An MCP server for YouTube operations that demonstrates core MCP concepts, including a tools framework and robust logging. It exposes a single main tool get_yt_video_info that, given a YouTube video ID, returns metadata via yt-info-extract (title, description, channel, publish date, view count) and retrieves a transcript via yt-ts-extract, with intelligent fallback logic to handle different transcript types and both manually created and auto-generated transcripts. The server works out of the box without an API key for basic functionality, using yt-info-extract and yt-ts-extract. Optional YouTube API key can be configured for enhanced metadata retrieval and as a fallback. The project emphasizes comprehensive error handling with graceful extraction failures, multiple fallback strategies, standardized error responses, and extensive logging for debugging. It includes development support with uv, tests, and a clear project structure. The MCP server can be run via mcp_youtube_extract or uv, and is designed to be easily integrated by clients via a simple configuration describing the MCP server and environment variables.