repowiserepowise

MCP server

kajisho5 / ffmpeg-skill

Local video editor for coding agents: 42 FFmpeg tools with a machine-readable contract and an MCP server, for Claude Code, Cursor and Codex. Cut, captions, reframe, loudness, platform exports. No cloud, no API keys.

★ 9,437stars on GitHubPythonMIT

This connects Repowise to kajisho5/ffmpeg-skill's code. It does not install ffmpeg-skill.

Repowise maps public repos from their source. Not affiliated with kajisho5. What is Repowise?

github.com/kajisho5/ffmpeg-skillIndexed at 39b9143

What is ffmpeg-skill?

ffmpeg-skill is an Agent Skill that lets AI coding agents such as Claude Code, Cursor and Codex cut, join, measure and export video and audio with local FFmpeg. It is a folder of instructions and scripts that an agent loads, so footage is edited on the user's own machine without being uploaded. The project is written mainly in Python and also provides an MCP server. To install it, run npx ffmpeg-skill setup and follow the steps below. Its code health is 3.3 out of 10 (At_risk), ranked #254 of 4,000+ indexed repos. The last commit to ffmpeg-skill was on Nov 21, 2026, and 8 people have contributed.

As of commit 39b9143, indexed Nov 22, 2026. What it is comes from a model reading the README; the rest is measured from the source and git history, not from the maintainer.

For agentsffmpeg-skill.md

Repo facts: type, license, health, rank, contributors
ffmpeg-skill facts
TypeMCP server
LicenseMIT
Code health3.3 / 10 (At_risk)
Health rank#254 of 4,000+ indexed repos
Last commitNov 21, 2026
Contributors8
Bus factor1
Size300 files, 80,363 lines

Install ffmpeg-skill

Commands copied from the README

  1. 1Install the skill for every agent found on the machine and install FFmpeg when it is missing
    npx ffmpeg-skill setup
  2. 2Check the machine for ffmpeg, ffprobe and the required FFmpeg components
    npx ffmpeg-skill doctor
  3. 3Install the skill for Cursor only
    npx ffmpeg-skill --cursor

Releases on GitHub (latest v4.6.2, Nov 22, 2026)

Full install notes on GitHub

Explore ffmpeg-skill

Ask about this code in your agent

This connects Repowise to ffmpeg-skill's code. It does not install ffmpeg-skill. To install ffmpeg-skill, use the install steps.

Useful if you want to change or fork ffmpeg-skill: your agent can ask how it is put together and get answers that cite its files.

Ask about ffmpeg-skill

Answers come from ffmpeg-skill's README and source, with the files they cite. Two free answers a day without an account.

Featured answer

ffmpeg-skill is an Agent Skill that lets coding agents such as Claude Code, Cursor and Codex cut, join, measure and export video and audio with local FFmpeg, so footage is never uploaded. The repository has 300 files, offers 42 FFmpeg tools with a machine-readable contract and an MCP server, and is written mainly in Python. The installer at bin/install.js is the entry point that copies the skill folder of instructions, scripts and templates into each agent's skill folder. The scripts folder holds one script per operation, scripts/render.py runs a whole project in one pass, and mcp/server.py lets MCP clients list and run the same tools, while evals/run.py grades saved agent runs and the fuzz folder tests the JSON parser and ASR word-timing reader. The layers are Application with 139 files, Config with 63, and Docs & Tooling with 76, so start reading at bin/install.js and then the scripts folder.

Files per layer in ffmpeg-skill3 layers.

Basis: Inferred · knowledge graph· as of 39b9143

Want this for your own codebase? Map your repo free

About this page

Get a page like this for your own codebase

Repowise built this map of kajisho5/ffmpeg-skill from its source and git history. It parses every symbol, draws the dependency graph, scores each file's health from complexity, duplication, coverage and churn, finds hotspots and owners in the history, and writes documentation for every module. Your coding agent reads the same index over MCP, so its answers cite real files.

Free for 2 public repos, with a page like this one. No card. See how Repowise works

Prefer local? pip install repowise && repowise init indexes your own checkout, no account needed. Star Repowise on GitHub

kajisho5/ffmpeg-skill FAQ

What is ffmpeg-skill?

ffmpeg-skill is an Agent Skill that lets AI coding agents such as Claude Code, Cursor and Codex cut, join, measure and export video and audio with local FFmpeg. It runs on the user's machine and does not upload footage.

How does ffmpeg-skill work?

ffmpeg-skill is loaded by an agent as a folder of instructions and scripts, and each media operation has its own script in the scripts folder. It also includes an MCP server at mcp/server.py that lets MCP clients list and run its FFmpeg tools.

How do I install ffmpeg-skill?

Run npx ffmpeg-skill setup to install the skill for each agent found and to install FFmpeg when it is missing. Then run npx ffmpeg-skill doctor to check that ffmpeg, ffprobe and the required components are present. See the install steps

What language is ffmpeg-skill written in?

ffmpeg-skill is written mainly in Python, using the Python standard library. Its installer at bin/install.js is written in JavaScript, and it also includes shell scripts.

How is the ffmpeg-skill code organized?

The bin folder holds the installer, and the scripts folder holds one script per media operation. The mcp folder holds the MCP server, while evals and fuzz hold the grading suite and the fuzzing harness.

Who is ffmpeg-skill for?

ffmpeg-skill is for people who want a coding agent to edit video on their own machine. It works with Claude Code, Cursor, Codex and any MCP client, with no cloud service or API keys required.

What license does ffmpeg-skill use?

ffmpeg-skill uses the MIT license, identified as MIT. The license applies to the whole repository.

Code health

3.3out of 10At risk

On the leaderboards:#71 of 71 Python repos

Key features of ffmpeg-skill

  • Cuts, joins, measures and exports video and audio with local FFmpeg, with no cloud service or API keys.
  • Works with Claude Code, Cursor and Codex as an Agent Skill, and with any MCP client through its MCP server.
  • Creates captions from speech with a local speech engine, so the audio never leaves the machine.
  • Checks the machine with the doctor command, which reports whether ffmpeg, ffprobe and the FFmpeg components the tools need are present.
  • Provides a machine-readable contract that agent frameworks can read with the contract command.

How kajisho5/ffmpeg-skill works

The installer at bin/install.js is the package's executable entry point, and it copies the skill folder of instructions, scripts and templates into the skill folders of Claude Code, Cursor and Codex. The core media work happens in the scripts folder, which holds one script per operation such as audio mixing, backgrounds, smart cutting and ASS overlays, and each script runs on its own from the command line. The scripts/render.py script runs a whole project in one pass. The mcp/server.py server lets MCP clients list and run the same ffmpeg tools. The evals/run.py entry point grades how an agent used the skill, using the transcript and files from each saved run. The fuzz folder holds a harness that searches for inputs that crash the JSON parser and the ASR word-timing reader.

The codebase has 300 files and 4,798 symbols in 10 modules, led by Python, Shell and Dockerfile. Code health is 3.3 out of 10, rated at risk.