op7418/youtube-clipper-skill
Overview
This skill automates downloading, AI-driven chaptering, clipping, subtitle translation, and subtitle burning for YouTube videos. It produces 2–5 minute topic-focused clips with bilingual (English/Chinese) SRTs, hardcoded subtitle videos, and ready-to-use social summaries. The tool emphasizes precise chapter boundaries and efficient batch translation to minimize API usage.
How this skill works
The skill runs a six-stage pipeline: environment checks, video and subtitle download, AI analysis of subtitles to produce fine-grained chapters, user selection of chapters and options, concurrent clip processing (cut, extract/translate subtitles, merge bilingual SRT, burn subtitles, generate summary), and organized output packaging. It uses yt-dlp for downloads, FFmpeg (ffmpeg-full with libass) for clipping and burning, and an AI model to analyze subtitle semantics and create meaningful 2–5 minute chapters.
When to use it
- You need short, topic-focused clips from long YouTube lectures or presentations.
- You want bilingual (English + Chinese) subtitles for clips or full videos.
- You need hardcoded subtitle versions for social platforms without subtitle support.
- You want concise social-media summaries and captions generated automatically.
Best practices
- Run environment detection first to verify yt-dlp, FFmpeg with libass, and Python dependencies.
- Prefer videos with English subtitles; fallback to auto-generated captions if needed.
- Limit each AI chapter to 2–5 minutes for coherent, shareable clips.
- Use batch translation (20 subtitles per batch) to reduce API calls and keep consistency.
- Avoid source file paths with spaces when using FFmpeg; the tool uses a temp directory workaround.
Example use cases
- Create multiple short clips from a 60-minute tech talk for TikTok or YouTube Shorts.
- Produce bilingual clips for bilingual audiences or educational content.
- Generate hardcoded subtitle videos for platforms that do not support external SRT files.
- Extract topic-focused excerpts from interviews for highlights or marketing.
- Quickly produce captioned clips with ready-made social summaries for posting.
FAQ
The skill detects this and prompts installation of ffmpeg-full, explaining the macOS path and providing the install command.
Can I select multiple chapters to process at once?
Yes. The UI supports multi-select; selected chapters are processed in parallel with progress reporting for each step.
How are filenames and special characters handled?
Filenames are sanitized: special characters removed, spaces replaced with underscores, and length capped to avoid filesystem issues.
