Expose DuckLM runtime status

duck_core/memory/vector_memory.py

@@ -94,6 +94,43 @@ class VectorMemory:
             response.raise_for_status()
         return response.json().get("result", [])
 
+    def config_status(self) -> dict[str, Any]:
+        if self.embeddings_base_url:
+            embedding_source = f"remote:{self.embeddings_base_url}"
+        elif self._local_model_path:
+            embedding_source = f"local:{self._local_model_path}"
+        else:
+            embedding_source = "none"
+        return {
+            "configured": bool(self.qdrant_url and embedding_source != "none"),
+            "qdrant_url": self.qdrant_url,
+            "collection": self.collection_name,
+            "embedding_source": embedding_source,
+        }
+
+    async def health(self) -> dict[str, Any]:
+        """Probe Qdrant without loading the local embedding model."""
+        import time
+
+        status = self.config_status()
+        started = time.perf_counter()
+        try:
+            async with httpx.AsyncClient(timeout=5.0, trust_env=False) as client:
+                response = await client.get(f"{self.qdrant_url}/")
+                response.raise_for_status()
+            return {
+                **status,
+                "ok": True,
+                "latency_ms": round((time.perf_counter() - started) * 1000, 1),
+            }
+        except httpx.HTTPError as exc:
+            return {
+                **status,
+                "ok": False,
+                "error": str(exc),
+                "latency_ms": round((time.perf_counter() - started) * 1000, 1),
+            }
+
     async def _embed(self, text: str) -> list[float]:
         """Generate embeddings using local model or remote endpoint."""
         # Prefer local model if available