Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Render the UI and prove it's balanced + usable: a deterministic layout audit (centroid / optical-center / pixel-oracle balance via explicit math + annotated screenshot) plus a vision-judged Nielsen usability audit by a separate fresh-eyes judge. The measurement layer taste-only design skills lack.
.claude/skills/sickn33-deterministic-design/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -44% | 0% |
Use to catch AI-generated UI that "looks off", is misaligned or centered-mush, or fails usability — when you need to PROVE a layout is balanced and usable instead of trusting the model's eye. Compose it with any taste/token design skill before reporting design "done".
_Source: connerkward/deterministic-design-skill (MIT)._
Thesis: determinism beats AI randomness. A model can't trust its own eye on layout — so don't. Render the UI and measure it.
Two sub-skills (load as needed):
+ 8-pt spacing, and layout-audit.js computes centroid / optical-center / pixel-oracle balance and draws an annotated screenshot. Numbers, not vibes. Plus a render-then- critique vision loop.
Nielsen's 10 + interaction heuristics via a SEPARATE fresh-eyes judge → prioritized fix list.
This improves existing design skills (including the default Anthropic one) by adding the layer they lack — it doesn't just advise on taste, it renders, measures, and judges the output. Composable with any design skill.
In central this lives as a subdir of ckw-design; it publishes separately as deterministic-design-skill (its own distribution) via publish-skill. One of the two flagship narratives — the determinism one; its sibling is human-in-the-loop (lookdev).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→fail | 15,226 | 17,679 | +16% | 1 | 1 | 0% | 2,455 | 4,102 | +67% | 0 | 0 | — |
case-02 | pass→pass | 14,140 | 8,175 | -42% | 1 | 1 | 0% | 2,728 | 2,076 | -24% | 0 | 0 | — |
case-03 | fail→pass | 23,944 | 13,546 | -43% | 1 | 1 | 0% | 5,331 | 3,748 | -30% | 0 | 0 | — |
case-04 | fail→pass | 11,501 | 1,699 | -85% | 1 | 1 | 0% | 2,016 | 852 | -58% | 0 | 0 | — |
case-05 | fail→pass | 15,648 | 7,765 | -50% | 1 | 1 | 0% | 2,397 | 1,689 | -30% | 0 | 0 | — |
case-06 | pass→pass | 14,568 | 5,496 | -62% | 1 | 1 | 0% | 2,401 | 1,561 | -35% | 0 | 0 | — |
case-07 | pass→pass | 11,986 | 13,214 | +10% | 1 | 1 | 0% | 1,856 | 1,259 | -32% | 0 | 0 | — |
case-08 | fail→pass | 12,605 | 6,721 | -47% | 1 | 1 | 0% | 1,980 | 1,544 | -22% | 0 | 0 | — |
case-09 | fail→pass | 13,750 | 4,504 | -67% | 1 | 1 | 0% | 2,281 | 1,266 | -44% | 0 | 0 | — |
case-10 | pass→pass | 16,132 | 6,504 | -60% | 1 | 1 | 0% | 2,433 | 1,541 | -37% | 0 | 0 | — |
case-11 | pass→pass | 16,207 | 7,951 | -51% | 1 | 1 | 0% | 2,817 | 1,620 | -42% | 0 | 0 | — |
case-16 | pass→fail | 13,213 | 1,696 | -87% | 1 | 1 | 0% | 2,054 | 790 | -62% | 0 | 0 | — |
case-12 | pass→pass | 14,993 | 7,362 | -51% | 1 | 1 | 0% | 2,196 | 1,613 | -27% | 0 | 0 | — |
case-13 | pass→pass | 4,443 | 2,136 | -52% | 1 | 1 | 0% | 757 | 894 | +18% | 0 | 0 | — |
case-14 | pass→pass | 4,272 | 2,398 | -44% | 1 | 1 | 0% | 707 | 938 | +33% | 0 | 0 | — |
case-15 | pass→pass | 6,637 | 1,839 | -72% | 1 | 1 | 0% | 1,076 | 784 | -27% | 0 | 0 | — |
case-21 | pass→pass | 12,359 | 7,529 | -39% | 1 | 1 | 0% | 1,953 | 1,715 | -12% | 0 | 0 | — |
case-17 | pass→pass | 12,803 | 6,173 | -52% | 1 | 1 | 0% | 2,095 | 1,482 | -29% | 0 | 0 | — |
case-18 | pass→pass | 13,891 | 5,871 | -58% | 1 | 1 | 0% | 2,041 | 1,456 | -29% | 0 | 0 | — |
case-19 | fail→pass | 11,857 | 7,573 | -36% | 1 | 1 | 0% | 1,884 | 1,900 | +1% | 0 | 0 | — |
case-20 | pass→pass | 11,873 | 8,348 | -30% | 1 | 1 | 0% | 1,873 | 1,908 | +2% | 0 | 0 | — |
case-22 | pass→pass | 9,626 | 6,989 | -27% | 1 | 1 | 0% | 1,734 | 1,750 | +1% | 0 | 0 | — |
case-23 | pass→pass | 9,224 | 8,157 | -12% | 1 | 1 | 0% | 1,654 | 1,844 | +11% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.