monkeypatch.setattr(
-            "chonkle.wasm_runner.download_https",
-            lambda url, **kw: wasm,
-        )
-        result = resolve_wasm_uri("https://example.com/codec.wasm")
-        assert result == wasm
-
-    def test_oci_uri_delegates_to_download(
-        self, monkeypatch: pytest.MonkeyPatch, tmp_path: Path
-    ) -> None:
-        wasm = tmp_path / "downloaded.wasm"
-        wasm.write_bytes(b"\x00")
-        monkeypatch.setattr(
-            "chonkle.wasm_runner.download_oci",
-            lambda uri, **kw: wasm,
-        )
-        result = resolve_wasm_uri("oci://ghcr.io/org/repo:v1")
-        assert result == wasm
-
-
-class TestWasmCoreDecode:
-    pytestmark = pytest.mark.usefixtures("ensure_core_wasm")
-
-    def test_small_uint16_array(self) -> None:
-        """Encode with Python, decode with Core Wasm, verify roundtrip."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        encoded = codec.encode(arr)
-        encoded_bytes = encoded.tobytes()
-
-        result_bytes = wasm_decode(
-            CORE_WASM_PATH,
-            encoded_bytes,
-            {"bytes_per_sample": 2, "width": 4},
-        )
-
-        result = np.frombuffer(result_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_small_uint8_array(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([[1, 3, 6, 10], [200, 198, 195, 199]], dtype=np.uint8)
-        encoded = codec.encode(arr)
-        encoded_bytes = encoded.tobytes()
-
-        result_bytes = wasm_decode(
-            CORE_WASM_PATH,
-            encoded_bytes,
-            {"bytes_per_sample": 1, "width": 4},
-        )
-
-        result = np.frombuffer(result_bytes, dtype=np.uint8).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
-
-
-class TestWasmCoreEncode:
-    pytestmark = pytest.mark.usefixtures("ensure_core_wasm")
-
-    def test_small_uint16_array(self) -> None:
-        """Encode with Core Wasm, decode with Python, verify roundtrip."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-
-        encoded_bytes = wasm_encode(
-            CORE_WASM_PATH,
-            arr.tobytes(),
-            {"bytes_per_sample": 2, "width": 4},
-        )
-
-        encoded = np.frombuffer(encoded_bytes, dtype=np.uint16).reshape(arr.shape)
-        result = codec.decode(encoded)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_small_uint8_array(self) -> None:
-        """Encode with Core Wasm, decode with Python, verify roundtrip."""
-        codec = TiffPredictor2()
-        arr = np.array([[1, 3, 6, 10], [200, 198, 195, 199]], dtype=np.uint8)
-
-        encoded_bytes = wasm_encode(
-            CORE_WASM_PATH,
-            arr.tobytes(),
-            {"bytes_per_sample": 1, "width": 4},
-        )
-
-        encoded = np.frombuffer(encoded_bytes, dtype=np.uint8).reshape(arr.shape)
-        result = codec.decode(encoded)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_core_encode_matches_python_encode(self) -> None:
-        """Core Wasm encode produces identical output to Python encode."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        python_encoded = codec.encode(arr)
-
-        core_encoded_bytes = wasm_encode(
-            CORE_WASM_PATH,
-            arr.tobytes(),
-            {"bytes_per_sample": 2, "width": 4},
-        )
-        core_encoded = np.frombuffer(core_encoded_bytes, dtype=np.uint16).reshape(
-            arr.shape
-        )
-
-        np.testing.assert_array_equal(core_encoded, python_encoded)
-
-
-class TestWasmCoreRoundtrip:
-    pytestmark = pytest.mark.usefixtures("ensure_core_wasm")
-
-    def test_encode_then_decode(self) -> None:
-        """Core Wasm encode followed by Core Wasm decode returns original data."""
