asyncapi-knowledge-mcp 0.0.2__tar.gz

asyncapi_knowledge_mcp-0.0.2/PKG-INFO

@@ -0,0 +1,6 @@
+Metadata-Version: 2.4
+Name: asyncapi-knowledge-mcp
+Version: 0.0.2
+Summary: MCP server for documentation search — built with OpenCrane
+Requires-Python: >=3.11
+Requires-Dist: opencrane>=0.18.8

asyncapi_knowledge_mcp-0.0.2/README.md

@@ -0,0 +1,117 @@
+# asyncapi-knowledge-mcp
+
+MCP server for documentation search — built with [OpenCrane](https://github.com/opencrane/opencrane).
+
+## Use locally
+
+```bash
+claude mcp add asyncapi-knowledge-mcp -- uvx --from /path/to/asyncapi-knowledge-mcp asyncapi-knowledge-mcp
+```
+
+## Share via GitHub
+
+Push this directory to a GitHub repository, then others can add it with:
+
+```bash
+claude mcp add asyncapi-knowledge-mcp -- uvx --from "git+https://github.com/you/asyncapi-knowledge-mcp" asyncapi-knowledge-mcp
+```
+
+## Share via PyPI
+
+Build and upload the wheel:
+
+```bash
+pip install build twine
+python -m build .
+twine upload dist/*
+```
+
+Then others can add it with:
+
+```bash
+claude mcp add asyncapi-knowledge-mcp -- uvx asyncapi-knowledge-mcp
+```
+
+> **Note:** PyPI has a 100 MB per-file size limit. If your `milvus.db` exceeds this,
+> distribute via GitHub or a direct path instead.
+
+> **Note:** When you re-pack with updated documentation, bump `--version` so that
+> `uvx` pulls the new version instead of serving its cache.
+
+## Share via Docker
+
+Build and publish a Docker image for HTTP deployment:
+
+```bash
+# Build
+docker build -t your-registry/asyncapi-knowledge-mcp:latest -f .opencrane/Dockerfile ..
+
+# Push
+docker push your-registry/asyncapi-knowledge-mcp:latest
+```
+
+Run the container:
+
+```bash
+docker run -p 8000:8000 your-registry/asyncapi-knowledge-mcp:latest
+```
+
+The MCP endpoint is available at `http://localhost:8000/http`.
+
+## MCP client configuration
+
+### Stdio (uvx)
+
+**Claude Code:**
+
+```bash
+claude mcp add asyncapi-knowledge-mcp -- uvx asyncapi-knowledge-mcp
+```
+
+**Cursor / Windsurf / VS Code (mcp.json):**
+
+```json
+{
+  "mcpServers": {
+    "asyncapi-knowledge-mcp": {
+      "command": "uvx",
+      "args": ["asyncapi-knowledge-mcp"]
+    }
+  }
+}
+```
+
+**Zed (settings.json):**
+
+```json
+{
+  "context_servers": {
+    "asyncapi-knowledge-mcp": {
+      "command": {
+        "path": "uvx",
+        "args": ["asyncapi-knowledge-mcp"]
+      }
+    }
+  }
+}
+```
+
+### HTTP (Docker)
+
+**Claude Code:**
+
+```bash
+claude mcp add asyncapi-knowledge-mcp --transport http --url http://localhost:8000/http
+```
+
+**Cursor / Windsurf / VS Code (mcp.json):**
+
+```json
+{
+  "mcpServers": {
+    "asyncapi-knowledge-mcp": {
+      "url": "http://localhost:8000/http"
+    }
+  }
+}
+```

asyncapi_knowledge_mcp-0.0.2/asyncapi_knowledge_mcp/__main__.py

@@ -0,0 +1,17 @@
+"""MCP server with bundled documentation data."""
+import os
+from pathlib import Path
+
+
+def main():
+    data_dir = Path(__file__).parent / "data"
+    os.environ.setdefault("MILVUS_DB_PATH", str(data_dir / "milvus.db"))
+    os.environ.setdefault("AI_DOCS_CHUNKS_FILE", str(data_dir / "chunks.json"))
+
+    import asyncio
+    from opencrane.mcp.server import main as serve_main
+    asyncio.run(serve_main())
+
+
+if __name__ == "__main__":
+    main()