The fix for an agent that keeps getting your design system wrong isn't a better prompt.

It's putting the rules where the agent reads them.

In the repo. In skills. In checks it can run itself.

Writing a course about exactly this.

What sells it for me is the refusal telling the agent exactly where the token should go. A bare "no hex" error usually just sends it looping. Does guard.mjs run only as an agent hook, or in pre-commit too so it catches humans as well?

I was working on the visual identity for the Ivy and wanted to experiment with stippling. Thousands of tiny dots forming an image. Simple enough as an idea. Figma had other plans. I tried plugins, but the number of SVG elements was too much. My laptop slowed down, and eventually the workflow became unusable. I remember thinking, "Why am I trying so hard to make this work inside Figma?" I didn't need another plugin. I needed a different tool. So I built one. It's called Stippling Studio. Nothing revolutionary. Just a tool that lets me experiment with the visual treatment I had in mind. But the experience stuck with me. For years, I learned to work around the limitations of my tools. Now I can sometimes build my way around them. https://stippling.pash.website/

So cool!!