skill

reel-studio

🎬 Make Instagram reels in YOUR voice from one line: /reel <topic>. Free & local (Windows + Mac). Setup: paste 1 command from the README → record 20s of voice → done.

AwaisZulfqar/reel-studiocommunitylistedunreviewed

Use skill

Set up

Skills are installed by placing their folder where your agent loads skills.

Download ZIP from GitHub Open the repository

The download comes from github.com/AwaisZulfqar/reel-studio itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Nov 2, 2026

Reel Studio — interview-prep reels from one line

Type a topic in Claude Code, get a finished Instagram reel in your own voice:

/reel Day 2 of JavaScript interview prep: hoisting

Claude writes the script, your cloned voice reads it, and code, diagrams, captions, music and sound effects are animated in sync. Out comes a 1080×1920 MP4, ready to post. Everything runs on your computer and costs nothing.

Built for tech interview content (JavaScript, React, system design, DSA, backend, behavioral), and easy to bend to any explainer.

What you get

🎙️ Your voice: record 20 seconds once and every reel is spoken in it (Chatterbox voice cloning, local)
⌨️ Code that types itself, with syntax colors, highlighted lines, and console output (errors in red)
🧩 System-design diagrams whose boxes and arrows appear exactly when you say them
🆚 Comparisons, step lists, big takeaways, and a Follow button that flips to Following
💬 Word-by-word captions matched to the real voice
🎵 Music + sound effects: a fresh lo-fi beat per reel that dips under your voice, plus whooshes, key clicks, pops, ding/buzz
✅ Self-checking voice: every sentence is transcribed back, and any take that skips words is re-recorded automatically
🎨 Settings page: voice, handle, colors, music, effects and pronunciations, no config files needed

🚀 Easy setup guide (one time, about 15 minutes)

No coding needed. Just copy and paste.

Step 1: Install Claude Code

Get it from claude.com/claude-code and sign in. This is the app you'll type /reel into.

Step 2: Open the terminal

Windows	Mac
Press the Windows key, type PowerShell, press Enter	Press Cmd + Space, type Terminal, press Enter

Step 3: Paste one line and press Enter

Windows:

irm https://raw.githubusercontent.com/AwaisZulfqar/reel-studio/main/install.ps1 | iex

Mac:

curl -fsSL https://raw.githubusercontent.com/AwaisZulfqar/reel-studio/main/install.sh | bash

It installs everything it needs on its own and shows the progress, 1/5 → 5/5. The first time it downloads about 3.5 GB, so it takes 10–15 minutes. Wait until you see:

READY

💡 If Windows asks "Do you want to allow this app to make changes?", click Yes.

Step 4: Record your voice (20 seconds)

Open Claude Code and type:

/reel settings

A page opens in your browser. Click ● Record 20s and read the text on the screen out loud. Then type your Instagram handle and pick your colors and music. Everything saves on its own.

Other ways to open the settings page: double-click settings.cmd (Windows) or settings.command (Mac) in the .claude/skills/reel folder.

Step 5: Make your first reel 🎬

In Claude Code, type:

/reel what is a closure in JavaScript

Wait about 5 minutes. Claude tells you where the video is saved, and it also shows up at the bottom of the settings page.

That's it. From now on it's only Step 5.

Problem	Fix
`...isn't on PATH yet. Close this window...`	Close PowerShell, open a new one, paste the same line again. It continues where it stopped.
`running scripts is disabled on this system`	Paste this first, then the install line: `Set-ExecutionPolicy -Scope CurrentUser RemoteSigned`
`winget is not recognized` (Windows)	Update App Installer from the Microsoft Store, then try again.
Download stuck at the same % for minutes	Press Ctrl + C, paste the line again. It resumes.
`/reel` not found in Claude Code	Fully quit and reopen Claude Code.
Microphone doesn't record	Click the 🔒 next to the address bar → allow Microphone. Or upload a voice memo instead.
Voice says a word wrong	Settings page → Pronunciation → add how it sounds (e.g. `const` → `konst`).
Anything else	Run the install line again; it's safe and also updates you to the newest version. Still stuck? Open an issue and paste the error.

Tool	Why	Windows	macOS
Git	download the skill	`winget install Git.Git`	`xcode-select --install`
Node.js 18+	video engine (Remotion)	`winget install OpenJS.NodeJS.LTS`	`brew install node`
uv	sets up Python 3.12	`winget install astral-sh.uv`	auto-installed
Google Chrome	renders the frames	`winget install Google.Chrome`	download

Everything goes into ~/.claude/skills/reel (Windows: C:\Users\<you>\.claude\skills\reel). Manual install: git clone https://github.com/AwaisZulfqar/reel-studio ~/.claude/skills/reel, then run setup.ps1 (Windows: powershell -ExecutionPolicy Bypass -File setup.ps1) or bash setup.sh inside it. To uninstall, delete that folder.

Use it

/reel what is a closure in JavaScript
/reel design a URL shortener
/reel SQL vs NoSQL, make it a quick comparison
/reel Day 5: debounce vs throttle, start with a search box example

Each reel takes about 5 minutes on an M1 Mac (voice ~3 min, video ~2 min) and lands in jobs/<topic>/<topic>.mp4. The settings page shows all your reels.

Not happy with something? Just say it: "slower voice", "change the hook", "bigger code", "music quieter". Voice clips are cached, so visual tweaks re-render in ~2 minutes.

Settings

Everything lives in config.json (created from config.example.json). The settings page edits it for you.

Setting	What it does
`handle`	Shown at the top of every reel
`speed`, `exaggeration`, `cfg`	Voice speed, energy and pacing
`music`	`"auto"` (generated lo-fi), `"music/"` (your tracks, random pick) or `null`
`musicVolume`, `sfx`, `sfxVolume`	Mix levels; turn effects on/off
`theme`	Background, text, accent and highlight colors
`pronounce`	How the voice should say tech words (`const` → `konst`); captions keep the real spelling

How it works

topic → Claude writes script.json → Chatterbox speaks each sentence in your voice
      → Whisper checks every word was said + times each word
      → sound.py mixes voice + ducked music + effects (synced to animations)
      → Remotion renders the scenes in Chrome → MP4

File	Role
`SKILL.md`	Instructions Claude follows for `/reel` (script rules, scene types)
`make.py`	Pipeline: voice → timings → mix → render
`sound.py`	Music synth, sound effects, mixing
`video/src/`	Remotion scenes: hook, code, diagram, compare, list, big, cta
`ui.py`, `ui.html`	Settings page (Python standard library only)
`jobs/example/script.json`	A full example script

You can also run it without Claude: write a script.json like the example and run .venv/bin/python make.py jobs/<name>/script.json.

Platform notes

macOS (Apple Silicon): tested on M1, uses the GPU for the voice.
Windows 10/11: install.ps1 / setup.ps1. With an NVIDIA GPU, the CUDA build of PyTorch is installed automatically. Without one, the voice runs on the CPU and a reel takes noticeably longer (roughly 10–20 min).
Linux: install.sh; install node and Chrome/Chromium with your package manager first. CUDA is used if available.

Licenses and fair use

This project: MIT (see LICENSE).
Remotion is free for individuals and companies of up to 3 people; larger companies need a Remotion company license.
Chatterbox (MIT) adds an inaudible watermark to generated speech.
Fonts (Inter, JetBrains Mono, Instrument Serif) are under the SIL Open Font License.
Only clone your own voice, or a voice you have explicit permission to use.
Add only music you have the rights to. The built-in beat is generated locally and is yours to use.