Analyze YouTube videos by aligning visual frames with audio transcripts.
npx skills add https://github.com/thiansit/LuPang --skill watch-visual
Please help me install this Agent Skill. Skill: watch-visual Source: https://github.com/thiansit/LuPang/tree/main/.claude/skills/watch-visual Command: npx skills add https://github.com/thiansit/LuPang --skill watch-visual
This Skill helps users learn from YouTube videos when important information is conveyed visually—code on screen, diagrams, UI flows, or demonstrations—that transcripts alone cannot capture, enabling a complete multimodal understanding.
Ask the skill to analyze this YouTube URL and specify the analysis detail level (quick, standard, or detailed).
High-intent search queries and answers about installing and using this skill.
To extract code snippets and diagrams from a YouTube video tutorial, this Skill downloads the video, extracts periodic frames, and applies visual recognition to capture on-screen content, aligning it with the audio transcript for a complete understanding.
Yes, you can analyze YouTube lectures by combining visual frame analysis with audio transcription. This multimodal approach extracts periodic frames and aligns visual recognition results with transcript timestamps to build a complete timeline of spoken and visual content.
You can get a timestamped timeline of visual changes from a video walkthrough by extracting periodic frames and identifying notable visual elements. The Skill aligns these visual changes with transcript timestamps to map demonstrations and actionable steps.
Yes, this video analysis approach works specifically for UI flow demonstrations and on-screen code. It applies OCR and visual intelligence to read on-screen text, identify diagrams, and highlight visual changes essential for learning from walkthroughs.
The best way to perform multimodal video analysis for programming tutorials is extracting periodic frames at configurable intervals and combining OCR visual recognition with audio transcripts. This produces a combined timeline with screenshots and key insights.
FFmpeg is used to download videos and extract periodic frames and audio for YouTube video transcription. The Skill handles frame extraction and audio processing internally to align visual analysis with transcript timestamps automatically.
joegoldinDownload videos and analyze frames and transcripts with yt-dlp and ffmpeg.
zephyrwang6Extract YouTube subtitles and convert them into Chinese transcripts.
Tommy-ywTranscribe spoken audio to text in 99 languages using OpenAI's Whisper model.
michalparkolaDownload YouTube video transcripts via yt-dlp with Whisper fallback.