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        config = {"bytes_per_sample": 2, "width": 4}
-
-        encoded_bytes = wasm_encode(CORE_WASM_PATH, arr.tobytes(), config)
-        decoded_bytes = wasm_decode(CORE_WASM_PATH, encoded_bytes, config)
-
-        result = np.frombuffer(decoded_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_decode_then_encode(self) -> None:
-        """Core Wasm decode followed by Core Wasm encode returns original data."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        differenced = codec.encode(arr)
-        config = {"bytes_per_sample": 2, "width": 4}
-
-        decoded_bytes = wasm_decode(CORE_WASM_PATH, differenced.tobytes(), config)
-        reencoded_bytes = wasm_encode(CORE_WASM_PATH, decoded_bytes, config)
-
-        result = np.frombuffer(reencoded_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, differenced)
-
-
-class TestWasmComponentDecode:
-    pytestmark = pytest.mark.usefixtures("ensure_component_wasm")
-
-    def test_small_uint16_array(self) -> None:
-        """Python encode → component decode → verify roundtrip."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        encoded = codec.encode(arr)
-
-        result_bytes = wasm_decode(
-            COMPONENT_WASM_PATH,
-            encoded.tobytes(),
-            {"bytes_per_sample": 2, "width": 4},
-        )
-
-        result = np.frombuffer(result_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_small_uint8_array(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([[1, 3, 6, 10], [200, 198, 195, 199]], dtype=np.uint8)
-        encoded = codec.encode(arr)
-
-        result_bytes = wasm_decode(
-            COMPONENT_WASM_PATH,
-            encoded.tobytes(),
-            {"bytes_per_sample": 1, "width": 4},
-        )
-
-        result = np.frombuffer(result_bytes, dtype=np.uint8).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
-
-
-class TestWasmComponentEncode:
-    pytestmark = pytest.mark.usefixtures("ensure_component_wasm")
-
-    def test_small_uint16_array(self) -> None:
-        """Component encode → Python decode → verify roundtrip."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-
-        encoded_bytes = wasm_encode(
-            COMPONENT_WASM_PATH,
-            arr.tobytes(),
-            {"bytes_per_sample": 2, "width": 4},
-        )
-
-        encoded = np.frombuffer(encoded_bytes, dtype=np.uint16).reshape(arr.shape)
-        result = codec.decode(encoded)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_small_uint8_array(self) -> None:
-        codec = TiffPredictor2()
-        arr = np.array([[1, 3, 6, 10], [200, 198, 195, 199]], dtype=np.uint8)
-
-        encoded_bytes = wasm_encode(
-            COMPONENT_WASM_PATH,
-            arr.tobytes(),
-            {"bytes_per_sample": 1, "width": 4},
-        )
-
-        encoded = np.frombuffer(encoded_bytes, dtype=np.uint8).reshape(arr.shape)
-        result = codec.decode(encoded)
-        np.testing.assert_array_equal(result, arr)
-
-    def test_component_encode_matches_python_encode(self) -> None:
-        """Component encode produces identical output to Python encode."""
-        codec = TiffPredictor2()
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        python_encoded = codec.encode(arr)
-
-        component_encoded_bytes = wasm_encode(
-            COMPONENT_WASM_PATH,
-            arr.tobytes(),
-            {"bytes_per_sample": 2, "width": 4},
-        )
-        component_encoded = np.frombuffer(
-            component_encoded_bytes, dtype=np.uint16
-        ).reshape(arr.shape)
-
-        np.testing.assert_array_equal(component_encoded, python_encoded)
-
-
-class TestWasmComponentRoundtrip:
-    pytestmark = pytest.mark.usefixtures("ensure_component_wasm")
-
-    def test_encode_then_decode(self) -> None:
-        """Component encode followed by component decode returns original data."""
-        arr = np.array(
-            [[10, 12, 15, 11], [100, 98, 95, 99]],
-            dtype=np.uint16,
-        )
-        config = {"bytes_per_sample": 2, "width": 4}
-
-        encoded_bytes = wasm_encode(COMPONENT_WASM_PATH, arr.tobytes(), config)
-        decoded_bytes = wasm_decode(COMPONENT_WASM_PATH, encoded_bytes, config)
-
-        result = np.frombuffer(decoded_bytes, dtype=np.uint16).reshape(arr.shape)
-        np.testing.assert_array_equal(result, arr)
diff --git a/tests/test_wasm_signature.py b/tests/test_wasm_signature.py
new file mode 100644
index 0000000..a683a3c
--- /dev/null
+++ b/tests/test_wasm_signature.py
@@ -0,0 +1,138 @@
+"""Tests for chonkle:signature custom section reader/writer."""
+
+import json
+
+import pytest
+
+from chonkle.wasm_signature import (
+    SECTION_NAME,
+    embed_signature,
+    read_signature,
+    read_signature_bytes,
+)
+
+# Minimal valid Wasm headers (magic + version only, no sections).
+COMPONENT_HEADER = b"\x00asm\x0d\x00\x01\x00"
+CORE_HEADER = b"\x00asm\x01\x00\x00\x00"
+
+SAMPLE_SIGNATURE = {
+    "codec_id": "zlib",
+    "implementation": "zlib-rs",
+    "inputs": {
+        "bytes": {"type": "bytes", "required": True},
+        "level": {"type": "int", "required": False, "default": 6, "encode_only": True},
+    },
+    "outputs": {"bytes": {"type": "bytes"}},
+}
+
+
+class TestEmbedAndRead:
+    """Round-trip: embed then read back."""
+
+    def test_roundtrip_component_model(self) -> None:
+        result = embed_signature(COMPONENT_HEADER, SAMPLE_SIGNATURE)
+        assert read_signature_bytes(result) == SAMPLE_SIGNATURE
+
+    def test_roundtrip_core_wasm(self) -> None:
+        result = embed_signature(CORE_HEADER, SAMPLE_SIGNATURE)
+        assert read_signature_bytes(result) == SAMPLE_SIGNATURE
+
+    def test_roundtrip_via_file(self, tmp_path) -> None:
+        wasm_path = tmp_path / "test.wasm"
+        wasm_path.write_bytes(embed_signature(COMPONENT_HEADER, SAMPLE_SIGNATURE))
+        assert read_signature(wasm_path) == SAMPLE_SIGNATURE
+
+    def test_embed_replaces_existing(self) -> None:
+        first = embed_signature(COMPONENT_HEADER, {"codec_id": "old"})
+        second = embed_signature(first, SAMPLE_SIGNATURE)
+        assert read_signature_bytes(second) == SAMPLE_SIGNATURE
+        # Only one custom section should exist.
