youtube-transcript-analyzer

YouTube transcript analysis extracts text from videos using yt-dlp, then chunks and processes the content to surface key insights, timestamps, and project relevance. This Skill automates transcript retrieval and generates structured summaries—Video Overview, Key Insights, Relevance, and Recommendations—from long-form technical talks and tutorials.

yt-dlp retrieves auto-generated or manual transcripts where available. The Skill uses yt-dlp with `--write-auto-sub` to fetch English subtitles in VTT format, then processes them into actionable chunks. Transcript availability depends on the video's caption settings.

Chunked processing splits lengthy transcripts into meaningful segments, analyzes each chunk separately to generate intermediate summaries, then synthesizes all chunks into a final structured output. This approach maintains context for videos spanning hours while keeping analysis efficient and granular.

The Skill outputs a structured format containing Video Overview, Key Insights with timestamps, Relevance to Your Project, and Specific Recommendations. This structured format links transcript content to your work and enables quick reference back to video moments.

No. The Skill uses yt-dlp with the `--skip-download` flag to extract transcripts without downloading video files, reducing storage overhead and processing time while focusing analysis purely on text content.

Both auto-generated and manual captions work with this Skill's transcript extraction. Manual captions typically offer higher accuracy; auto-generated captions provide broader coverage. The Skill processes whichever is available, prioritizing accurate chunked analysis over caption source.