Remove enable_retrieval flag — always attempt vtk-index on startup

Retrieval is silently skipped on ImportError (vtk-index not installed)
or any other init failure, so a separate toggle adds no value.
Also drop enable_cpp_scraping (no scraping tools remain).

Author: vicentebolea

README.md

@@ -43,8 +43,7 @@ Key environment variables:
 |---|---|---|
 | `VTK_MCP_VTK_VERSION` | `9.3.0` | VTK version to fetch artifacts for |
 | `VTK_MCP_KNOWLEDGE_ARTIFACT_PATH` | _(auto)_ | Local JSONL path; skips auto-download |
-| `VTK_MCP_ENABLE_RETRIEVAL` | `false` | Enable vtk-index semantic search |
-| `VTK_MCP_QDRANT_URL` | _(auto)_ | Qdrant URL; if unset uses embedded storage |
+| `VTK_MCP_QDRANT_URL` | _(auto)_ | Qdrant server URL; if unset uses embedded storage from ghcr.io |
 | `VTK_MCP_ENABLE_VALIDATION` | `true` | Enable vtk-validate code validation |
 | `VTK_MCP_TRANSPORT` | `stdio` | `stdio` or `http` |
 

deploy.Dockerfile

@@ -41,7 +41,6 @@ ENV VTK_MCP_TRANSPORT=http
 ENV VTK_MCP_HTTP_HOST=0.0.0.0
 ENV VTK_MCP_HTTP_PORT=8000
 ENV VTK_MCP_ENABLE_VALIDATION=true
-ENV VTK_MCP_ENABLE_RETRIEVAL=true
 
 EXPOSE 8000
 

src/vtk_mcp/composition.py

@@ -43,29 +43,25 @@ def __init__(self, settings: Settings) -> None:
             self.api_index.vtk_version,
         )
 
-        # Layer 2: retrieval (optional)
+        # Layer 2: retrieval — always attempted; stays None if vtk-index not installed
         self.retriever = None
-        if settings.enable_retrieval:
-            try:
-                from vtk_index import Retriever
-
-                if settings.qdrant_url:
-                    # Connect to a running Qdrant server
-                    self.retriever = Retriever(
-                        qdrant_url=settings.qdrant_url,
-                        vtk_version=self.api_index.vtk_version,
-                    )
-                    logger.info("Retriever connected to %s", settings.qdrant_url)
-                else:
-                    # Download pre-built embedded storage (no server required)
-                    logger.info(
-                        "Downloading embedded Qdrant storage for VTK %s",
-                        self.api_index.vtk_version,
-                    )
-                    self.retriever = Retriever.from_artifact(self.api_index.vtk_version)
-                    logger.info("Retriever ready (embedded storage)")
-            except Exception as exc:
-                logger.warning("Retrieval disabled: %s", exc)
+        try:
+            from vtk_index import Retriever
+
+            if settings.qdrant_url:
+                self.retriever = Retriever(
+                    qdrant_url=settings.qdrant_url,
+                    vtk_version=self.api_index.vtk_version,
+                )
+                logger.info("Retriever connected to %s", settings.qdrant_url)
+            else:
+                logger.info("Downloading embedded Qdrant storage for VTK %s", self.api_index.vtk_version)
+                self.retriever = Retriever.from_artifact(self.api_index.vtk_version)
+                logger.info("Retriever ready (embedded storage)")
+        except ImportError:
+            logger.info("vtk-index not installed — vector search unavailable")
+        except Exception as exc:
+            logger.warning("Retriever init failed: %s", exc)
 
         # Layer 3: validation (optional — library call, no subprocess)
         self.validate = None

src/vtk_mcp/config.py

@@ -16,11 +16,8 @@ class Settings(BaseSettings):
     knowledge_artifact_path: Optional[Path] = None
     vtk_version: str = "9.3.0"
 
-    # Layer 2 — retrieval
-    # When enable_retrieval is True and vtk_version is set, uses Retriever.from_artifact()
-    # which downloads pre-built embedded Qdrant storage (no server required).
-    # Set qdrant_url to a real Qdrant instance to use a server instead.
-    enable_retrieval: bool = False
+    # Layer 2 — retrieval (always attempted; silently disabled if vtk-index not installed)
+    # Set qdrant_url to a running Qdrant server; if unset, uses embedded storage from ghcr.io.
     qdrant_url: Optional[str] = None
 
     # Layer 3 — validation
@@ -31,9 +28,5 @@ class Settings(BaseSettings):
     http_host: str = "0.0.0.0"
     http_port: int = 8000
 
-    # C++ docs scraping
-    enable_cpp_scraping: bool = True
-    vtk_docs_base_url: str = "https://vtk.org/doc/nightly/html/"
-
     class Config:
         env_prefix = "VTK_MCP_"

tests/test_tools.py

@@ -54,7 +54,6 @@ def ctx():
 
     settings = Settings(
         knowledge_artifact_path=None,
-        enable_retrieval=False,
         enable_validation=True,
     )