An experiment in AI & interface design
Same prompt.
Different skills.
What changes when you give an AI coding agent a UI skill? We gave the same model the same briefs and put the results side by side.
See the difference
ONE VARIABLE. EVERY POSSIBILITY.
—Skills compared
—Shared briefs
—Pages generated
—One model throughout
01 / The comparison
The proof is in the pixels.
Explore the outputs. Find the skills that make a difference.
Gathering the results…
Loading skills, briefs, and screenshots.
No matching skills.
Try a different skill name or author.
Click a preview to explore full results ↗Scroll to compare
02 / A fair starting point
Let the skill speak for itself.
Every page starts with the same product brief and the same model. The only variable is the UI skill added to the system prompt. The first row uses no skill at all: our control group.
We run each skill and brief pairing three times. Open a skill or preview to see its full results on skillarena.dev and judge the differences for yourself.
Same brief+Same model+Different skill
03 / Stay in the loop
The experiment continues.
Get updates as SkillArena explores what makes AI-generated interfaces better.