devinilabs

devinilabs / claude-watch

Public

Turn any tutorial or lecture video into structured study notes — scene-aware frames, persistent library, Claude-vision OCR.

45
13
100% credibility
Found May 12, 2026 at 45 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A Claude AI skill that processes tutorial or lecture videos to create structured markdown study notes with embedded screenshots, timestamped transcripts, and AI synthesis, saved to a personal library.

How It Works

1
💡 Discover claude-watch

You hear about a simple tool that transforms lecture videos into organized study notes with pictures and smart summaries right in your AI chat.

2
📥 Add to your AI assistant

In your Claude chat, you grab the tool with one easy command or download, making it ready for your conversations.

3
🔧 Handle quick setup

It checks your setup and gently guides you to install free helpers for grabbing videos and optional voice-to-text if needed.

4
🔗 Paste video link

Share a YouTube lecture URL or local video file in chat, optionally noting the topic like 'machine learning basics'.

5
⏳ Watch it work

Step away as it pulls the video, captures key scenes with snapshots, grabs the spoken words, and builds your notes.

📚 Get perfect study notes

Return to find a saved folder with a tidy markdown file full of summaries, timestamped highlights, embedded images, code blocks, and review sections – ready for studying anytime.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 45 to 45 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is claude-watch?

Claude-watch is a Python-based Claude plugin that turns any tutorial or lecture video—like a YouTube URL or local file—into structured markdown study notes. Drop a command like `/claude-watch https://youtu.be/ backprop intuition`, and it delivers a notes.md with embedded screenshots from scene changes, timestamped transcripts, TLDR summaries, key concepts, code blocks, and open questions, all synthesized by Claude's vision capabilities. It handles downloads via yt-dlp, scene detection with ffmpeg, free captions or cheap Whisper transcription (Groq preferred), and saves everything to a persistent ~/claude-watch/library for quick re-runs.

Why is it gaining traction?

Unlike basic video tools with uniform frame grabs that waste tokens on static slides, claude-watch uses scene-aware extraction and coverage floors for smarter sampling—up to 80 frames max, tunable resolution for tiny code text. The persistent library and strict notes template mean notes stick around beyond chat sessions, beating ephemeral claude-video chats. Developers dig the "claude watching me code" automation without manual OCR or note-taking drudgery.

Who should use this?

Data scientists bingeing ML lectures, backend devs dissecting AWS tutorials, or frontend folks following UI/UX YouTube series. Ideal for anyone clipping video segments with --start/--end flags to focus on dense 20-30 minute sections, skipping the claude watch cainte of re-watching full vids.

Verdict

Worth a spin for video-heavy learners—solid docs, pytest suite, and MIT license make it dev-friendly despite 45 stars and 1.0% credibility score signaling early maturity. Fork or contribute if you hit 30+ minute limits; otherwise, install via Claude marketplace and level up your claude watch github workflow.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.