+        assert second.count(SECTION_NAME.encode()) == 1
+
+    def test_minimal_signature(self) -> None:
+        sig = {"codec_id": "identity"}
+        result = embed_signature(COMPONENT_HEADER, sig)
+        assert read_signature_bytes(result) == sig
+
+
+class TestReadErrors:
+    """Error conditions for the reader."""
+
+    def test_not_wasm_raises(self) -> None:
+        with pytest.raises(ValueError, match="Not a valid Wasm binary"):
+            read_signature_bytes(b"not wasm")
+
+    def test_too_short_raises(self) -> None:
+        with pytest.raises(ValueError, match="Not a valid Wasm binary"):
+            read_signature_bytes(b"\x00asm")
+
+    def test_no_signature_section_raises(self) -> None:
+        with pytest.raises(ValueError, match=SECTION_NAME):
+            read_signature_bytes(COMPONENT_HEADER)
+
+    def test_file_not_found(self, tmp_path) -> None:
+        with pytest.raises(FileNotFoundError):
+            read_signature(tmp_path / "nonexistent.wasm")
+
+    def test_no_section_in_wasm_with_other_sections(self) -> None:
+        """A wasm binary with a non-signature custom section still raises."""
+        # Build a custom section with a different name.
+        name = b"other"
+        payload = b"data"
+        name_field = bytes([len(name)]) + name
+        section_body = name_field + payload
+        section = b"\x00" + bytes([len(section_body)]) + section_body
+        wasm = COMPONENT_HEADER + section
+        with pytest.raises(ValueError, match=SECTION_NAME):
+            read_signature_bytes(wasm)
+
+
+class TestEmbedErrors:
+    """Error conditions for the writer."""
+
+    def test_not_wasm_raises(self) -> None:
+        with pytest.raises(ValueError, match="Not a valid Wasm binary"):
+            embed_signature(b"not wasm", {})
+
+
+_CODEC_PARAMS = [
+    ("codec/identity-c/identity.wasm", "identity", "identity-c"),
+    (
+        "codec/tiff-predictor-2-c/tiff-predictor-2.wasm",
+        "tiff-predictor-2",
+        "tiff-predictor-2-c",
+    ),
+    ("codec/zlib-rs/zlib.wasm", "zlib", "zlib-rs"),
+]
+
+
+class TestRealCodecFiles:
+    """Verify reading signatures from the actual built codec .wasm files."""
+
+    @pytest.fixture(params=_CODEC_PARAMS)
+    def codec_info(self, request):
+        from pathlib import Path
+
+        rel_path, codec_id, implementation = request.param
+        wasm_path = Path(__file__).parent.parent / rel_path
+        if not wasm_path.exists():
+            pytest.skip(f"{rel_path} not built")
+        return wasm_path, codec_id, implementation
+
+    def test_embedded_signature_has_expected_fields(self, codec_info) -> None:
+        wasm_path, codec_id, implementation = codec_info
+        sig = read_signature(wasm_path)
+        assert sig["codec_id"] == codec_id
+        assert sig["implementation"] == implementation
+        assert "inputs" in sig
+        assert "outputs" in sig
+
+    def test_embedded_signature_matches_source(self, codec_info) -> None:
+        wasm_path, _codec_id, _implementation = codec_info
+        source_sig_path = wasm_path.parent / "signature.json"
+        if not source_sig_path.exists():
+            pytest.skip("source signature.json not found")
+        with source_sig_path.open() as f:
+            expected = json.load(f)
+        actual = read_signature(wasm_path)
+        assert actual == expected
diff --git a/uv.lock b/uv.lock
index c02be61..5267e26 100644
--- a/uv.lock
+++ b/uv.lock
@@ -271,12 +271,16 @@ name = "chonkle"
 version = "0.0.0"
 source = { editable = "." }
 dependencies = [
-    { name = "numcodecs" },
-    { name = "numpy" },
     { name = "oras" },
     { name = "wasmtime" },
 ]
 
+[package.optional-dependencies]
+native = [
+    { name = "numcodecs" },
+    { name = "numpy" },
+]
+
 [package.dev-dependencies]
 dev = [
     { name = "mypy" },
@@ -293,11 +297,12 @@ notebook = [
 
 [package.metadata]
 requires-dist = [
-    { name = "numcodecs" },
-    { name = "numpy" },
+    { name = "numcodecs", marker = "extra == 'native'" },
+    { name = "numpy", marker = "extra == 'native'" },
     { name = "oras" },
     { name = "wasmtime" },
 ]
+provides-extras = ["native"]
 
 [package.metadata.requires-dev]
 dev = [