Make a launch video with an AI agent

Agents are good at writing HTML and CSS. Video skills use that: the agent writes each scene as a web page with animations, and a renderer records the page frame by frame into an MP4. Nothing is generated by a video model, so text is always sharp and spelled right.

What you need

• An agent that reads SKILL.md, such as Claude Code or Codex
• Node.js 22 or newer
• FFmpeg on your PATH

The workflow a launch-film skill follows

1. Read your product URL or README and pick the three things worth showing.
2. Write a storyboard with scene lengths and on-screen text, and show it to you.
3. Capture your real interface, then compose each scene in HTML.
4. Render a low-resolution draft, check timing and text fit, then render the final file.

How to avoid the generic look

Most agent-made videos look alike because they use the same default fonts, gradients and easing. A skill fixes the art direction up front: a type pairing, a palette, a motion curve and rules about what never to do. That is most of what you pay for in a Lookbook video skill.