AIBit-Discover Open Source Projects AIBit-Discover Open Source Projects

video-use: Edit Videos with Claude Code

Discover video-use, the open-source Claude Code skill that transforms raw footage into polished final.mp4 files. Drop your videos in a folder, chat with Claude, and get professional edits with auto color grading, filler word removal, smooth audio fades, custom subtitles, and Manim animations. No presets needed—works for talking heads, tutorials, interviews, and more. Learn how this smart agent reads transcripts and visual composites to deliver production-ready videos without ever 'watching' the footage.

video-use: Revolutionize Video Editing with Claude Code

Imagine dropping raw footage into a folder, chatting with an AI, and getting a polished final.mp4 back—complete with perfect cuts, cinematic color grading, and custom subtitles. That's video-use, a groundbreaking 100% open-source Claude Code skill that's changing video production forever.

What Makes video-use Different?

Unlike traditional editors with rigid presets, video-use works for any content type:

  • Talking head interviews
  • Tutorial montages
  • Travel vlogs
  • Corporate explainers

Key features:

  • Filler word removal (umm, uh, false starts)
  • Auto color grading (warm cinematic, neutral punch, custom FFmpeg)
  • 30ms audio fades at every cut
  • Burned subtitles in your style
  • Animation overlays via Manim/Remotion/PIL
  • Self-evaluation at every cut boundary
  • Session memory persistence

How the Magic Happens (Without 'Watching' Video)

The LLM never processes raw frames. Instead, it uses a two-layer system:

Layer 1: Audio Transcript (~12KB takes_packed.md)

C0103 (43.0s, 8 phrases)
[002.52-005.36] S0 Ninety percent of what a web agent does...
[006.08-006.74] S0 We fixed this.

Layer 2: Visual Composite (timeline_view PNGs)—generated only when needed for cut decisions.

Pipeline: Transcribe → Pack → LLM Reasons → EDL → Render → Self-Eval → Repeat (max 3x)

Get Started in 3 Minutes

git clone https://github.com/browser-use/video-use
cd video-use
ln -s "$(pwd)" ~/.claude/skills/video-use
pip install -e .
brew install ffmpeg yt-dlp  # macOS
cp .env.example .env  # Add ElevenLabs API key

Navigate to your video folder and run:

cd /path/to/raw/videos
claude

In Claude: "edit these into a launch video"

The skill inventories sources, proposes strategy, gets your OK, then delivers edit/final.mp4.

Production-Ready Design Principles

  1. Text-first: Transcript is the primary surface
  2. Audio-primary: Cuts follow speech boundaries
  3. Ask-confirm-execute: Strategy approval required
  4. 12 hard rules: Production correctness guaranteed
  5. Zero content assumptions: Adapts to any footage

Check SKILL.md for the complete production rules.

Why This Matters

video-use proves LLMs can handle complex creative workflows through structured interfaces (like browser-use does for web). With 414+ stars and growing, it's becoming the go-to for AI-powered video editing.

Try it today and never waste time on filler cuts again. Perfect for creators, marketers, educators, and anyone who makes videos.