G.STANCUTA
Agent skill · open

/videoThe video skill

Turn a prompt into a finished, narrated, captioned, animated video.

I built a skill for AI coding agents (Claude Code, Codex) that takes a single prompt and produces a complete short video: it writes the script, generates a natural voiceover, syncs word-by-word captions to the audio, animates scenes that illustrate what is said, sizes everything for the target platform, and renders the final MP4 — all without leaving the agent.

Unzip into your agent's skills folder, add an OpenAI key, done.

Watch the demo

This video was made by the skill itself.

What it does

  • Writes a platform-sized script from your topic, hook to call-to-action.
  • Generates a natural voiceover (OpenAI gpt-4o-mini-tts; ElevenLabs optional).
  • Transcribes it for word-level timestamps so captions land exactly on the voice.
  • Animates SVG scenes that illustrate the script and cut on the narration's pauses.
  • Sizes for TikTok, Reels, Shorts or YouTube and renders the final MP4 with Remotion.
  • Writes the social copy — title, caption and hashtags — for the chosen platform.

The pipeline

  1. 01Interview — the agent asks for platform, script, voice, tone and look.
  2. 02Voice — gpt-4o-mini-tts generates the narration.
  3. 03Timestamps — whisper-1 returns word-level timing.
  4. 04Scaffold — a Remotion (React) project is created automatically.
  5. 05Scenes — the agent authors animations that illustrate this exact script.
  6. 06Render — Remotion renders an MP4 whose length matches the voice.
  7. 07Social copy — a ready title, caption and hashtags are written for you.

Install & run

The fastest way is the Claude Code plugin — two lines and /video is available everywhere. Prefer another agent, or want the source? Grab the .zip below or the GitHub repo.

Claude Code · plugin
/plugin marketplace add jumpino27/Video-Skill-Remotion
/plugin install video@video-skill-remotion
github.com/jumpino27/Video-Skill-Remotion

Requirements

  • Node.js 18–22 (newer Node breaks Remotion's renderer).
  • An OpenAI API key — used for the voice and the captions.
  • Works with Claude Code, Codex and similar agents.
  • Optional: ElevenLabs (richer voice) and Suno (theme music).

Steps

  1. 01Download the .zip and unzip it into your agent's skills folder (~/.claude/skills/video).
  2. 02Set OPENAI_API_KEY in your environment.
  3. 03In the agent, type /video and describe the video you want.
  4. 04Approve the script, and the agent renders the finished MP4 for you.
manual · zip (any agent)
# 1. download the .zip and unzip into your skills folder
mkdir -p ~/.claude/skills/video
unzip video-skill.zip -d ~/.claude/skills/video

# 2. set your key (voice + captions)
export OPENAI_API_KEY=sk-...

# 3. in your agent, just ask:
/video a 30s TikTok about my new app
Good to know

The whole pipeline runs inside your agent — no web app, no upload, nothing leaves your machine except the API calls for the voice and captions. The demo on this page was produced entirely by the skill.

Download the skill (.zip)
Portfolio · Drawing Stamp
Drawn by
G. STANCUTA
Discipline
AI & AUTOMATION
Location
MORTER · SÜDTIROL
Status
Available
Languages
IT · EN · RO · DE+
Stack
PLOI · HETZNER
Revision
REV 2026.A
2026

© 2026 Gabriel Stancuta · jumpinotech.com — Architected with AI, built to run itself.