Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apple-tier soft web prototype. Silver/cream canvas, double-bezel cards, button-in-button CTAs, generous squircle radii, spring motion, ambient mesh. Distilled from Leonxlnx/taste-skill `soft-skill` + sections 4–8 of `taste-skill`.
.claude/skills/nexu-io-web-prototype-taste-soft/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 79% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 182% | 0% |
For briefs that ask for "Apple-like", "Linear-tier", "premium consumer", "calm SaaS", or "$150k agency" finish. The aesthetic is soft, weighty, and obsessive about nested architecture.
Distilled from Leonxlnx/taste-skill — skills/soft-skill/SKILL.md ("Vanguard UI Architect") with the haptic micro-aesthetics, double-bezel architecture, and motion choreography rules. See example.html in this directory for a fully-realized template.
#F2F2F0 or #FDFBF7). Not pure white.clamp(48px, 7vw, 96px), letter-spacing: -0.035em, line-height: 0.96. Heavy weight (700+).border-radius: 28px–40px (rounded-[2rem] to rounded-[2.5rem]) on major surfaces.p-1.5, hairline border, soft outer shadow) containing an inner core with its own background and a concentric smaller radius (e.g., outer 2rem, inner calc(2rem - 0.375rem)).rounded-full px-6 py-3); the trailing arrow lives inside its own circular wrapper flush-right.opacity ≤ 0.18, pointer-events: none, fixed.letter-spacing: 0.2em, text-[10px].border-black/5 / ring-1 ring-black/5).shadow-md, shadow-lg, hard drop shadows. Only diffuse, wide-spread, low-opacity (0 20px 40px -15px rgba(0,0,0,0.05)).mt-6 mx-auto w-max rounded-full).ease-in-out transitions. Use cubic-bezier(0.32, 0.72, 0, 1) or spring physics.h-screen (use min-h-[100dvh]).width / height / top / left. Only transform and opacity.#000. Use Zinc-950 / Off-Black.max-w-[52ch], two CTAs (primary pill with button-in-button arrow + ghost).-100% / 20s), pause on hover.transition: transform 700ms cubic-bezier(0.32,0.72,0,1) minimum.0.98 on :active, inner trailing-icon circle translates (+1px, -1px) and scale(1.05).translateY(16px) blur(8px) opacity(0) → translateY(0) blur(0) opacity(1). IntersectionObserver only.transform: translateX(-50%) over 30s linear infinite, applied to a duplicated track.backdrop-blur + hairline ringpy-24min-h-[100dvh], never 100vhlinear / ease-in-out| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | fail→pass | 29,642 | 26,054 | -12% | 1 | 1 | 0% | 4,600 | 6,254 | +36% | 0 | 0 | — |
case-07 | pass→pass | 28,784 | 25,163 | -13% | 1 | 1 | 0% | 6,340 | 6,384 | +1% | 0 | 0 | — |
case-22 | fail→pass | 20,119 | 28,879 | +44% | 1 | 1 | 0% | 3,895 | 6,954 | +79% | 0 | 0 | — |
case-08 | pass→pass | 17,647 | 34,296 | +94% | 1 | 1 | 0% | 3,885 | 6,726 | +73% | 0 | 0 | — |
case-01 | fail→pass | 110,296 | 61,893 | -44% | 1 | 1 | 0% | 8,301 | 9,523 | +15% | 0 | 0 | — |
case-02 | fail→fail | 53,907 | 41,440 | -23% | 1 | 1 | 0% | 8,272 | 9,494 | +15% | 0 | 0 | — |
case-03 | fail→pass | 53,621 | 54,093 | +1% | 1 | 1 | 0% | 8,286 | 9,508 | +15% | 0 | 0 | — |
case-04 | pass→pass | 25,077 | 36,394 | +45% | 1 | 1 | 0% | 3,829 | 7,020 | +83% | 0 | 0 | — |
case-05 | fail→fail | 53,303 | 51,892 | -3% | 1 | 1 | 0% | 8,245 | 9,467 | +15% | 0 | 0 | — |
case-06 | pass→pass | 49,823 | 37,191 | -25% | 1 | 1 | 0% | 8,235 | 9,457 | +15% | 0 | 0 | — |
case-09 | fail→pass | 8,863 | 12,619 | +42% | 1 | 1 | 0% | 1,277 | 3,596 | +182% | 0 | 0 | — |
case-10 | fail→pass | 48,491 | 58,096 | +20% | 1 | 1 | 0% | 6,661 | 9,458 | +42% | 0 | 0 | — |
case-11 | fail→fail | 17,652 | 32,549 | +84% | 1 | 1 | 0% | 2,699 | 6,062 | +125% | 0 | 0 | — |
case-12 | fail→pass | 44,485 | 40,218 | -10% | 1 | 1 | 0% | 7,485 | 7,717 | +3% | 0 | 0 | — |
case-13 | pass→pass | 21,520 | 37,745 | +75% | 1 | 1 | 0% | 4,405 | 7,184 | +63% | 0 | 0 | — |
case-14 | pass→pass | 13,701 | 13,279 | -3% | 1 | 1 | 0% | 2,685 | 3,744 | +39% | 0 | 0 | — |
case-16 | fail→pass | 22,760 | 22,252 | -2% | 1 | 1 | 0% | 3,041 | 4,666 | +53% | 0 | 0 | — |
case-17 | fail→pass | 22,239 | 37,864 | +70% | 1 | 1 | 0% | 4,458 | 9,444 | +112% | 0 | 0 | — |
case-18 | pass→pass | 20,744 | 26,713 | +29% | 1 | 1 | 0% | 3,489 | 5,486 | +57% | 0 | 0 | — |
case-19 | fail→fail | 17,389 | 16,939 | -3% | 1 | 1 | 0% | 3,799 | 4,510 | +19% | 0 | 0 | — |
case-20 | fail→fail | 29,223 | 41,399 | +42% | 1 | 1 | 0% | 3,545 | 8,065 | +128% | 0 | 0 | — |
case-21 | pass→pass | 18,600 | 39,967 | +115% | 1 | 1 | 0% | 2,830 | 9,450 | +234% | 0 | 0 | — |
case-23 | fail→fail | 65,824 | 51,008 | -23% | 1 | 1 | 0% | 7,867 | 9,458 | +20% | 0 | 0 | — |
case-24 | fail→fail | 24,982 | 36,173 | +45% | 1 | 1 | 0% | 5,174 | 8,778 | +70% | 0 | 0 | — |
case-25 | pass→pass | 47,405 | 38,677 | -18% | 1 | 1 | 0% | 8,237 | 9,355 | +14% | 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. 25 cases were attempted. The headline lift of +36 percentage points is the difference between those two pass rates over the 25 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.