Video Use: AI Video Editing for Claude Code — Drop Footage, Get Final Cut

By Prahlad Menon 1 min read

The Browser Use team just dropped Video Use — a skill that turns Claude Code into a video editor. Drop raw footage in a folder, describe what you want, get final.mp4 back.

No timeline. No menus. No presets. Just conversation.

Repo: github.com/browser-use/video-use

What It Does

Point Claude Code at a folder of raw takes:

cd /path/to/your/videos
claude
> edit these into a launch video

The agent:

  1. Inventories your source files
  2. Proposes an editing strategy
  3. Waits for your OK
  4. Produces edit/final.mp4

Works for any content type — talking heads, montages, tutorials, travel vlogs, interviews.

Automatic Editing Features

FeatureWhat It Does
Filler word removalCuts “umm”, “uh”, false starts, dead space between takes
Color gradingAuto-grades every segment (warm cinematic, neutral punch, or custom ffmpeg chain)
Audio smoothing30ms fades at every cut — no pops
SubtitlesBurns in your style — 2-word UPPERCASE chunks by default, customizable
Animation overlaysGenerates via HyperFrames, Remotion, Manim, or PIL — parallel sub-agents
Self-evaluationChecks rendered output at every cut boundary before showing you
Session memoryPersists to project.md so next week’s session picks up where you left off

Installation

One-Liner (Claude Code, Codex, Hermes, OpenClaw)

Paste this into your agent:

Set up https://github.com/browser-use/video-use for me.

Read install.md first to install this repo, wire up ffmpeg, register the skill with whichever agent you're running under, and set up the ElevenLabs API key — ask me to paste it when you need it. Then read SKILL.md for daily usage, and always read helpers/ because that's where the editing scripts live. After install, don't transcribe anything on your own — just tell me it's ready and wait for me to drop footage into a folder.

The agent handles clone, dependencies, skill registration, and prompts you for your ElevenLabs API key.

Manual Install

# Clone and symlink
git clone https://github.com/browser-use/video-use ~/Developer/video-use
ln -sfn ~/Developer/video-use ~/.claude/skills/video-use  # Claude Code
# ln -sfn ~/Developer/video-use ~/.codex/skills/video-use # Codex

# Install deps
cd ~/Developer/video-use
uv sync  # or: pip install -e .
brew install ffmpeg  # required
brew install yt-dlp  # optional, for downloading online sources

# Add ElevenLabs key
cp .env.example .env
$EDITOR .env  # ELEVENLABS_API_KEY=...

How It Works (The Smart Part)

The LLM never watches the video. It reads it through two layers:

Layer 1: Audio Transcript (Always Loaded)

One ElevenLabs Scribe call per source gives:

  • Word-level timestamps
  • Speaker diarization
  • Audio events ((laughter), (applause), (sigh))

All takes pack into a single ~12KB takes_packed.md:

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

Layer 2: Visual Composite (On Demand)

timeline_view produces a filmstrip + waveform + word labels PNG for any time range. Called only at decision points:

  • Ambiguous pauses
  • Retake comparisons
  • Cut-point sanity checks

The efficiency gain:

  • Naive approach: 30,000 frames × 1,500 tokens = 45M tokens of noise
  • Video Use: 12KB text + a handful of PNGs

Same philosophy as Browser Use giving an LLM a structured DOM instead of a screenshot — but for video.

The Pipeline

Transcribe → Pack → LLM Reasons → EDL → Render → Self-Eval
                                                    │
                                        issue? fix + re-render (max 3)

The self-eval loop runs timeline_view on the rendered output at every cut boundary — catches visual jumps, audio pops, hidden subtitles. You see the preview only after it passes.

Design Principles

  1. Text + on-demand visuals — No frame-dumping. Transcript is the surface.
  2. Audio is primary, visuals follow — Cuts come from speech boundaries and silence gaps
  3. Ask → confirm → execute → self-eval → persist — Never cuts without strategy approval
  4. Zero assumptions about content type — Look, ask, then edit
  5. 12 hard rules, artistic freedom elsewhere — Production-correctness is non-negotiable. Taste isn’t.

Always-On Editing

For editing from your own VPS or via Telegram, run the agent through Browser Use Box. There’s a 15-second demo on TikTok.

Compared to Other AI Video Tools

Video Use is specifically designed for coding agents — it’s a skill, not a standalone app. This makes it:

  • Composable — Chain with other skills (research, animation, publishing)
  • Scriptable — Automate batch editing workflows
  • Persistent — Session memory means long-running projects work

For other AI video approaches we’ve covered:

  • Pixelle-Video — Full generation pipeline (script → images → TTS → video)
  • HyperFrames — Code-to-video for animations and explainers
  • OpenCut — Open-source CapCut alternative with timeline UI
  • Palmier Pro — AI-assisted traditional editor
  • Director by VideoDB — Framework for video agent pipelines

Video Use fills the “raw footage → polished edit” gap that the others don’t address.

Requirements

  • Claude Code, Codex, Hermes, OpenClaw, or any agent with shell access
  • ffmpeg — For all video operations
  • ElevenLabs API key — For transcription (Scribe)
  • yt-dlp (optional) — For downloading online sources

Limitations

  • Requires ElevenLabs subscription for transcription
  • Best for speech-driven content (relies heavily on transcript)
  • Self-eval adds rendering time (but catches issues)

Links: