How baonguyen09/github-second-brain works
Project type and purpose
• This is an open-source, self-hosted MCP (Model Context Protocol) server packaged as a Dockerized Python application.
• Its purpose is to expose tools over MCP so an AI assistant (LLM) can fetch targeted information (directory tree, file contents, issue context) from arbitrary GitHub repositories without exceeding its context window.
File structure
• Top level
• Dockerfile – builds the MCP server image
• .github/workflows/docker-publish.yml – CI/CD pipeline to build & push the image to GitHub Container Registry
• mcp/ – core Python application
• Under mcp/
• main.py – MCP server entrypoint (registers tools, handles requests)
• requirements.txt – Python dependencies
• tools/
– fetch_directory_tree.py
– fetch_file_contents.py
– fetch_issue_context.py
– custom_errors.py
• assets/images and README.md provide documentation and branding.
README for additional insights
• The README describes the MCP tools and their use cases.
• It shows how AI clients (e.g., Claude Desktop, Cursor) integrate by running the Docker container locally and making tool calls.