Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Master Frontend & Web Architecture.
.claude/skills/dokhacgiakhoa-modern-web-architect/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | 44% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 13% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 20% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 80% | 0% |
You are a Principal Frontend Architect and Design Engineer. You build web applications that are technically flawless, performant, and visually stunning.
Before coding, calculate the Frontend Feasibility & Complexity Index (FFCI):
FFCI = (Architectural Fit + Reusability + Performance) − (Complexity + Maintenance)
'use client' only for interactivity.Vite for development and Vitest for lightening-fast testing.useActionState, useOptimistic, and the use API.<Suspense>. No manual isLoading flags.useState for component-level.Zustand for complex global state.Context for subtree configuration.Every UI must have an Intentional Aesthetic (e.g., Editorial Brutalism, Luxury Minimal).
Evaluate via Design Feasibility & Impact Index (DFII): DFII = (Impact + Context Fit + Feasibility + Performance) − Consistency Risk
React.lazy) for heavy modules.<Image> for automatic optimization.npm list or bundle analyzers to avoid bloat. Prefer lightweight libraries (e.g., lucide-react over font-awesome).Merged and optimized from 11 legacy frontend, React, and Next.js skills.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 49,321 | 37,537 | -24% | 1 | 1 | 0% | 7,494 | 7,783 | +4% | 0 | 0 | — |
case-02 | fail→fail | 39,745 | 52,171 | +31% | 1 | 1 | 0% | 7,438 | 9,288 | +25% | 0 | 0 | — |
case-03 | fail→fail | 43,494 | 33,004 | -24% | 1 | 1 | 0% | 8,219 | 7,622 | -7% | 0 | 0 | — |
case-04 | pass→pass | 15,157 | 10,198 | -33% | 1 | 1 | 0% | 2,440 | 2,752 | +13% | 0 | 0 | — |
case-05 | pass→pass | 10,306 | 5,543 | -46% | 1 | 1 | 0% | 1,640 | 1,962 | +20% | 0 | 0 | — |
case-06 | pass→pass | 16,132 | 21,024 | +30% | 1 | 1 | 0% | 2,853 | 5,127 | +80% | 0 | 0 | — |
case-07 | fail→fail | 18,144 | 16,525 | -9% | 1 | 1 | 0% | 2,983 | 3,988 | +34% | 0 | 0 | — |
case-08 | pass→pass | 14,630 | 18,995 | +30% | 1 | 1 | 0% | 2,322 | 4,236 | +82% | 0 | 0 | — |
case-09 | pass→pass | 15,141 | 16,741 | +11% | 1 | 1 | 0% | 2,709 | 4,120 | +52% | 0 | 0 | — |
case-10 | pass→pass | 13,645 | 15,011 | +10% | 1 | 1 | 0% | 2,110 | 3,749 | +78% | 0 | 0 | — |
case-11 | pass→pass | 8,109 | 8,277 | +2% | 1 | 1 | 0% | 1,466 | 2,522 | +72% | 0 | 0 | — |
case-12 | pass→pass | 10,868 | 11,917 | +10% | 1 | 1 | 0% | 1,944 | 3,074 | +58% | 0 | 0 | — |
case-13 | pass→pass | 6,052 | 7,730 | +28% | 1 | 1 | 0% | 1,060 | 2,528 | +138% | 0 | 0 | — |
case-14 | pass→pass | 4,110 | 6,297 | +53% | 1 | 1 | 0% | 769 | 2,219 | +189% | 0 | 0 | — |
case-15 | pass→pass | 12,149 | 10,163 | -16% | 1 | 1 | 0% | 1,819 | 2,651 | +46% | 0 | 0 | — |
case-16 | pass→pass | 15,306 | 18,047 | +18% | 1 | 1 | 0% | 2,542 | 4,204 | +65% | 0 | 0 | — |
case-17 | fail→pass | 19,263 | 18,136 | -6% | 1 | 1 | 0% | 3,011 | 4,341 | +44% | 0 | 0 | — |
case-18 | fail→pass | 6,370 | 3,932 | -38% | 1 | 1 | 0% | 1,048 | 1,687 | +61% | 0 | 0 | — |
case-19 | pass→pass | 6,648 | 8,775 | +32% | 1 | 1 | 0% | 1,108 | 2,548 | +130% | 0 | 0 | — |
case-20 | pass→pass | 6,066 | 7,372 | +22% | 1 | 1 | 0% | 1,168 | 2,212 | +89% | 0 | 0 | — |
case-21 | pass→pass | 10,355 | 9,097 | -12% | 1 | 1 | 0% | 2,130 | 2,984 | +40% | 0 | 0 | — |
case-22 | pass→pass | 7,572 | 8,459 | +12% | 1 | 1 | 0% | 1,440 | 2,668 | +85% | 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. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.