docs-mcp
The Docs MCP Server is a flexible server designed for creating searchable documentation or codebases that can interact with AI assistants using the Model Context Protocol. It leverages the Probe search engine to facilitate natural language queries across configured Git repositories or local directories, enhancing the accessibility of technical content.
Docs MCP Server
This project provides a flexible Model Context Protocol (MCP) server powered by Probe, designed to make documentation or codebases searchable by AI assistants. It allows users to chat with code or documents by pointing the server to a Git repository or a folder. The server can integrate with AI assistants and MCP clients, making it easy to query documentation and codebases.
Features
- Powered by Probe: Uses the Probe search engine for efficient results.
- Flexible Content Sources: Supports local directories and Git repositories.
- Dynamic Configuration: Configure content sources via config file, CLI arguments, or environment variables.
- Automatic Git Updates: Keeps content fresh by updating from a Git repository.
- Customizable Integration: Define the name and description of the MCP tool for AI use.
Usage
The server is primarily used with npx for easy integration with AI assistants and MCP clients. Installation can be automated via Smithery, and integration with MCP clients like IDEs is supported. Configuration involves setting options for content sourcing from directories or Git.
Configuration
A docs-mcp.config.json file can be used to set default content sources and MCP tool details, while runtime configurations override these defaults. The server provides tools for dynamic searching through AI assistants using configurable queries.
Creating Your Own Pre-built MCP Server
Users can create and publish their own MCP server using this as a template, providing a zero-configuration setup for utilizing standardized documentation or codebase searches.