bayramannakov/notebooklm-youtube-skill
Overview
This skill automates creating NotebookLM notebooks from YouTube videos by extracting featured people, researching them online, adding sources to a NotebookLM notebook, and generating an Audio Overview. It uses a screenshot-first automation pattern to reliably detect UI state and target the correct NotebookLM elements. The result is a ready-made notebook with the video and curated research as sources plus an audio summary.
How this skill works
Given a YouTube URL, the skill opens the video page, parses visible metadata and spoken participants, and runs targeted web searches to compile concise research documents. It then opens NotebookLM, creates a new notebook, inserts the YouTube link and the research text as sources using precise textarea selectors and synthetic input events, and triggers Audio Overview generation. Screenshots are taken before and after key actions to verify UI state and confirm success.
When to use it
- Create research-backed NotebookLM notebooks from single YouTube videos
- Rapidly aggregate background on people mentioned in a video
- Automate source insertion into NotebookLM for teaching or briefing materials
- Generate an audio summary for quick consumption or accessibility
- As part of a workflow to batch-process multiple videos into notebooks
Best practices
- Always use screenshot-before-action flow to confirm the correct modal or textarea is active
- Target specific selectors: textarea[placeholder="Paste any links"] and textarea[placeholder="Paste text here"] to avoid the sidebar
- Inject textarea content by calling the native setter and dispatching input/change events rather than setting .value directly
- Check button enabled state and class names before clicking to avoid no-op actions
- Allow recommended waits: navigation (~3s), source processing (8–10s), audio generation (5–10 min)
Example use cases
- Instructor converts a lecture video into a NotebookLM lesson with speaker bios and references
- Researcher compiles profiles of interviewees from a recorded panel discussion
- Content creator builds source-rich show notes and an audio overview for a published episode
- Accessibility lead generates audio overviews for video content archives
- Team lead batches multiple speaker-focused videos into study notebooks for onboarding
FAQ
Retake a screenshot, clear the sidebar input if present, and re-target the center textarea using the placeholder selector; then re-inject content with the native setter and dispatch input/change events.
Why use synthetic input events instead of setting .value directly?
NotebookLM listens for Angular-style input events. Using the native setter and dispatching input/change ensures the app registers changes and enables action buttons.
How long does audio generation take?
Expect 5–10 minutes for Audio Overview generation; monitor the Studio panel and verify with a final screenshot.
