Rippr Mcp

Pulls YouTube transcripts through a multi-strategy extraction pipeline that hits YouTube's Innertube API, scrapes ytInitialPlayerResponse from page HTML, and falls back to the transcript panel when needed. Exposes a single tool that takes a video URL and saves the transcript locally to ~/rippr/transcripts/, returning the file path so Claude can read it. Output formats include plain text optimized for RAG chunking, timestamped JSON segments, and markdown. Runs entirely local over stdio, so it works in Claude Desktop and Cursor but not web or mobile clients. Useful when you need to analyze long-form video content without manual copy-paste, or when building agents that need to reference spoken material from YouTube at scale.