Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
- Expanded the description in SKILL.md to clarify when and how to use the skill, especially for requests involving YouTube, X/Twitter, or TikTok video URLs. - Added practical usage examples to the description, detailing scenarios such as summarizing videos, extracting key points, and analyzing video content. - No changes to any code or functionality; documentation improvements only.
What This Skill Does
Downloads, transcribes, and analyzes videos from YouTube, X/Twitter, and TikTok using a two-tier system: yt-dlp for fast subtitle extraction and local whisper-cpp as fallback. Outputs timestamped transcripts, TL;DRs, key moments, and actionable insights. Responds in the user's language regardless of the video's language.
Local Whisper processing keeps transcription private and avoids API costs, while the yt-dlp fast-path means most videos with existing subtitles are processed in seconds.
When to use it
- Get a timestamped summary of a 1-hour YouTube conference talk
- Find the exact moment a specific topic is mentioned in a podcast
- Download a TikTok video as MP4 to your Desktop
- Transcribe an X/Twitter video that has no subtitles
- Extract key takeaways from a product demo in a foreign language
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Can you summarize this YouTube video? https://youtube.com/watch?v=abc123
- 1Runs analyze_video.py with --action transcript and the provided URL
- 2yt-dlp fetches subtitles if available; falls back to whisper-cpp if not
- 3Reads the output .txt file containing timestamped transcript
- 4Formats response with TL;DR, Key Moments with timestamps, and Actionable Insights
A structured Markdown response with a 3-sentence summary, 3-7 timestamped key moments, and a numbered list of practical takeaways