claude-watch

Convert tutorial and lecture videos into structured markdown study notes.

91|26|Updated May 3, 2026
One-click install
npx skills add https://github.com/devinilabs/claude-watch --skill claude-watch
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: claude-watch
Source: https://github.com/devinilabs/claude-watch/tree/main
Command: npx skills add https://github.com/devinilabs/claude-watch --skill claude-watch

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ffmpeg, ffprobe, yt-dlp, and includes scripts (resource) components.

What problem does it solve?

This Skill removes the manual work of watching long videos while taking notes, so learners can return to a clean, searchable summary instead of scrubbing through the recording.

Core Features & Use Cases

  • Downloads a video from a URL or uses a local file.
  • Detects scene changes and keeps coverage through long static sections with screenshot frames.
  • Pulls captions first and falls back to Whisper when captions are missing.
  • Writes a structured notes file with TLDR, key concepts, per-scene notes, code blocks, diagrams, and open questions.
  • Saves everything in a persistent library so re-running the same source reuses the cached artifact.
  • Use it for lecture review, tutorial recap, coding walkthroughs, and focused study passes.

Quick Start

Ask the skill to watch a tutorial or lecture video from a URL or local file and produce a structured markdown study-notes file with screenshots, timestamps, and transcript synthesis.

Frequently Asked Questions about claude-watch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn lecture videos into study notes with timestamps and screenshots?

You can turn lecture videos into study notes by downloading the video, detecting scene changes for screenshots, extracting captions, and generating a structured markdown file with timestamped per-scene sections and key concepts.

Can I generate video notes from local recordings or only from URLs?

You can generate video notes from both local recordings and URLs. The tool processes local files directly and downloads online videos using yt-dlp to produce the same structured markdown study notes.

Do I need ffmpeg and yt-dlp installed to extract transcripts from tutorial videos?

Yes, you need ffmpeg and yt-dlp installed to download videos and detect scenes. For transcript extraction, the tool uses available captions first and falls back to Whisper transcription if captions are missing.

What is the best way to review a coding walkthrough video without scrubbing through the recording?

The best way to review a coding walkthrough is to generate a structured markdown summary containing a TLDR, key concepts, code blocks, and open questions. This creates a searchable recap with timestamped screenshots.

Are cached video transcripts and notes reused if I process the same lecture twice?

Yes, cached library outputs are reused when re-running the same video source. This persistent library saves the generated notes, screenshots, and transcripts so repeated processing is skipped.

Why does my video notes output include screenshots from long static sections without scene changes?

Screenshots are captured from long static sections to maintain coverage. Even without scene changes, the tool periodically extracts frames to ensure the transcript summary has visual context throughout the video.