CodeDox

Crawls documentation websites and extracts code snippets using tree-sitter parsing, automatically categorizing and enriching them with metadata to build searchable code databases for technical documentation.

This MCP server provides AI assistants with comprehensive documentation crawling and code extraction capabilities, built by Chris Scott using Python with FastAPI and PostgreSQL. The implementation combines web crawling through Crawl4AI with tree-sitter-based code parsing to automatically extract, categorize, and enrich code snippets from documentation sites, featuring real-time progress tracking via WebSockets, LLM-powered metadata enhancement, and a React-based web interface for search and management. Built with Docker support, health monitoring for long-running crawl jobs, and both REST API and MCP tool interfaces, it serves developers needing to build searchable code databases from documentation sites, teams requiring automated extraction of code examples from multiple sources, and organizations wanting to create internal knowledge bases from technical documentation with intelligent code categorization and semantic search capabilities.