video-use

Edit videos by chatting with Claude Code: drop raw footage in a folder and get a final.mp4 back.

Stars: 30.4k
Forks: 3.5k
Licence: MIT
Language: Python
Last update: 28 days ago
Type: AI model / agent

What it is

video-use is a skill for Claude Code (and similar coding agents) that edits video from plain-language requests. You put raw clips in a folder, describe the edit you want, and the agent transcribes the audio, proposes a plan, cuts filler words and dead space, grades color, adds fades and subtitles, and renders a final video. The README says the model reads a text transcript rather than watching every frame, which keeps it cheap to run.

Turn a folder of raw takes into a single launch or explainer video with one chat request
Remove umm, uh, and false starts from an interview recording
Add burned-in uppercase subtitles in two-word chunks to a short clip
Generate simple animation overlays through HyperFrames, Remotion, Manim, or PIL

Great for

Creators who have talking-head, interview, or tutorial footage and want a rough cut quickly
People already using Claude Code or Codex who want video editing inside the same chat
Users comfortable installing ffmpeg and an API key from the command line
Anyone who wants a session memory file so later edits pick up earlier context

Not ideal for

People who want a point-and-click editor with a timeline interface, since the README describes a chat-driven workflow
Users who cannot or will not create an ElevenLabs API key, because transcription depends on the ElevenLabs Scribe service
Anyone needing fully offline processing, since the README relies on an external transcription API
Users who want a hosted product now, as the README says the cloud version is only a waitlist

Step-by-step guide: how to use video-useInstall, first run, examples, tips and FAQs

How to get it

1
Clone and build from source
You'll need: Claude Code, Codex, or another agent with shell access · ffmpeg · uv or pip · ElevenLabs API key
Clone and symlink into Claude Code skills directory
git clone https://github.com/browser-use/video-use ~/Developer/video-use && ln -sfn ~/Developer/video-use ~/.claude/skills/video-use
Install Python dependencies
cd ~/Developer/video-use && uv sync
Install ffmpeg (macOS, Homebrew)
brew install ffmpeg
Add ElevenLabs API key
cp .env.example .env
2
Run it
- Complete the install and add your ElevenLabs API key to .env
- Put raw video files in a folder and change into that folder
- Start Claude Code in that folder with the claude command
- Ask for an edit, such as 'edit these into a launch video', and confirm the strategy the agent proposes
- Find the result in the edit/ subfolder next to your sources
3
Get the most out of it
- Read SKILL.md for the production rules and editing craft the README points to
- Use the session memory in project.md so follow-up edits build on earlier sessions
- Specify a custom ffmpeg color grade or subtitle style if the defaults do not fit
- Install yt-dlp if you want to pull in online source footage, as the README suggests

Downloads

Source code (.zip)main branch · straight from GitHub Official website & docsvideo-use.com

Strengths

Works from plain-language instructions without presets or menus, according to the README
Uses a text transcript as its main input, which the README says avoids sending every frame to the model
Renders a preview only after its own self-check of cut boundaries passes

Trade-offs

Depends on an ElevenLabs API key and ffmpeg, so setup involves several external services and tools
The README does not describe output quality limits or supported video formats in detail
The cloud-hosted version is not available yet, only a waitlist

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.