Turns YouTube videos into clipped segments with bilingual subtitles burned in. Downloads the video, uses Claude to analyze subtitles and generate fine-grained chapters (2-5 minute chunks based on topic shifts, not arbitrary timestamps), then lets you pick which segments to extract. The translation happens in batches of 20 subtitles at a time to cut down on API calls. Handles the annoying FFmpeg path issues with spaces by copying to a temp directory. You end up with the raw clip, a subtitled version, the SRT file, and an optional social media summary. Built for Chinese users making bilingual content, but the chapter analysis approach works for anyone who needs semantic chunking instead of just time-based splits.
npx skills add https://github.com/op7418/youtube-clipper-skill --skill youtube-clipper