chore: bump v0.7.0, update changelog and README

Add memory_edit docs, stats/export CLI commands, relay config
examples, Phase 7 roadmap entry. Closes #12, #13, #14.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: singularityjason

CHANGELOG.md

@@ -1,5 +1,31 @@
 # Changelog
 
+## [0.7.0] - 2026-03-24
+
+### Added
+- `memory_edit` MCP tool for correcting stored memories with edit tracking (21 → 22 tools)
+- Optional semantic search via ONNX embeddings (all-MiniLM-L6-v2, 384-dim)
+  - Install with `pip install lightning-memory[semantic]`
+  - Hybrid FTS5 + cosine similarity with reciprocal rank fusion
+  - Hash-based fallback when onnxruntime not installed
+- Memory deduplication using Jaccard word-set similarity (per-type thresholds)
+- `lightning-memory stats` CLI command — memory statistics dashboard
+- `lightning-memory export [json|csv]` CLI command
+- Schema migration system using PRAGMA user_version
+- Per-relay circuit breaker (auto-skip down relays, exponential backoff)
+- L402 payment idempotency (duplicate tokens detected, payment logged once)
+- Relay configuration examples in README
+
+### Fixed
+- Vendor name normalization across all subsystems — `bitrefill.com`, `www.bitrefill.com`, `BITREFILL.COM` now match consistently in reputation, budget, trust, and anomaly detection
+- Intelligence engine double-scan — `vendor_report()` now counts failures in a single pass instead of scanning all memories twice
+- Relay sync batching — `push_memories()` uses single event loop instead of per-memory `asyncio.run()`
+- `pull_trust_assertions()` now fetches all vendors concurrently instead of sequentially
+
+### Changed
+- `GatewayClient` reuses persistent `httpx.Client` with connection pooling
+- `PhoenixdClient` reuses persistent `httpx.AsyncClient` with connection pooling
+
 ## [0.6.0] - 2026-03-14
 
 ### Added

README.md

@@ -279,6 +279,19 @@ ln_remote_query(
 # → {status: "success", data: {count: 3, memories: [...]}}
 ```
 
+### `memory_edit`
+
+Edit an existing memory's content or metadata. Tracks edit history.
+
+```
+memory_edit(
+  id="7222215a5eba8547",
+  content="Bitrefill now charges 300 sats (was 500)",
+  metadata='{"note": "price updated March 2026"}'
+)
+# → {id: "7222...", content: "...", old_content_preview: "...", edit_count: 1}
+```
+
 ### `memory_sync`
 
 Sync memories with Nostr relays (push and/or pull). Also pulls NIP-85 trust assertions and gateway announcements.
@@ -429,11 +442,84 @@ Generate a `.well-known/lightning-memory.json` manifest for DNS-based gateway di
 lightning-memory-manifest > .well-known/lightning-memory.json
 ```
 
+### `lightning-memory stats`
+
+Show a memory statistics dashboard:
+
+```bash
+lightning-memory stats
+# Lightning Memory — Statistics Dashboard
+# ==========================================
+#   Agent: 7222215a5eba8547...
+#   Total memories: 147
+#
+#   By type:
+#     transaction              89
+#     vendor                   23
+#     decision                 18
+#     error                    12
+#     attestation               5
+#
+#   Spending (30d): 42,500 sats across 89 txns
+#   Top vendors:
+#     bitrefill.com                    15,200 sats
+#     openai.l402.io                    8,300 sats
+#
+#   Semantic search: active
+```
+
+### `lightning-memory export [json|csv]`
+
+Export memories to JSON or CSV:
+
+```bash
+lightning-memory export json > backup.json
+lightning-memory export csv > backup.csv
+```
+
+## Relay Configuration
+
+Memories sync to Nostr relays via NIP-78 events. Default relays:
+
+```json
+["wss://relay.damus.io", "wss://nos.lol", "wss://relay.nostr.band"]
+```
+
+To customize, create `~/.lightning-memory/config.json`:
+
+```json
+{
+  "relays": [
+    "wss://relay.damus.io",
+    "wss://nos.lol",
+    "wss://relay.nostr.band",
+    "wss://relay.primal.net"
+  ],
+  "sync_timeout_seconds": 30,
+  "max_events_per_sync": 500
+}
+```
+
+**Choosing relays:**
+
+| Relay | Speed | Reliability | Notes |
+|-------|-------|-------------|-------|
+| `wss://relay.damus.io` | Fast | High | Most popular, good uptime |
+| `wss://nos.lol` | Fast | High | Reliable, good NIP-78 support |
+| `wss://relay.nostr.band` | Medium | Medium | Search-focused, may be slow |
+| `wss://relay.primal.net` | Fast | High | Primal's relay, well-maintained |
+| `wss://nostr.wine` | Fast | High | Paid relay, less spam |
+
+**Tips:**
+- Use 2-3 relays for redundancy — Lightning Memory deduplicates across relays
+- Check relay health: `lightning-memory relay-status`
+- Relays with a circuit breaker tripped are automatically skipped until they recover
+
 ## How It Works
 
 1. **First run**: A Nostr keypair is generated and stored at `~/.lightning-memory/keys/`
 2. **Storing**: Memories go to local SQLite with FTS5 indexing. Each memory is tagged with your agent's public key.
-3. **Querying**: Full-text search with BM25 ranking returns the most relevant memories.
+3. **Querying**: Full-text search with BM25 ranking returns the most relevant memories. With `pip install lightning-memory[semantic]`, queries also use ONNX-based semantic similarity for concept matching.
 4. **Identity**: Your agent's public key is a globally unique, cryptographically verifiable identifier. No accounts needed.
 
