youtube-summarize_skill

Overview

This skill summarizes YouTube videos by extracting transcripts and captions, cleaning the text, and generating concise summaries or key points. It is designed to speed content analysis so you can get the gist of a video without watching it end-to-end. The skill supports auto-generated and manual subtitles and can handle multiple languages when available.

How this skill works

The skill uses yt-dlp to fetch subtitles or transcripts in VTT or json3 formats, then removes timestamps, headers, and markup to produce clean plain text. It can also dump video metadata and chapters. The cleaned transcript is ready to be sent to an LLM for summarization, highlighting, or further analysis.

When to use it

• You need a quick summary of a lecture, tutorial, interview, or podcast video.
• You want to extract key points, action items, or a transcript for research or notes.
• You need multilingual subtitle extraction for non-English content.
• You want to analyze or archive video content without downloading full video files.
• You need chapter titles or metadata for indexing and search.

Best practices

• Ensure openclawcli is installed and running before attempting transcript extraction.
• Prefer json3 or vtt output and run a cleaning step to remove timestamps and duplicate lines.
• Request manual subtitles first if available; fall back to auto-generated captions when necessary.
• Use language codes with --sub-lang to extract specific subtitle languages.
• Check transcript quality for heavily accented or noisy audio before relying on summaries.

Example use cases

• Summarize a 90-minute webinar into a one-paragraph executive summary and bullet list of takeaways.
• Extract and clean a podcast transcript for searchable notes and citation.
• Generate chapter-based summaries from tutorial videos to create a study guide.
• Archive subtitles from a multilingual talk for later translation and analysis.
• Quickly preview video content to decide if a full viewing is required.