Video Skill Remotion
jumpino27/Video-Skill-Remotion
Crie vídeos, músicas e imagens narradas end-to-end com um comando
Ver no GitHubSkill Python para Claude Code que gera vídeos narrativos completos, músicas originais e imagens animadas a partir de prompts simples. Integra síntese de voz (ElevenLabs), transcrição (Whisper), geração de imagens (DALL-E) e renderização de vídeos (Remotion) com legendas sincronizadas e otimização automática para diferentes plataformas.
Instalação
npx convez-ai jumpino27/Video-Skill-Remotion1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Gerar vídeos virais com roteiro, voiceover e legendas karaokê automáticas
- Criar conteúdo para YouTube Shorts, TikTok, Instagram Reels otimizado por plataforma
- Produzir músicas originais ou instrumentais a partir de descrições textuais
- Gerar e animar imagens com remoção de fundo e sprite sheets
Veredito

Ferramenta robusta para produção de mídia programática, com bom suporte a workflows compostos. Demanda configuração de múltiplas APIs externas (ElevenLabs, OpenAI, Remotion), mas entrega resultados sólidos para criadores e automatizadores. Recomendado para quem já trabalha com vídeo e busca acelerar produção.
README
A tradução para o português deste README ainda está na fila.
🎬 Media Skills for Claude Code
Turn one sentence into a finished video, song, or image — without ever leaving your editor.
📲 Follow @jumpino27 for AI, news, skills & repos — ❤️ Like these skills? Support me on Revolut.
/video · /music · /image · /crop
/videouses/musicand/imageunder the hood, and/imageuses/crop— but each skill works on its own too.
⚡ Install
🅰️ Option A — Let the AI install it for you (easiest, works in any agent)
Paste this prompt into Claude Code, Codex, or any coding agent:
Read the GitHub repo https://github.com/jumpino27/Video-Skill-Remotion and install its
four skills (video, music, image, crop) into my GLOBAL skills folder — that's
~/.claude/skills/ for Claude Code, or the equivalent skills folder for whatever agent
you are. Then install the Python requirements with `pip install -r requirements.txt`,
and tell me exactly which API keys I need to put in my .env (ElevenLabs + OpenAI).
That's it — the agent clones the repo, copies the four skill folders into place, installs the dependencies, and tells you what keys to add.
🅱️ Option B — Native plugin command (Claude Code)
/plugin marketplace add jumpino27/Video-Skill-Remotion
/plugin install media-skills@video-skill-remotion
Installs all four skills globally (available in every project), with managed updates via /plugin marketplace update. When installed as a plugin they're invoked namespaced — /media-skills:video, /media-skills:music, … — and still auto-trigger when you simply ask for "a video / a song / an image".
🅲 Option C — Manual
Copy the four folders into your skills directory and restart Claude Code:
skills/video → ~/.claude/skills/video
skills/music → ~/.claude/skills/music
skills/image → ~/.claude/skills/image
skills/crop → ~/.claude/skills/crop
Then install the Python deps: pip install -r requirements.txt
🔑 API keys
Copy the template and fill in your keys:
cp .env.example .env # then edit .env
| Key | Used by | Get it |
|---|---|---|
ELEVEN_LABS | /video voiceover · /music | elevenlabs.io → Profile → API Keys |
OPENAI_API_KEY | /video captions · /image | platform.openai.com |
The skills find your keys automatically, in this order (an env var of the same name always wins over a file):
process env → $SKILL_ENV_FILE / $VIDEO_ENV_FILE → skills/<name>/.env
→ ~/.claude/skills/.env → ~/.claude/.env → ~/Documents/.env
The simplest setup is one shared .env at ~/.claude/skills/.env.
🔒
.envand your personalpersona.jsonare git-ignored — they never get committed.
🎨 Make /video yours
Out of the box /video ships with neutral defaults. To personalize it — your name, brand, links, voice, presenter, accent color — drop in a persona file:
cp skills/video/config/persona.example.json skills/video/config/persona.json
{
"creatorName": "Ada Lovelace",
"brandName": "Analytical Engine",
"links": { "website": "https://ada.dev", "youtube": "", "instagram": "", "tiktok": "", "x": "" },
"voice": { "id": "eXO2dNYP4zo4oPWzuLta", "name": "Ivy", "model": "eleven_v3", "stability": "natural" },
"presenter": { "name": "Eve", "skill": "eve" },
"accentColor": "#6C5CE7",
"defaultPlatform": "shorts",
"distribution": { "linkPlacement": "description" }
}
- Voice id — grab yours from the ElevenLabs Voice Library (each voice has an id). Override per render with
--voice-idor theVIDEO_VOICE_IDenv var. - Links / accent / distribution — the script writer and social-copy step use these so spoken CTAs and captions point where you want viewers to go.
- Presenter — only matters if you want an on-screen avatar host (default points at a separate
/eveskill).
If persona.json is absent, the skill falls back to persona.example.json — so it always runs.
🕹️ Usage
Just talk to your agent:
- 🎥
/video— "a 30-second TikTok on why my note-taking app saves an hour a week." → interviews you, writes & shows the script for approval, then voices, captions, animates, and renders toout/video.mp4+ acontent.mdwith title/caption/hashtags. - 🎵
/music— "a dreamy 80s synthwave instrumental, 110 BPM, for a launch video." → asks the missing details, crafts a producer-grade prompt, renders tomusic_output/. - 🖼️
/image— "a glossy transparent fox logo, sprite sheet, 6 frames." → asks the quality tier (with prices), generates, auto-keys to transparent PNGs via/crop.
Each skill's full playbook lives in its SKILL.md.
🧩 Requirements
| Node.js 18–22 | /video only (Remotion is unstable on Node 23+). Not needed for the others. |
| Python 3.9+ | /music, /image, /crop |
| ffmpeg | on your PATH (audio trim/concat in /video) |
| Keys | ElevenLabs + OpenAI |
pip install -r requirements.txt
💸 Costs
These skills call paid APIs (ElevenLabs, OpenAI) billed to your accounts. /image always asks you to pick a quality tier with prices before spending; /video and /music bill per character / track length. Nothing is generated without your go-ahead.
📁 Layout
.
├── .claude-plugin/ marketplace.json + plugin.json (Option B)
├── .env.example API-key template → copy to .env
├── requirements.txt Python deps for /image, /crop
└── skills/
├── video/ SKILL.md · scripts/ · assets/remotion-template/ · config/persona.example.json
├── music/ SKILL.md · generate_music.py · generate_music_lyria.py
├── image/ SKILL.md · image.py
└── crop/ SKILL.md · crop.py
Enjoying these skills?
AI · news · skills & repos on Instagram · support keeps new skills coming
MIT © jumpino27 — see LICENSE.
You're responsible for complying with the underlying APIs (ElevenLabs, OpenAI) and for anything you generate and publish.