claude-video-vision
Claude Code Video Vision is a plugin that gives Claude the ability to watch and understand videos. It extracts frames with ffmpeg and processes audio through one of several backends, then hands Claude the frames as images plus a transcription with timestamps. The README stresses that the plugin is a perception layer, not an interpretation layer: Claude does the interpreting.
Three audio backends are offered. The Gemini API handles speech and non-speech events natively and has a free tier of 1,500 requests a day. A fully local option uses `whisper.cpp` or the Python `openai-whisper`, with models downloaded automatically on first use. The OpenAI Whisper API is the paid option. Frame extraction is adaptive: Claude adjusts fps, time range and resolution to the question, for example low fps for a one-hour lecture or high resolution on a narrow window to read on-screen text.
Install inside Claude Code with `/plugin marketplace add https://github.com/jordanrendric/claude-video-vision` and `/plugin install claude-video-vision`; the Node.js MCP server installs itself via `npx` on first use. Run `/setup-video-vision` for the setup wizard, then `/watch-video path/to/video.mp4 "your question"` or just mention a video file in conversation. It exposes four MCP tools (`video_watch`, `video_info`, `video_configure`, `video_setup`) and needs Node.js 20+ and ffmpeg. The license is MIT and the last push was 2026-10-07. This review is based on the README and repository data; we have not run it ourselves.
Features
Why choose it
Trade-offs
Maintenance check
- Last commit
- 2026-10-06
- Commits, last 90 days
- 17
- Releases, last 90 days
- 1
- Latest release
- v1.4.0 (2026-10-06)
- Open issues
- 9
- Recent issues with at least one comment
- 7/11
Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.
Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.
Most discussed open issues
- #48 Duplicate MCP server instances spawn after reconnect, causing Claude Code Task dispatches to be silently rejected1 comments · 2026-05-18
- #41 Whisper.cpp backend: support audio tagging (Whisper-AT or parallel classifier) for music / non-speech annotation1 comments · 2026-05-15
- #58 video_analyze: 600s ffmpeg timeout silently returns empty analysis — decode cost is unbounded at native resolution (likely root cause of #470 comments · 2026-09-07
Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.
Compatibility
Quick start
Use cases
Alternatives
Comments
No comments yet. Be the first!