DevDocs-MCP provides a local, version-aware, deterministic documentation layer for AI agents. It serves as a middleware between your coding agents and the DevDocs data, caching documentation offline and mapping docs to the specific versions in your project. Agents can query the MCP for API metadata, signatures, and usage examples without network calls, reducing hallucinations and latency. The server exposes a standard MCP-compatible interface, so you can configure agents like RooCode, Claude, or Cline to search, explain, and retrieve structured documentation content via the MCP protocol.

To use it, start the server and point your agent configuration to the devdocs MCP server. The agent’s mcp_settings.json should reference the server name (e.g., devdocs) and the command used to launch the server. Typical workflows include: requesting API signatures and usage for dependencies pinned in your package.json, ingesting new documentation datasets via the built-in tooling, and performing fast offline searches or explain requests against the cached docs. The result is a clean, LLM-friendly JSON payload containing metadata and relevant content for the requested API surface.

Prerequisites:

• Node.js 18+ installed on your machine
• pnpm (recommended) for development

Installation steps:

1. Clone the repository git clone https://github.com/madhan-g-p/DevDocs-MCP.git cd DevDocs-MCP

2. Install dependencies pnpm install

3. Configure environment (optional but recommended) cp .env.example .env

   Edit .env to customize DEVDOCS_DATA_PATH, MCP_DB_PATH, LOG_LEVEL, etc.

4. Build and run in production mode pnpm build pnpm start:prod

5. If you prefer Docker (optional):

   • Follow the Docker usage section in the README to run with Docker Compose or docker run.

Note: Strictly use pnpm for development and contributions, as the project expects pnpm-based workflows.

Tips and notes:

• This is a Node-only project that uses sql.js for a zero-native-dependency stack. No Python or C++ builds are required.
• The server caches DevDocs data locally to provide offline, deterministic documentation. Ensure you have adequate disk space for large documentation datasets.
• Ingest documentation for your stack by providing a dependencies object (e.g., {"react": "18.2.0", "typescript": "5.0.0"}) and using the ingest tooling once the server is up.
• The default environment variables live in .env. Key variables include DEVDOCS_DATA_PATH (where docs are stored), MCP_DB_PATH (the SQLite database location), and LOG_LEVEL.
• If you deploy with Docker, both the database and the downloaded docs are stored under a single volume for easy backup and relocation.
• The MCP configuration snippet in your agent should point to the server name (e.g., devdocs) and use the appropriate command and arguments (node dist/main.js) to start the server.