Open source / Reel
Edit reels with Claude
You record yourself talking on your phone, and an agent builds animated graphics on top, timed to what you say: karaoke subtitles, camera moves, music and sound effects.
Free and open source. Honestly, it’s not zero setup: you need a render environment, details below.
What it does for you
From phone footage to a finished reel
Timed graphics
Cards, mockups and numbers come in one at a time, right on the word you’re saying. Never a panel that assembles all at once.
Karaoke subtitles
Word by word, the one you’re saying lights up as you say it. Synced to the real audio, not by eye.
Camera moves
Digital zooms that follow the rhythm of your speech: a push on the payoff, room to breathe on the line. The footage stays still, only the framing moves.
Music and sound effects
A music bed under your voice and a sound effect at every graphic change, balanced so they never cover what you say.
Your own style skill
From the second reel on, you start from your own choices: palette, fonts, how the cards come in. Never from scratch again.
Free and open source
The engine is HyperFrames (open source) plus ffmpeg and Playwright. The guide is free, and so is the skill’s code.
What you need
Honestly: it’s not zero setup
Unlike other skills (PDFs, for example, need nothing installed), here you build a real layer of animated graphics: you need a render environment.
-
Claude Code
Required
In the terminal: it’s the agent that runs the whole edit.
-
Node.js + HyperFrames CLI
Required
The engine that renders the animated graphics.
-
FFmpeg
Required
Cuts, assembles the final video, burns in the subtitles.
-
Playwright
Required
Checks the graphics are right before the real render.
-
ElevenLabs
Optional, audio layer
Word-by-word transcription for the karaoke subtitles. It has a free plan; Whisper running locally also works, for free.
-
HeyGen
Optional, audio layer
Library music and sound effects. Skip it and the reel still works, just without the extra audio.
Don’t have these tools yet? Ask Claude Code to read the repo’s SKILL.md file: it checks what’s missing and installs it, asking you first before touching anything global.
git clone https://github.com/marcogalluccio/claude-reel
Prefer a file to keep? The PDF guide (in Italian): what you need, the block-by-block method with a preview at every step, the ready-made graphic archetypes, the mistakes that cost you a whole render.
Example
See a real reel
The reel below was edited with this skill: press play and listen to how subtitles, graphics and music follow the speech.
Want this for your team?
I teach teams to work with AI agents, in-company or in class in Rome. We start from tools like this one.