Prompting motion graphics with Remotion and Claude Code

If you needed a slick motion graphics clip for your product launch yesterday, you basically had two options: suffer through After Effects keyframes or wait two weeks for a motion designer.

Now, you can just vibe-code it. On January 19, developers demonstrated using Remotion's official Agent Skills alongside Claude Code to script, build, and render a complex 3D motion graphics video entirely through natural language.

No Adobe tools, no manual timeline scrubbing. Claude Code handled everything from writing React components to camera math and final rendering.

An 8-second promo built strictly in the terminal

The demo project started with a simple prompt: create an animated promo showing a specific CLI install command.

The initial canvas was a clean 1280x1000 macOS terminal window in light mode. Claude Code explored the existing Remotion workspace, generated a dedicated terminal component, and fine-tuned fonts and backgrounds based on conversational feedback.

From there, the agent created a master composition. It placed the terminal inside, then applied smooth 3D transforms including bottom slide-ins, continuous Y-axis rotation, and scaling.

It even ran the CLI command itself, turned the resulting ASCII art into typewriter text effects, combined brand logos, and flipped the window 90 degrees toward the camera for a clean exit. The final 8-second asset was rendered straight to disk via the agent.

What Remotion Agent Skills actually are

Remotion lets developers build videos programmatically using standard React and web technologies. The newly packaged Agent Skills act as an official playbook tailored for AI coding tools like Claude Code, Cursor, Codex, or Kimi Code.

Developers can install the skills with a single command in their terminal. Once added, the agent instantly knows Remotion's recommended patterns instead of guessing.

The skills cover everything from base typography, audio, and easing curves to specialized tasks like CesiumJS 3D flyovers, subtitle syncing, SaaS architecture setups, and firing up the local preview studio.

Why it matters

We think this demo highlights an important alternative path for AI video creation.

While generative diffusion models fight with blurry text, hallucinated UI, and uncontrollable camera drifts, the Remotion approach relies on deterministic code. For indie hackers and software teams, generating UI promos, changelog animations, or end-of-year data recaps via code offers pixel-level precision and git-trackable iterations.

Instead of praying a video model gets your dashboard layout right, you let an agent write clean React components and hand off rendering to a proven browser-based engine.

The trade-off is clear: you won't be generating cinematic movie scenes this way, as it strictly fits UI demos, infographics, and motion typography. But if you already live in a terminal or code editor, bringing motion design into your vibe-coding pipeline is a massive productivity boost worth testing immediately.