Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generates a frontend interface that is visually pleasing and experientially sound by jointly optimizing aesthetics, motion graphics, and user experience (UX) using quantified evaluation. Use only when the user explicitly invokes this skill by name to redesign a frontend. Trigger keywords: use human-optimized-frontend, redesign frontend, redesign interface.
.claude/skills/sundial-org-human-optimized-frontend/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -41% | 0% |
Activate this skill only when the user explicitly instructs the agent to redesign a frontend and references this skill by name.
Do not activate for:
Score each dimension from 0–10:
Typography
Color
Layout
Background
Motion
UX
Cross-Dimensional Harmony
Weighted Total Score
Produce a single declarative frontend specification including:
No alternatives. No explanations. No theory.
If activation conditions are not met, output a minimal statement indicating the skill cannot be activated.
If after maximum iterations UX or harmony thresholds are not met, output a minimal statement indicating that a satisfactory frontend cannot be generated under the given constraints and terminate.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 32,049 | 14,750 | -54% | 1 | 1 | 0% | 6,204 | 4,031 | -35% | 0 | 0 | — |
case-02 | fail→pass | 28,481 | 18,914 | -34% | 1 | 1 | 0% | 4,874 | 4,741 | -3% | 0 | 0 | — |
case-03 | fail→pass | 35,450 | 14,298 | -60% | 1 | 1 | 0% | 6,197 | 3,720 | -40% | 0 | 0 | — |
case-09 | fail→pass | 31,197 | 14,784 | -53% | 1 | 1 | 0% | 6,173 | 3,874 | -37% | 0 | 0 | — |
case-04 | pass→fail | 18,748 | 2,241 | -88% | 1 | 1 | 0% | 4,052 | 1,713 | -58% | 0 | 0 | — |
case-05 | pass→fail | 15,393 | 1,786 | -88% | 1 | 1 | 0% | 2,234 | 1,607 | -28% | 0 | 0 | — |
case-06 | pass→fail | 18,167 | 1,595 | -91% | 1 | 1 | 0% | 2,739 | 1,588 | -42% | 0 | 0 | — |
case-07 | fail→pass | 27,394 | 13,159 | -52% | 1 | 1 | 0% | 6,169 | 3,627 | -41% | 0 | 0 | — |
case-08 | fail→pass | 17,061 | 11,382 | -33% | 1 | 1 | 0% | 2,331 | 3,325 | +43% | 0 | 0 | — |
case-10 | fail→pass | 33,179 | 18,799 | -43% | 1 | 1 | 0% | 6,176 | 4,367 | -29% | 0 | 0 | — |
case-11 | fail→pass | 31,385 | 13,518 | -57% | 1 | 1 | 0% | 6,172 | 3,555 | -42% | 0 | 0 | — |
case-12 | fail→pass | 33,273 | 14,262 | -57% | 1 | 1 | 0% | 6,175 | 3,765 | -39% | 0 | 0 | — |
case-13 | fail→pass | 30,672 | 12,279 | -60% | 1 | 1 | 0% | 5,103 | 3,855 | -24% | 0 | 0 | — |
case-14 | fail→fail | 30,152 | 20,162 | -33% | 1 | 1 | 0% | 6,170 | 4,685 | -24% | 0 | 0 | — |
case-15 | fail→fail | 32,778 | 10,094 | -69% | 1 | 1 | 0% | 6,172 | 3,028 | -51% | 0 | 0 | — |
case-16 | fail→pass | 31,155 | 16,359 | -47% | 1 | 1 | 0% | 6,175 | 3,893 | -37% | 0 | 0 | — |
case-17 | fail→fail | 31,531 | 15,451 | -51% | 1 | 1 | 0% | 6,170 | 3,968 | -36% | 0 | 0 | — |
case-18 | pass→pass | 29,185 | 13,433 | -54% | 1 | 1 | 0% | 6,167 | 3,794 | -38% | 0 | 0 | — |
case-19 | fail→pass | 30,350 | 16,835 | -45% | 1 | 1 | 0% | 6,170 | 4,247 | -31% | 0 | 0 | — |
case-20 | fail→fail | 32,120 | 15,746 | -51% | 1 | 1 | 0% | 6,168 | 3,866 | -37% | 0 | 0 | — |
case-21 | fail→pass | 26,198 | 18,584 | -29% | 1 | 1 | 0% | 4,695 | 4,687 | -0% | 0 | 0 | — |
case-22 | fail→pass | 30,555 | 19,539 | -36% | 1 | 1 | 0% | 6,181 | 4,895 | -21% | 0 | 0 | — |
case-23 | fail→pass | 25,989 | 14,893 | -43% | 1 | 1 | 0% | 5,018 | 3,988 | -21% | 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 +52 percentage points is the difference between those two pass rates over the 23 comparable cases. 3 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.