← All skills

Claude Video

Gibt Claude die Fähigkeit, jedes Video zu sehen: /watch lädt herunter, extrahiert Frames, transkribiert und übergibt alles an Claude.

View on GitHub Media & Video Health: Excellent Security: Passed

7.8k stars 869 forks Python MIT Updated 2 days ago

Install

Skills CLI (Claude Code, Codex, Cursor, Gemini CLI and more). Or straight from the GitHub repo.

Health

Excellent

This skill is actively maintained.

Hygiene
Present: readme, license, description. Missing: contributing guide.
Maintenance
Last commit 2 days ago. 420 commits in the past year. Last release 11 days ago.
Responsiveness
Closed 82% of 180 issues. 8 contributors active in the past year.
Adoption
7.8k stars on GitHub.

Security check

Passed
  • No outbound network calls

    Heuristik: erwartete Aufrufe an Video-Plattformen und die Whisper-API; keine Hinweise auf Exfiltration an fremde Hosts.

  • No obfuscated code

    Heuristik: kein verschleierter Code (kein eval-auf-Dekodiertem, keine großen Base64-Blöcke) im geprüften Code.

  • No hardcoded secrets

    Heuristik: keine hartcodierten API-Keys, Tokens oder Zugangsdaten gefunden.

Automatically checked with a heuristic (no AI model configured). No guarantee — review before use.

About

Claude Video

Give Claude the ability to watch any video. /watch downloads a video, extracts frames, transcribes it, and hands everything to Claude.

Features

  • Downloads from YouTube, TikTok, Vimeo, local files and 500+ sources via yt-dlp
  • Scene-aware or keyframe-based frame extraction
  • Free native caption extraction, with a Whisper API fallback
  • Configurable detail modes (transcript, efficient, balanced, token-burner)
  • Frame deduplication to keep token costs low
  • Auto-fps calculation based on video duration

Installation

Install via the Claude Code marketplace, the Agent Skills CLI (npx skills add bradautomates/claude-video), by uploading to claude.ai, or manually via symlink.

Usage

/watch https://www.youtube.com/watch?v=...

Claude downloads the video, extracts frames and transcript, and can then answer questions about what happens on screen.