Watch Video

Give Claude eyes and ears on any online video, then analyze it. Downloads a video from a URL (YouTube, Instagram, X, Vimeo, and 1000+ yt-dlp sites), extracts key frames, and gets a transcript (native captions, else local Whisper), then hands the frames + timestamped transcript to Claude so it can produce a full breakdown — hook, format, storytelling structure, pacing and visual layout — grounded in what's actually on screen AND said. Pass several URLs to batch-analyze dozens at once and surface the patterns the best ones share. Use whenever the user shares a video link (or many) and wants it w

8 stars 1 forksImages, Video & Audio

Install

skills CLI (Claude Code, Cursor, Codex and more)

npx skills add Ootto-AI/ootto-watch --skill watch-video

Manual install

git clone --depth 1 https://github.com/Ootto-AI/ootto-watch.git /tmp/ootto-watch
mkdir -p ~/.claude/skills
cp -r /tmp/ootto-watch/. ~/.claude/skills/watch-video

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# watch-video — let Claude actually watch a video

Most tools only read a transcript and miss everything visual. This skill gives Claude the **frames + the words, aligned on one timeline**, so it can reason about cuts, on-screen text, demos, and pacing — not just the audio.

## How to use it

When the user gives a video URL:

1. Run the helper (it downloads, samples frames, and transcribes):
   ```
   python scripts/watch.py "<VIDEO_URL>" --out watch_out
   ```
   Useful flags: `--max-frames 60` (cap frames), `--no-whisper` (captions only, fully free/offline), `--every 0` (auto-pace by duration).

2. It writes:
   - `watch_out/frames/f000.jpg …` — sampled frames (named by index)
   - `watch_out/transcript.txt` — the spoken words
   - `watch_out/index.json` — `[{frame, t, file}]` so each frame maps to its timestamp

3. **Read the frames as images** (use the Read tool on the PNG/JPGs) alongside `transcript.txt`, lining each frame up with what's being said at its `t`. Then answer the user's request (summary, breakdown, hook analysis, study notes, a fresh script, etc.) grounded in BOTH the visuals and the audio.

## Analyze — the full breakdown

Once `watch.py` has run, read the frames + `transcript.txt` aligned by `index.json`, then write a structured breakdown. Ground every point in a specific frame `t` and/or transcript line:

- **Hook (first ~3s)** — exactly what's on screen and said in the opening, and *why* it stops the scroll.

Read the full skill on GitHub

Details

Skill name: watch-video
Repository: Ootto-AI/ootto-watch
License: MIT
Repo updated: 13 August 2026

FAQ

What does the Watch Video skill do?

How do I install the Watch Video skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add Ootto-AI/ootto-watch --skill watch-video. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Watch Video popular?

It lives in Ootto-AI/ootto-watch, which has 8 GitHub stars and 1 forks.

Related skills

Aeon Music Video

AEON-7/aeon-music-video

Compose audio-reactive music videos from existing video clips and an audio track. USE THIS SKILL when the user wants to: build a music video from clips that react to the song's beats / onsets / spectral content, create rapid-cut hard-cut edits synced to a song, build a DMT-flash / Tron-aesthetic video, or compose a video where visuals respond to audio dynamics. Two scripts: reactive_compositor.py (general reactive editing across mood-bucketed clips) and dmt_flash_editor.py (10-cut hard-cut editor with chromatic aberration, bloom, and aggressive per-beat post-processing).

Images, Video & Audio

Video Vision

Moh4696/claude-video-vision

Watch and understand local video files. Extracts sampled frames with ffmpeg and generates a timestamped transcript with local Whisper, then answers the user's question grounded in both visual and audio evidence. Use this skill whenever the user references a video file (.mp4, .mov, .mkv, .webm, .avi, .m4v) and wants it watched, summarized, analyzed, described, searched, captioned, transcribed, clipped, or asks questions like "what happens in this video", "what is said at 1:30", "find the part where X", "summarize this recording / lecture / meeting / clip", even if they don't explicitly say "wat

Images, Video & Audio

AI Video

0xadvait/ai-video-skill

Generate AI video of any kind, end to end — write the prompt, create first-frame stills, run the generation across six video models (Seedance 2.0, Kling 3.0, Wan 2.7, Veo 3.1, OmniHuman 1.5, VEED Fabric), synthesize voiceover, stitch clips into a finished cut, and upscale for delivery. Use whenever the user wants to make or improve a video: cinematic shots, character dialogue, action, dance, native SFX, image-to-video, motion transfer, lip-sync, talking-avatar / "a video of me", real-person parody, abstract motion graphics, montages and stitched episodes. Builds prompts with the canonical 5-pa

Images, Video & Audio

awrshift/skill-telegram

The only Telegram skill your AI agent needs. Read messages, parse channels, download and transcribe voice notes — all via Telethon user API. One session file, zero bots. Use when user says 'read telegram', 'check saved messages', 'fetch from telegram', 'parse channel', 'transcribe voice', 'telegram search', 'check my telegram', or when any agent needs Telegram data. Also trigger on 'what did I save in telegram', 'read my saved messages', 'get messages from channel X', 'download voice notes'. Covers: ad-hoc reading, channel parsing with sanitization, voice/audio download. Requires TG_API_ID, TG

Images, Video & Audio

Cue Omni Reader

sensedeal/cue-skills

Use when the user wants an external AI agent to parse or understand an HTTP(S) URL or an authorized local document, audio, or video source through Cue Omni Reader.

Images, Video & AudioIn 1 list

Mega Shot

stevenguyenbt/mega-shot

Makes finished narrated videos from an idea, driven by one video.yaml: researched, fact-checked script; narration in a cloned or built-in voice (local VoiceStudio TTS, voice cloning from any audio/video file); visuals in a style chosen with the user (stock footage and archival images, 2.5D parallax, whiteboard, flat 2D, kinetic type, 35 painterly styles, Three.js or Blender 3D, AI images, news graphics, satellite imagery, code-drawn story characters); music mix at -14 LUFS; high-quality export plus phone preview; automatic quality checks; English subtitles (SRT); YouTube thumbnails; publishing

Images, Video & Audio

Browse the most popular Claude skills, all images, video & audio skills, or write your own with the skills generator.