Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Standardized guidelines for organizing React applications using feature-based architecture. Use when structuring new projects, organizing large codebases, or implementing domain-driven folder structures.
.claude/skills/valec3-frontend-architecture-feature-based/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-03 | ✓→✗ | ▼ Worse | 16% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 22% | 0% |
| case-05 | ✓→✓ | = Same ✓ | 21% | 0% |
| case-06 | ✓→✓ | = Same ✓ | -18% | 0% |
src/
├── features/
│ ├── auth/
│ │ ├── components/
│ │ │ ├── LoginForm.tsx
│ │ │ └── RegisterForm.tsx
│ │ ├── hooks/
│ │ │ └── useAuth.ts
│ │ ├── api/
│ │ │ └── authApi.ts
│ │ ├── types/
│ │ │ └── auth.types.ts
│ │ └── index.ts # Public API
│ ├── products/
│ │ ├── components/
│ │ ├── hooks/
│ │ └── index.ts
│ └── cart/
├── shared/
│ ├── components/ # Shared UI
│ ├── hooks/
│ └── utils/
└── app/
├── App.tsx
└── routes.tsx1. Feature Colocation Keep all code related to a feature together:
typescript// features/products/index.ts (Public API) export { ProductList } from './components/ProductList'; export { useProducts } from './hooks/useProducts'; export type { Product } from './types/product.types';
2. Import Rules
shared/3. Shared Code Only extract to shared/ when:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 17,095 | 16,350 | -4% | 1 | 1 | 0% | 3,144 | 3,001 | -5% | 0 | 0 | — |
case-03 | pass→fail | 22,276 | 22,652 | +2% | 1 | 1 | 0% | 3,597 | 4,183 | +16% | 0 | 0 | — |
case-04 | pass→pass | 10,884 | 10,063 | -8% | 1 | 1 | 0% | 1,932 | 2,354 | +22% | 0 | 0 | — |
case-05 | pass→pass | 16,637 | 16,575 | -0% | 1 | 1 | 0% | 2,717 | 3,281 | +21% | 0 | 0 | — |
case-06 | pass→pass | 13,327 | 8,978 | -33% | 1 | 1 | 0% | 2,550 | 2,086 | -18% | 0 | 0 | — |
case-01 | fail→fail | 14,468 | 13,330 | -8% | 1 | 1 | 0% | 2,603 | 2,979 | +14% | 0 | 0 | — |
case-07 | fail→pass | 10,057 | 8,388 | -17% | 1 | 1 | 0% | 1,905 | 1,962 | +3% | 0 | 0 | — |
case-08 | fail→fail | 11,081 | 8,977 | -19% | 1 | 1 | 0% | 1,977 | 2,031 | +3% | 0 | 0 | — |
case-09 | pass→pass | 12,001 | 7,054 | -41% | 1 | 1 | 0% | 2,077 | 1,613 | -22% | 0 | 0 | — |
case-10 | pass→pass | 8,168 | 3,525 | -57% | 1 | 1 | 0% | 1,399 | 993 | -29% | 0 | 0 | — |
case-11 | pass→pass | 12,225 | 7,953 | -35% | 1 | 1 | 0% | 2,050 | 1,868 | -9% | 0 | 0 | — |
case-12 | pass→pass | 10,442 | 6,879 | -34% | 1 | 1 | 0% | 2,005 | 1,754 | -13% | 0 | 0 | — |
case-13 | pass→pass | 14,424 | 7,910 | -45% | 1 | 1 | 0% | 2,331 | 1,986 | -15% | 0 | 0 | — |
case-14 | pass→pass | 13,359 | 4,089 | -69% | 1 | 1 | 0% | 2,477 | 1,166 | -53% | 0 | 0 | — |
case-15 | pass→pass | 13,147 | 9,590 | -27% | 1 | 1 | 0% | 2,274 | 2,218 | -2% | 0 | 0 | — |
case-16 | pass→pass | 3,219 | 2,566 | -20% | 1 | 1 | 0% | 579 | 959 | +66% | 0 | 0 | — |
case-17 | pass→pass | 8,656 | 3,759 | -57% | 1 | 1 | 0% | 1,492 | 1,149 | -23% | 0 | 0 | — |
case-18 | pass→pass | 14,847 | 10,436 | -30% | 1 | 1 | 0% | 2,554 | 2,335 | -9% | 0 | 0 | — |
case-19 | pass→pass | 9,063 | 7,076 | -22% | 1 | 1 | 0% | 1,783 | 1,811 | +2% | 0 | 0 | — |
case-20 | pass→pass | 12,060 | 6,413 | -47% | 1 | 1 | 0% | 2,058 | 1,554 | -24% | 0 | 0 | — |
case-21 | pass→pass | 12,653 | 5,047 | -60% | 1 | 1 | 0% | 2,145 | 1,235 | -42% | 0 | 0 | — |
case-22 | pass→pass | 9,397 | 6,592 | -30% | 1 | 1 | 0% | 1,707 | 1,723 | +1% | 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 0 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.