video-use
Browser Use's conversational video editing Skill for cutting pauses, filler, and retakes, adding subtitles and color treatment, and reviewing rendered cuts from transcripts and on-demand timeline views.
Project overview
Browser Use's conversational video editing Skill for cutting pauses, filler, and retakes, adding subtitles and color treatment, and reviewing rendered cuts from transcripts and on-demand timeline views. Its main documented capabilities are Talking-head and interview rough cuts, Subtitles and color treatment, Timeline visual inspection, Rendered-output review. This page is based on public GitHub material and does not present repository claims as hands-on testing.
Core capabilities
Talking-head and interview rough cuts
Uses transcripts to identify pauses, filler, false starts, and retakes and produce reviewable edit decisions.
Subtitles and color treatment
Burns customizable subtitles and applies FFmpeg-based color treatment to segments.
Timeline visual inspection
Inspects ambiguous pauses and cut points in timeline views containing waveforms, word labels, and filmstrips.
Rendered-output review
Checks rendered cut boundaries for visual jumps, audio pops, and hidden subtitles, then repairs and rerenders when needed.
Installation and usage
Clone the complete repository to a stable path and link the whole directory into the agent's Skills directory. Install Python dependencies, confirm FFmpeg, and configure the ElevenLabs transcription key only with user authorization.
Let an AI Agent install it
Send this prompt to Codex, Claude Code, or another AI agent that can work with your local environment.
Help me install video-use from https://github.com/browser-use/video-use. Read install.md, SKILL.md, helpers/, pyproject.toml, and LICENSE first. Clone the complete repository to a stable path, register the whole directory with the current agent, install Python dependencies, and confirm FFmpeg/ffprobe. Ask before requesting an ElevenLabs API key, installing system software, writing `.env`, replacing a directory, or transcribing billable media, and never echo the key. Perform only the quota-free help and ffprobe checks until I provide a test video.- Python with uv or pip
- FFmpeg and ffprobe
- An ElevenLabs API key
- An agent with a Skill directory or system-prompt imports
git clone https://github.com/browser-use/video-use ~/Developer/video-use
ln -sfn ~/Developer/video-use "${CODEX_HOME:-$HOME/.codex}/skills/video-use"
cd ~/Developer/video-use
uv sync- 1Review prerequisites and boundaries
Read the repository installation material and confirm: Python with uv or pip; FFmpeg and ffprobe; An ElevenLabs API key; An agent with a Skill directory or system-prompt imports. Stop for confirmation before credentials, cost, additional permissions, or replacement.
- 2Run the documented installation
Use the current README command: git clone https://github.com/browser-use/video-use ~/Developer/video-use ln -sfn ~/Developer/video-use "${CODEX_HOME:-$HOME/.codex}/skills/video-use" cd ~/Developer/video-use uv sync
- 3Run a minimal verification
Run `python ~/Developer/video-use/helpers/timeline_view.py --help` and `ffprobe -version`. A real transcription consumes Scribe quota, so wait for a user-provided test clip and confirmation before running it.
- 4Move to a real task
After the minimal check passes, use nonsensitive test media first. Ask separately before external uploads, billable calls, publishing, or irreversible changes.
Run `python ~/Developer/video-use/helpers/timeline_view.py --help` and `ffprobe -version`. A real transcription consumes Scribe quota, so wait for a user-provided test clip and confirmation before running it.
- ElevenLabs Scribe transcription consumes quota and should not run automatically during setup
- Audio is sent to ElevenLabs for processing, so sensitive footage requires a data-boundary review
- The README's manual examples use Unix/macOS paths; Windows requires adapted linking and dependency steps
- No installation or real edit was performed for this catalog entry
Use cases
Talking-head cleanup
Keep the best takes while removing filler, dead space, and false starts.
Tutorials and interviews
Structure footage from word-level transcripts, then add subtitles, color treatment, and audio fades.
Cut quality review
Use timeline views and rendered-boundary checks to reduce visual jumps and audio pops before delivery.
Assessment
Based on the README, installation material, and repository structure, video-use turns Talking-head and interview rough cuts, Subtitles and color treatment, Timeline visual inspection into explicit agent workflows. Its limitations are mainly dependency, host, and data-boundary concerns. This assessment is based on public material, without installation or a real media task.
Why it may be useful
- Uses transcripts to identify pauses, filler, false starts, and retakes and produce reviewable edit decisions.
- Burns customizable subtitles and applies FFmpeg-based color treatment to segments.
- Inspects ambiguous pauses and cut points in timeline views containing waveforms, word labels, and filmstrips.
What to know first
- ElevenLabs Scribe transcription consumes quota and should not run automatically during setup
- Audio is sent to ElevenLabs for processing, so sensitive footage requires a data-boundary review
- The README's manual examples use Unix/macOS paths; Windows requires adapted linking and dependency steps
- No installation or real edit was performed for this catalog entry
README
video-use
Overview
Browser Use's conversational video editing Skill for cutting pauses, filler, and retakes, adding subtitles and color treatment, and reviewing rendered cuts from transcripts and on-demand timeline views. Edit videos with coding agents
Getting started
- Clone the complete repository to a stable path and link the whole directory into the agent's Skills directory. Install Python dependencies, confirm FFmpeg, and configure the ElevenLabs transcription key only with user authorization.
- Project focus: Browser Use's conversational video editing Skill for cutting pauses, filler, and retakes, adding subtitles and color treatment, and reviewing rendered cuts from transcripts and on-demand timeline views.
- Published characteristics: Word-timestamp-driven editing; On-demand timeline views; Cut-boundary self-evaluation with up to three repairs; Requires FFmpeg and an ElevenLabs API key.
- Run `python ~/Developer/video-use/helpers/timeline_view.py --help` and `ffprobe -version`. A real transcription consumes Scribe quota, so wait for a user-provided test clip and confirmation before running it.
Configuration
git clone https://github.com/browser-use/video-use ~/Developer/video-use
ln -sfn ~/Developer/video-use "${CODEX_HOME:-$HOME/.codex}/skills/video-use"
cd ~/Developer/video-use
uv sync