claude-video
Give Claude and other agents the ability to watch and transcribe videos.
Extract frames, pull transcripts, and analyze media files directly inside your development environment.
Claude-video turns local files and URLs into timestamped frames and transcripts that AI agents can analyze. It suits developers and video professionals who want automated visual inspection and transcription inside tools like Claude Code and Codex.
What it is
The tool downloads video streams, extracts individual frames using FFmpeg and yt-dlp, and coordinates native or fallback transcriptions via WhisperX or Groq/OpenAI. It functions as an agent skill, allowing local LLMs to ingest video evidence for summarization and technical queries.
Who it's for
It suits developers and technical video producers running local agent environments like Claude Code or Codex. It does not work in browser-based AI chats or Cowork apps, and requires comfort with CLI tools and dependency management.
Key features
Agent Integration
Installs directly into Claude Code, Codex, and other Agent Skills hosts.
Dual Engine Processing
Route videos to Google AI Studio for fast analysis or run local FFmpeg extraction.
Flexible Transcription
Uses native captions or falls back to local WhisperX, Groq, or OpenAI.
Timestamped Summaries
Outputs frame-accurate descriptions mapped directly to video timestamps.
Up and running in 3 steps
Our honest verdict
What's great
- Integrates natively into popular developer environments like Claude Code and VS Code
- Supports direct video URL ingestion from YouTube and other web sources
- Offers a fast Gemini API path for rapid video ingestion without local hardware overhead
- Performs local frame extraction and transcription via FFmpeg and yt-dlp
Where it falls short
- Requires manual installation of external system dependencies like Python 3.10+, FFmpeg, and yt-dlp
- Does not work in browser apps, Claude Chat, or Cowork environments
- Local WhisperX transcription and video processing demand local machine resources and setup time
- Cloud fallback requires an external API key from Google AI Studio, Groq, or OpenAI
Pricing & license
The software is free under its MIT licence. The only financial costs depend on whether you use external APIs like Google AI Studio, Groq, or OpenAI for transcription and model analysis.
Open remote AI-training roles
876 open remote AI-training roles, updated November 15, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does this work in standard Claude Chat or web browser applications?
No. The README explicitly states that Watch does not work in Claude Chat, Cowork, or browser apps; it requires local agent environments like Claude Code or Codex.
What external tools do I need to run local video processing?
You need Python 3.10+, FFmpeg/ffprobe, and a current yt-dlp setup installed on your machine.
How can I avoid installing local transcription models?
You can use the fastest path by providing a free Google AI Studio key so Gemini handles the video processing directly.
