Skip to main content
Glama
Sign In
rag-knowledge-mcp

rag-knowledge-mcp

by evokateur
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

RAG Knowledge Base MCP Server

This is an MCP server I use to connect a RAG knowledge base to Claude Desktop, Claude Code, and other things.

Setup

With pytest tests:

uv sync --extra dev

Configuration

Defaults live in the committed settings.yaml. Override them (all optional, merged in order, later wins):

settings.yaml                       committed defaults
~/.rag-knowledge-mcp/settings.yaml  per-user override
settings.local.yaml                 per-machine override (gitignored)

Under pytest, settings.test.yaml is merged over the defaults instead, so tests run against an isolated database.

Ingesting Documents

Embedding and retrieval are done with Chroma (other types of embedding/retrieval are possible by implementing AbstractRagBackend)

Default configuration assumes the KB docs will be in ./knowledge-base – which can be a symlink.

What my directory looks like (more or less):

knowledge-base
├── companies
│   └── frobozz-co.md
│   └── acme.md
├── developers
│   └── wesley-hinkle.md
└── projects
|   ├── magic-api-gateway.md
|   ├── zork-legacy-cms.md
|   ├── torch-saas.md
|   ├── grue-detector.md
|   ├── zorkmid-sdk.md
|   ├── hello-footpad.md
|   ├── anvil.md
└── skills-mapping.md

Once the documents are linked up/in place:

uv run python ingest.py

or just

make embeddings

Related MCP server: RAGBrain MCP

Client Configuration

Claude Desktop

Add to Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "rag-knowledge": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/absolute/path/to/this/project",
        "python",
        "rag_knowledge_mcp.py"
      ],
      "env": {
        "LOG_LEVEL": "INFO"
      }
    }
  }
}

uv may need to be an absolute path, depending on how it's installed.

Claude Code

https://code.claude.com/docs/en/mcp

Installs with --scope local by default (claude in CWD)

claude mcp add rag-knowledge -- uv run --directory /absolute/path/to/this/project python rag_knowledge_mcp.py

Add --scope user for claude in any directory:

claude mcp add --scope user rag-knowledge -- uv run --directory /absolute/path/to/this/project python rag_knowledge_mcp.py

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    This server cannot be deployed

    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Resources

    Related MCP Connectors

    Related MCP Servers

    • Claude RAG MCP Pipeline

      kenjisekino
      A
      license
      Not graded
      quality
      D
      maintenance
      Enables Claude Desktop to search and query personal document collections (PDF, Word, Markdown, text) using semantic search and conversational AI with full context preservation across exchanges.
      MIT

    • RAGBrain MCP

      ragbrain
      A
      license
      A
      quality
      F
      maintenance
      Connects Claude Desktop to a RAGBrain knowledge base to enable semantic search, document retrieval, and namespace management. It allows users to browse collections, discover documents by topic, and access full text content through natural language.
      5
      MIT

    • RAG MCP Server

      torkian
      A
      license
      Not graded
      quality
      D
      maintenance
      Enables Claude Desktop to search custom knowledge bases using retrieval-augmented generation via a simple MCP tool.
      MIT

    • Multimodal RAG MCP Server

      brett-hardiman
      A
      license
      Not graded
      quality
      D
      maintenance
      Connects Claude to a personal knowledge base with hybrid search over code, documents, and images. Allows the AI assistant to retrieve and answer from your own files during conversations.
      1
      MIT