Extract YouTube video transcriptions and generate summaries at multiple detail levels. Auto-trigger on YouTube URLs (youtube.com, youtu.be) in conversation when user wants content extracted, summarized, or transcribed. Also trigger on: "summarize this video", "get transcript", "transcribe this", "what does this video say", "what's this video about", or any request involving YouTube video content extraction or summarization. Manually invokable via /yt-summary. Saves transcripts as Obsidian-compatible markdown to ~/Knowledge/Life/Transcriptions organized by topic.
yt-dlp installed via brew (brew install yt-dlp)--cookies-from-browser chrome)python3 ~/.claude/skills/yt-summary/scripts/extract_transcript.py "<URL>" --topic "<Topic>"
Args:
--topic — subfolder in ~/Knowledge/Life/Transcriptions/<Topic>/. Infer from content if not specified.--no-timestamps — omit timestamps from saved transcript.md includes timestamps + YAML frontmatterRead the stdout output from the script. Summarize at the requested level (default: ).
standardbrief — 3-5 bullet points, one sentence each, core ideas only.
standard — Structured with H3 topic headers, 2-3 sentences per topic, key takeaways at end.
detailed — Comprehensive: all major/minor points, impactful quotes, timestamped references
formatted as clickable links [MM:SS](<url>&t=<seconds>), core thesis + arguments + examples,
actionable takeaways section.
Append summary under ## Summary in the saved transcript file using Edit tool.
When unspecified, infer topic from content:
~/Knowledge/Life/Transcriptions/
└── <Topic>/
└── <video-slug>.md # frontmatter + transcript + summary