What problem does it solve? Fetching YouTube captions programmatically often fails in production: datacenter IPs get blocked, language selection returns the wrong track, and output formats vary across consumers. This Skill guides you through the extract-youtube package, a browserless transcript extractor that runs on serverless and edge platforms without an API key. ## Core Features & Use Cases - Transcript Fetching & Language Control: Use YouTubeTranscriptApi with ordered language preferences, manual-vs-auto track selection, and machine translation between tracks. - Proxy & Error Handling: Configure Webshare or generic HTTP proxies to bypass IP blocks, and diagnose typed errors like RequestBlocked, TranscriptsDisabled, NoTranscriptFound, and PoTokenRequired. - Output Formats & Tooling: Render transcripts as JSON, text, SRT, WebVTT, or article HTML, compute reading-speed encodings, embed the React modal, or run the CLI for one-off extractions. - Use Case: A serverless function needs German subtitles for a video URL. Extract the video id with extractVideoId, call fetchTranscript with languages ["de", "en"], and format the result as SRT — falling back to a Webshare proxy if YouTube blocks the datacenter IP. ## Quick Start Ask the assistant to fetch the English transcript for a given YouTube URL using extract-youtube and format it as SRT, configuring a Webshare proxy if the request gets blocked.