Episode → show notes & chapters
Show notes written from memory miss half the episode, and chapters guessed from the timeline land in the wrong place. With a timed transcript, both come from what was actually said and when.
What you get
- YouTube chapters (00:00 Intro …) at the real topic changes
- Show notes with a summary, guest quotes and links mentioned
- A description that matches the episode
With an AI assistant
Connect the MCP server to Claude, ChatGPT or Cursor, then paste:
Read the transcript of <episode URL> with timestamps. Write YouTube chapters starting at 00:00, one per real topic change, with short titles. Then write show notes: a 3-sentence summary, 5 key takeaways, 3 verbatim guest quotes with timestamps, and every book, tool or link mentioned.
With the API
Set GYT_API_KEY to a key from the developers page. Each transcript, list page or channel check costs one credit.
1. Get the episode with per-line timestamps
curl -s -H "Authorization: Bearer $GYT_API_KEY" \ "https://getyoutubetranscript.com/api/v1/transcript?v=VIDEO_ID×tamps=true"
Limits to know
- YouTube needs at least three chapters, the first at 00:00, each at least 10 seconds long.
Tools for this workflow
Questions
Does it work before the episode is public?
Only public or unlisted videos with captions can be read. Upload as unlisted, let captions generate, then run it.
More workflows
EducationYouTube playlist → study notesA 20-lecture course is ten hours of video and about an hour of reading. Get the reading.EducationLectures → flashcards & quizzesTesting yourself beats rewatching. Build the questions from the transcript.MarketingCompetitor channels → weekly briefKnow what three competitors published this week without watching any of it.DevelopersPlaylist → RAG corpusIndex what a channel said, chunked on caption timings, with links that open at the right second.ContentYouTube video → blog postA post built from what was said, not a generic article on the same topic.ContentOne video → a week of postsThe best hooks are lines someone already said out loud.
