You run an MCP server that provides tooling to retrieve and process documentation using vector search, enabling AI assistants to augment their responses with relevant docs. This server can ingest documentation sources, perform semantic searches, and deliver context-aware results to your AI workflows.

How to use

Launch the MCP server alongside your client to enable documentation-aware responses. You will configure one or more MCP servers to manage embeddings, sources, and search operations. Use the standard toolset to search documentation, list sources, and manage the processing queue. You can run different embedding providers (Ollama or OpenAI) and point the server at a Qdrant vector store for fast retrieval.

How to install

# Prerequisites
- Node.js installed (recommended LTS version)
- npm available
- Access to a Qdrant instance (local or cloud) with a URL and API key if needed

# Install dependencies for the MCP server project
- Clone or download the MCP Ragdocs project
- Navigate to the project directory
- Install dependencies

npm install

Run an MCP server instance using npx (example for rag-docs)

npx -y @sanderkooger/mcp-server-ragdocs

Start a specific configuration (as shown in examples) with environment variables set

EMBEDDINGS_PROVIDER=ollama \
OLLAMA_BASE_URL=http://localhost:11434 \
QDRANT_URL=http://localhost:6333 \
QDRANT_API_KEY=YOUR_QDRANT_KEY \
OPENAI_API_KEY=YOUR_OPENAI_KEY \
node ./build/index.js

Verify

• Access Qdrant at http://localhost:6333
• Access Ollama at http://localhost:11434
• Confirm the server starts without errors

Configuration

The MCP Ragdocs server provides multiple configuration variants. Use the appropriate setup for your embedding provider and deployment. The following examples show the exact MCP configuration blocks you can run to start the server with different embedding providers.

json
{
  "mcpServers": {
    "rag_docs": {
      "command": "npx",
      "args": ["-y", "@sanderkooger/mcp-server-ragdocs"],
      "env": {
        "EMBEDDINGS_PROVIDER": "ollama",
        "QDRANT_URL": "your-qdrant-url",
        "QDRANT_API_KEY": "your-qdrant-key"
      }
    }
  }
}

json
{
  "mcpServers": {
    "rag_docs_openai": {
      "command": "npx",
      "args": ["-y", "@sanderkooger/mcp-server-ragdocs"],
      "env": {
        "EMBEDDINGS_PROVIDER": "openai",
        "OPENAI_API_KEY": "your-openai-key-here",
        "QDRANT_URL": "your-qdrant-url",
        "QDRANT_API_KEY": "your-qdrant-key"
      }
    }
  }
}

json
{
  "mcpServers": {
    "rag_docs_ollama": {
      "command": "npx",
      "args": ["-y", "@sanderkooger/mcp-server-ragdocs"],
      "env": {
        "EMBEDDINGS_PROVIDER": "ollama",
        "OLLAMA_BASE_URL": "http://localhost:11434",
        "QDRANT_URL": "your-qdrant-url",
        "QDRANT_API_KEY": "your-qdrant-key"
      }
    }
  }
}

Ollama run from this codebase

json
"ragdocs-mcp": {
      "command": "node",
      "args": [
        "/home/sander/code/mcp-server-ragdocs/build/index.js"
      ],
      "env": {
        "QDRANT_URL": "http://127.0.0.1:6333",
        "EMBEDDINGS_PROVIDER": "ollama",
        "OLLAMA_URL": "http://localhost:11434"
      },
      "alwaysAllow": [
        "run_queue",
        "list_queue",
        "list_sources",
        "search_documentation",
        "clear_queue",
        "remove_documentation",
        "extract_urls"
      ],
      "timeout": 3600
    }

Security and maintenance notes

Keep your Qdrant and embedding services secured behind appropriate access controls. Use API keys where provided, rotate credentials regularly, and monitor for unusual activity in the processing queue and search endpoints.

Troubleshooting

If you encounter startup errors, verify that all required services (Qdrant and embedding provider) are reachable at their configured URLs. Check environment variable names and values for typos, and ensure the MCP server process has permission to read its configuration and start the build index.

Notes

This server focuses on providing vector-based documentation search and retrieval, integrating with LLMs to augment responses with relevant passages and excerpts. It supports multiple documentation sources and can operate with local Ollama embeddings or OpenAI embeddings.

Available tools

search_documentation

Search through stored documentation using natural language queries and return contextually relevant excerpts.

list_sources

List all indexed documentation sources with details like URLs, titles, and last update times.

extract_urls

Extract and analyze all URLs from a given web page and optionally add them to the processing queue.

remove_documentation

Remove specific documentation sources from the system by their URLs.

list_queue

Show all URLs currently waiting in the documentation processing queue.

run_queue

Process and index all URLs in the documentation queue with progress updates.

clear_queue

Clear all pending URLs from the documentation processing queue.