Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@RAGBrain MCP search for information about the project architecture"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
This project has moved to stache-ai - it is no longer maintained here.
MCP server that connects Claude Desktop to your RAGBrain knowledge base.
pip install ragbrain-mcpRelated MCP server: Personal RAG MCP Server
Add to your config file:
OS | Path |
macOS |
|
Windows |
|
Linux |
|
{
"mcpServers": {
"ragbrain": {
"command": "ragbrain-mcp",
"env": {
"RAGBRAIN_URL": "http://localhost:8000"
}
}
}
}Restart Claude Desktop. Done.
Tool | Description |
| List namespaces with doc counts |
| Semantic search |
| List docs in a namespace |
| Get full document by ID |
| Find docs by topic via summary search |
"What namespaces do I have?"
"Search for machine learning"
"Show docs in work/projects"
"Get document abc-123"
Environment variables:
Variable | Default | Description |
|
| RAGBrain API URL |
|
| Request timeout (seconds) |
|
| DEBUG, INFO, WARNING, ERROR |
|
| Max search results |
|
| Max doc length (chars) |
git clone https://github.com/ragbrain/ragbrain-mcp.git
cd ragbrain-mcp
uv venv && source .venv/bin/activate
uv pip install -e ".[dev]"
pytest┌─────────────────┐ stdio ┌─────────────────┐ HTTP ┌─────────────────┐
│ Claude Desktop │ ◄────────────► │ RAGBrain MCP │ ◄───────────► │ RAGBrain API │
└─────────────────┘ └─────────────────┘ └─────────────────┘Can't connect? Check RAGBrain is running: curl http://localhost:8000/health
Tools not showing? Verify config path, test with ragbrain-mcp, restart Claude Desktop.
Timeouts? Set RAGBRAIN_TIMEOUT higher.
MIT
List all documents stored in a specific namespace. Returns document names, IDs, chunk counts, and creation dates. Use this to see what's in a particular knowledge area.
Inputs: namespace (required), limit.
Discover documents by semantic search over their summaries. Use this to find documents about a topic BEFORE searching for specific content. Returns document titles, headings, and relevance scores. Example queries: 'documents about leadership', 'notes on valuation', 'files covering conflict resolution'. After discovering relevant documents, use ragbrain_search to find specific content.
Inputs: query (required), top_k, namespace.
Retrieve the full content of a specific document by its ID. Returns the complete reconstructed text from all chunks. Use this when you need to read an entire document.
Inputs: doc_id (required).
List all namespaces in the RAGBrain knowledge base. Returns namespace names, descriptions, document counts, and hierarchy. Use this to discover what knowledge areas are available.
Inputs: tree_view.
Perform semantic search across the RAGBrain knowledge base. Returns relevant text chunks that match the query. Use this to find information on any topic stored in RAGBrain.
Inputs: query (required), top_k, namespace.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Each tool serves a distinct purpose: listing namespaces, browsing a namespace, discovering documents via summaries, searching content chunks, and retrieving full documents. No ambiguity in their roles.
All tools follow a consistent 'ragbrain_verb_noun' pattern with clear verbs (list_namespaces, browse_namespace, discover_documents, search, get_document) and snake_case throughout.
Five tools is well-scoped for a knowledge retrieval server, covering all necessary operations without being excessive or insufficient.
The tool set provides complete coverage for querying a RAG knowledge base: exploring namespaces, discovering and searching documents, and retrieving full content. No obvious gaps for read-only operations.
Connect your team's living knowledge base — docs, data, issues, CRM — to Claude and ChatGPT.
Connect Claude, Cursor, or ChatGPT to your business data. Ask questions, get answers.
Your personal knowledge base and memory for AI agents: search, explore and feed your myBrain.
Connect Claude to your Platform7n workspaces: chat, links, work and decks. One-click OAuth.
