music-mv skill
A skill that makes a music video for your song: every frame is drawn by code and six steps with pauses lead from analysis to edit
- 130 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs analysis and render scripts, downloads a renderer and browser on first run and writes files into the project working folder.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/music-mvDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs analysis and render scripts, downloads a renderer and browser on first run and writes files into the project working folder.
This is third-party code. Review the repository files before installing.
What it does
music-mv is a skill and a tutorial for making a clip in code. No video model is involved: the agent draws each frame in code and assembles the clip through HyperFrames and FFmpeg. You tell the agent to make a clip for song.mp3 and it runs six steps: analyses the song, picks a style, draws a character, writes the script and storyboard, shows a draft and only then builds the full clip, then revises after review. At each key step the agent pauses for your approval. There are four styles with their own drawing techniques: pixel game, paper collage, crayon picture book and a paper pop-up book. Scripts detect tempo, beats and song segments, make contact sheets for review and put the original audio back into the finished clip.
Who it is for. Musicians and short-video authors who need a clip for their own song without video generation or an editing app.
Good fit when
- The song is finished and there is no budget for shooting or commissioned animation.
- You need a pixel or paper-style clip with lyrics shown in time.
- The clip must follow the rhythm: cuts tied to beats and track segments.
- You want to check each step yourself: a 5-second draft first, then the whole clip.
Not a fit when
- You need photorealistic footage: the result is built from code and looks drawn.
- ffmpeg, Node.js or Python are missing: the analysis and render scripts will not run without them.
- The agent runs a weak model: the author notes that a stronger model yields finer imagery.
Example request
Make a clip for song.mp3 in a paper collage style: first analyse the song and show the style and character.Limitations
The documentation is in Chinese with an English caption under the title only. Python 3.9 or newer with numpy, librosa and Pillow, ffmpeg and Node.js 18 or newer are required. The HyperFrames renderer and a headless browser download on first run. Watercolor and crayon frames need puppeteer-core in the project. Lyrics can be transcribed with Whisper or you prepare a lyrics.srt file yourself. Rights to the song stay with its author and the repository contains no clips with audio. The latest commit is from October 2026.
How to disable. Delete the installed skill folder from the agent's skills directory, for example `~/.claude/skills/music-mv`. The downloaded HyperFrames renderer and Chrome are removed separately, as are clip working folders.
FAQ
Which steps does the agent follow?
Six: song analysis, style choice, character drawing, script and storyboard, a draft before the full build, then review and edits. The agent waits for your approval at each important step.
Which styles exist?
A 320 by 180 pixel game with a 24-color palette, paper collage, p5.brush crayon drawing and a layered paper pop-up book with shadows.
How do I check the result?
`contact_sheet.py` builds a contact sheet of frames at a set interval, `video_rhythm.py` checks cuts, static stretches and black frames, and `finalize_audio.py` restores the original audio.
Does it work without installing the skill?
Yes. Instead of the skill you can use the tutorial file from the repository and send the prompts to the agent in parts.
Which agents does it work with?
Any agent that reads and writes files and runs commands, including Claude Code and Codex. The author states that stronger models draw in finer detail.
Discussion
No questions or reviews yet.
Related
SkillsFour skills turn a song with lyrics into a watercolour animated music video: storyboard, parallel painters and render.
A skill that turns your narration and real assets into a 20–60 second video in the style you pick, rendered locally as MP4.
motion-video-production: a skill for motion videos
motion-video-production
A skill that takes an agent through seven stages for a 15 to 90 second video, a script checking each stage before the next
A skill that makes promo videos from code: direction, in-browser animation with motion blur, its own score and a 60 fps MP4
A skill that takes Claude Code from a song and lyrics to a finished Vocaloid-style lyric video, rendered in a browser.
The Video by Reference skill learns style from a sample clip and builds an animated short from code with an original soundtrack
README in short
The repository offers two modes: a skill and a step-by-step tutorial. Two example clips are shown, and each style has a sample scene with source code.