<aside> 🧬

The idea in one line. A design loop finds you one beautiful thing. Design DNA turns that one thing into a skill you keep forever, so you can make more of it, in any format, without the look drifting.

</aside>

design-dna-card-v3.png

<aside> 📦

Important: this is not only about "designs". It is about Claude codifying anything it built. The winner from your loop might be a website, a poster, a motion graphic, a carousel or a deck. Same system, same file, same tests. The format is an output, not the subject.

</aside>

The problem

You run a design loop. Lots of models make lots of options. One comes out beautiful.

Then it dies.

Nobody wrote down why it was beautiful. Next time you ask for "the same style" you get something close, then something a bit off, then something generic. By the fifth one it looks like every other AI design on the internet.

<aside> ⚠️

This is not the model being lazy. Every rule you do not write down, the model has to guess. And it does not guess randomly. It guesses the average of everything it has ever seen. That average is what people call AI slop. Slop is just the sound of an unwritten rule.

</aside>

So the answer is: write the rules down. But which rules, and how many? That turns out to be the whole game.

The 60-second version

  1. Find a design you love.
  2. Take it apart. Measure it, do not describe it.
  3. Argue with yourself about which bits actually matter.
  4. Write the rules down as a file.
  5. Write tests that a bad copy would fail.
  6. Rebuild the original from your rules alone. Whatever you get wrong is a rule you forgot. Add it. Go again.
  7. Save the whole thing as a skill.