 ## Data Storage
@@ -458,6 +544,7 @@ All data is stored locally:
 - [x] Phase 5.1: Community reputation — live NIP-85 trust attestation sync
 - [x] Phase 5.2: Compliance integration — KYA attestations, LNURL-auth sessions, compliance reports
 - [x] Phase 6: Memory marketplace — gateway discovery (Nostr + DNS), remote L402 queries, gateway client
+- [x] Phase 7: Agent reliability — vendor normalization, deduplication, memory edit, semantic search, circuit breakers, L402 idempotency
 
 ## Star History
 

lightning_memory/__init__.py

@@ -1,3 +1,3 @@
 """Lightning Memory: Decentralized agent memory for the Lightning economy."""
 
-__version__ = "0.6.0"
+__version__ = "0.7.0"

lightning_memory/server.py

@@ -908,13 +908,98 @@ def gateway_manifest_main() -> None:
     print(json.dumps(manifest, indent=2))
 
 
+def _cmd_stats() -> None:
+    """Show memory statistics dashboard."""
+    engine = _get_engine()
+    stats = engine.stats()
+    intel = _get_intelligence()
+
+    print("Lightning Memory — Statistics Dashboard")
+    print("=" * 42)
+    print(f"  Agent: {stats['agent_pubkey'][:16]}...")
+    print(f"  Total memories: {stats['total']}")
+    print()
+
+    if stats["by_type"]:
+        print("  By type:")
+        for t, count in sorted(stats["by_type"].items(), key=lambda x: x[1], reverse=True):
+            print(f"    {t:20s} {count:>5}")
+    else:
+        print("  No memories stored yet.")
+    print()
+
+    # Spending summary
+    summary = intel.spending_summary("30d")
+    if summary.txn_count > 0:
+        print(f"  Spending (30d): {summary.total_sats:,} sats across {summary.txn_count} txns")
+        if summary.by_vendor:
+            print("  Top vendors:")
+            for vendor, sats in sorted(summary.by_vendor.items(), key=lambda x: x[1], reverse=True)[:5]:
+                print(f"    {vendor:30s} {sats:>8,} sats")
+    print()
+
+    # Budget rules
+    budget = _get_budget()
+    rules = budget.list_rules()
+    if rules:
+        print(f"  Budget rules: {len(rules)}")
+        for r in rules:
+            limits = []
+            if r.max_sats_per_txn:
+                limits.append(f"{r.max_sats_per_txn}/txn")
+            if r.max_sats_per_day:
+                limits.append(f"{r.max_sats_per_day}/day")
+            if r.max_sats_per_month:
+                limits.append(f"{r.max_sats_per_month}/mo")
+            print(f"    {r.vendor:30s} {', '.join(limits)}")
+
+    # Embedding status
+    try:
+        from .embedding import has_embeddings, get_embedding_info
+        info = get_embedding_info()
+        status = "active" if info["model_loaded"] else ("available" if info["available"] else "not installed")
+        print(f"\n  Semantic search: {status}")
+    except Exception:
+        print("\n  Semantic search: not installed")
+
+
+def _cmd_export(fmt: str = "json") -> None:
+    """Export memories to JSON or CSV."""
+    import csv
+    import io
+
+    engine = _get_engine()
+    memories = engine.list(limit=10000)
+
+    if fmt == "csv":
+        output = io.StringIO()
+        if memories:
+            fields = ["id", "content", "type", "created_at"]
+            writer = csv.DictWriter(output, fieldnames=fields, extrasaction="ignore")
+            writer.writeheader()
+            for m in memories:
+                writer.writerow(m)
+        print(output.getvalue(), end="")
+    else:
+        print(json.dumps(memories, indent=2, default=str))
+
+
 def main():
     """Run the MCP server, or a CLI subcommand."""
     import sys
 
-    if len(sys.argv) > 1 and sys.argv[1] == "relay-status":
-        _cmd_relay_status()
-        return
+    if len(sys.argv) > 1:
+        cmd = sys.argv[1]
+        if cmd == "relay-status":
+            _cmd_relay_status()
+            return
+        elif cmd == "stats":
+            _cmd_stats()
+            return
+        elif cmd == "export":
+            fmt = sys.argv[2] if len(sys.argv) > 2 else "json"
+            _cmd_export(fmt)
+            return
 
     mcp.run()
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "lightning-memory"
-version = "0.6.0"
+version = "0.7.0"
 description = "Decentralized agent memory for the Lightning economy. Nostr identity, L402 payments, semantic search."
 readme = "README.md"
 license = "MIT"