NewuxtremeNewuxtremeCommunity·1 Agent Skills Included

watch-video-skill

Extract transcripts and frames from videos into structured notes

Extracts time-synced transcripts and auto-scaled still frames from YouTube, TikTok, Vimeo, and local video files. Combines spoken words with on-screen visuals to produce structured markdown notes with summaries, timelines, and key quotes. Eliminates manual video scrubbing and note-taking while keeping token costs low with a 100-frame cap.
npx skills add Newuxtreme/watch-video-skill --all -g -y

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install watch-video-skill?

Run `npx skills add Newuxtreme/watch-video-skill --all -g -y` in your terminal to install it globally. You also need ffmpeg, yt-dlp, and Python 3.9 or later on your system.

How to make Claude watch a video?

Type the slash command `/watch-video` followed by a video URL or local file path. The skill extracts frames and a transcript so Claude can read both and write structured notes.

Which video sites does watch-video support?

It works on YouTube and every site yt-dlp supports, including Vimeo, TikTok, X, Twitch, Loom, and Instagram. It also handles local files like mp4, mov, mkv, webm, and avi.

Does it work on videos without captions?

Yes. If no captions exist, it sends the audio to Whisper via the Groq or OpenAI API. Without an API key, you still get frames-only output.

How do I analyze just one part of a long video?

Pass a time range like `--start 2:00 --end 3:00` to focus the pipeline on that section. It packs more frames into that window for denser detail.

Related Repositories in Content & Communication

View All in Content & Communication