diff --git a/.github/workflows/reusable-pypi-publish.yml b/.github/workflows/reusable-pypi-publish.yml
index b6700b626..e5279cf70 100644
--- a/.github/workflows/reusable-pypi-publish.yml
+++ b/.github/workflows/reusable-pypi-publish.yml
@@ -84,7 +84,8 @@ jobs:
           DIST_FILES=(./dist/nemo_retriever/dist/*.whl ./dist/nemo_retriever/dist/*.tar.gz)
           echo "After legacy glob: count=${#DIST_FILES[@]} files=${DIST_FILES[*]:-<none>}"
 
-          # Flat layout (older upload-artifact glob uploads): ./dist/*
+          # Flat layout (upload-artifact strips to the common parent, so wheels
+          # often land directly under ./dist/): ./dist/*
           if [ "${#DIST_FILES[@]}" -eq 0 ]; then
             DIST_FILES=(./dist/*.whl ./dist/*.tar.gz)
             echo "After flat glob: count=${#DIST_FILES[@]} files=${DIST_FILES[*]:-<none>}"
diff --git a/nemo_retriever/src/nemo_retriever/service/app.py b/nemo_retriever/src/nemo_retriever/service/app.py
index f7cef45b1..290dfb840 100644
--- a/nemo_retriever/src/nemo_retriever/service/app.py
+++ b/nemo_retriever/src/nemo_retriever/service/app.py
@@ -271,14 +271,15 @@ def create_app(config: ServiceConfig) -> FastAPI:
     else:
         logger.info("Bearer-token authentication DISABLED (no api_token configured)")
 
-    from nemo_retriever.service.routers import admin, ingest, metrics
+    from nemo_retriever.service.routers import admin, ingest, metrics, test
     from nemo_retriever.service.services.prometheus import instrument_app
 
     app.include_router(ingest.router, prefix="/v1")
     app.include_router(metrics.router, prefix="/v1")
-    # Admin/internal endpoints — pool_stats etc. Registered on every
-    # role; the handler self-reports an empty pool dict on gateway pods.
+    # Admin/internal endpoints — pool_stats etc.
     app.include_router(admin.router, prefix="/v1")
+    # General-purpose smoke-test route — no external deps.
+    app.include_router(test.router, prefix="/v1")
     instrument_app(app, role=config.mode)
 
     if config.mode == "gateway":
diff --git a/nemo_retriever/src/nemo_retriever/service/routers/test.py b/nemo_retriever/src/nemo_retriever/service/routers/test.py
new file mode 100644
index 000000000..560ad326d
--- /dev/null
+++ b/nemo_retriever/src/nemo_retriever/service/routers/test.py
@@ -0,0 +1,43 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES.
+# All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+"""Test endpoint for the retriever service.
+
+Provides a lightweight, no-dependency health-check route that validates
+the Python runtime and the current service mode are reachable at all.
+"""
+
+from __future__ import annotations
+
+import logging
+import platform
+import sys
+
+from fastapi import APIRouter, Request
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter(tags=["test"], include_in_schema=True)
+
+
+@router.get("/test", summary="Health-check that validates the Python runtime")
+async def test(request: Request) -> dict:
+    """Return a JSON blob describing the current process environment.
+
+    Response shape::
+
+        {
+          "status": "ok",
+          "mode": "gateway" | "realtime" | "batch" | "standalone",
+          "python": "3.12.1+linux",
+        }
+
+    Intended for cluster probes, load-balancer heart-beats, and manual
+    smoke-tests -- it has no external dependencies (no DB, no pipeline
+    pool, no media binaries).
+    """
+    config = getattr(request.app.state, "config", None)
+    mode = config.mode if config is not None else "unknown"
+    runtime = f"{sys.version.split()[0]}; {'/'.join(platform.system().split())}"
+    return {"status": "ok", "mode": mode, "python": runtime}