Open-Source AI Agent Skills for Motion Canvas

We just open-sourced a set of AI agent skills that give coding agents full autonomous control over Motion Canvas and Canvas Commons projects. The skills are available on GitHub at VideoZero/skills and can be installed with a single command:

npx skills add videozero/skills

This works with Claude Code, Codex, OpenCode, and any other coding agent that supports skills.

Saving Time on High-Quality Videos

Motion Canvas is a powerful tool for creating production-quality animated videos through code. But the learning curve is real — mastering the component system, understanding signal-based reactivity, and getting animation timing right takes time and practice.

These skills compress that learning curve dramatically. Instead of reading through documentation and experimenting with parameters, the agent already knows the framework and can apply the 12 principles of animation out of the box. What might take an experienced developer hours of tweaking can be done in minutes.

For teams producing educational content, explainer videos, or technical visualizations, this means faster iteration cycles and more time spent on the creative direction rather than the implementation details.

What’s Included

The repository contains three skills:

• motion-canvas — Framework reference covering components, signals, refs, layout, transitions, filters, gradients, shaders, and 18 reference docs. This gives the agent deep knowledge of how to write Motion Canvas code correctly.
• motion-canvas-agent — HTTP API for programmatic editor control. This is what enables autonomy: the agent can seek through the timeline, take screenshots, inspect the scene graph, adjust settings, and trigger renders — all without human intervention.
• motion-canvas-agent also teaches agents how to capture screenshots at specific frames, enabling visual verification of what they’ve built before exporting.
• motion-graphics-fundamentals — The 12 principles of animation and timing guidance, so the agent doesn’t just produce working code but produces animations that actually look good.

Fully Autonomous Video Creation

The key breakthrough here is end-to-end autonomy. With these skills installed, a coding agent can:

1. Set up a new Motion Canvas project from scratch, or improve an existing scene
2. Write animation code using the framework reference as context
3. Seek through the timeline and take screenshots to visually verify the result
4. Adjust and iterate based on what it sees
5. Export the final video — no manual steps required

This means you can describe what you want in natural language, and the agent handles everything from project scaffolding to final render. It can also be used more surgically — point it at an existing scene and ask it to refine a specific transition or improve the timing.

Contributions Welcome

The skills are open-source and we welcome contributions. Whether it’s improving the framework reference, adding new agent capabilities, or fixing edge cases — PRs are open at github.com/VideoZero/skills.

Need Something More Convenient?

These skills are built for developers who enjoy a code-first workflow and want precise control over every animation detail. If that’s you, we think you’ll love working with them.

But not everyone wants to set up a project, install dependencies, and wrangle code — and that’s exactly why we built VideoZero. If you prefer a more convenient service or need to generate videos at scale, VideoZero lets you go from a text prompt to a finished animated video without any setup. And with our recent ElevenLabs integration, your videos can include automated AI narration too.