React Production Engineering
SkillCloud & infraGives your agent guidance for building production React apps, covering architecture, state, testing, and deployment.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the React Production Engineering skill
About this capability
Complete methodology for building production-grade React applications with architecture decisions, component design, state management, performance optimization, testing, and deployment.
What this skill tells your AI
The instructions your AI receives, as published by leoyeai/openclaw-master-skills in skills/afrexai-react-production/SKILL.md and read by ahel’s review.
Complete methodology for building production-grade React applications. Covers architecture decisions, component design, state management, performance optimization, testing, and deployment — not just API reference, but engineering methodology with decision frameworks, templates, and scoring systems.
Phase 1: Architecture Assessment
Quick Health Check (score /16)
- Component tree depth < 6 levels (+2)
- No prop drilling past 2 levels (+2)
- Bundle size < 200KB gzipped (+2)
- LCP < 2.5s on 4G (+2)
- Test coverage > 70% on business logic (+2)
- Zero
anytypes in production code (+2) - No direct DOM manipulation (+2)
- Consistent error boundaries (+2)
Architecture Brief
project:
name: ""
type: "" # spa | ssr | hybrid | static
framework: "" # next | remix | vite-spa | astro
scale: "" # small (<20 routes) | medium (20-100) | large (100+)
team_size: "" # solo | small (2-5) | medium (6-15) | large (15+)
current_state:
react_version: "" # 18 | 19
typescript: true
router: "" # react-router | next-app | tanstack-router
state_management: "" # useState | zustand | jotai | redux | tanstack-query
styling: "" # tailwind | css-modules | styled-components | vanilla-extract
testing: "" # vitest | jest | playwright | cypress
ci_cd: "" # github-actions | gitlab-ci | vercel
pain_points: []
goals: []
Framework Selection Decision Matrix
| Factor | Vite SPA | Next.js | Remix | Astro |
|---|---|---|---|---|
| SEO needed | ❌ | ✅ Best | ✅ Good | ✅ Best |
| Dashboard/app | ✅ Best | ✅ Good | ✅ Good | ❌ |
| Content-heavy | ❌ | ✅ Good | ✅ Good | ✅ Best |
| Team familiarity | ✅ Simple | ⚠️ Learning curve | ⚠️ Web standards | ⚠️ Islands |
| Deployment | Anywhere | Vercel optimal | Anywhere | Anywhere |
| Bundle size | You control | Framework overhead | Smaller | Minimal JS |
Decision rules:
- Dashboard/internal tool with no SEO → Vite SPA
- Marketing + app hybrid → Next.js
- Content-first with some interactivity → Astro
- Web-standards-first, nested layouts → Remix
- Default for most SaaS products → Next.js
Phase 2: Project Structure & Conventions
Recommended Feature-Based Structure
src/
├── app/ # Routes/pages (framework-specific)
├── features/ # Feature modules (THE core pattern)
│ ├── auth/
│ │ ├── components/ # Feature-specific components
│ │ ├── hooks/ # Feature-specific hooks
│ │ ├── api/ # API calls & types
│ │ ├── utils/ # Feature utilities
│ │ ├── types.ts # Feature types
│ │ └── index.ts # Public API (barrel export)
│ ├── dashboard/
│ └── settings/
├── shared/ # Cross-feature shared code
│ ├── components/ # Generic UI components
│ │ ├── ui/ # Primitives (Button, Input, Card)
│ │ └── layout/ # Layout components
│ ├── hooks/ # Generic hooks
│ ├── lib/ # Utilities, constants
│ └── types/ # Global types
├── providers/ # Context providers
└── styles/ # Global styles
7 Structure Rules
- Feature isolation — features/ never import from other features directly; use shared/ or events
- Barrel exports — every feature has index.ts that defines its public API
- Colocation — tests, stories, and styles live next to their component
- Max file size — 300 lines. If bigger, split
- Max component size — 50 lines of JSX. If bigger, extract
- No circular deps — enforce with eslint-plugin-import
- Types colocated — feature types in feature, shared types in shared/types
Naming Conventions
Components: PascalCase.tsx (UserProfile.tsx)
Hooks: useCamelCase.ts (useAuth.ts)
Utilities: camelCase.ts (formatCurrency.ts)
Types: PascalCase.ts (User.ts) or types.ts
Constants: SCREAMING_SNAKE.ts (API_ENDPOINTS.ts)
Test files: *.test.tsx (UserProfile.test.tsx)
Story files: *.stories.tsx (Button.stories.tsx)
Phase 3: Component Design Patterns
Component Anatomy Template
// 1. Imports (grouped: react → third-party → internal → types → styles)
import { useState, useCallback, memo } from 'react'
import { clsx } from 'clsx'
import { Button } from '@/shared/components/ui'
import type { User } from '../types'
// 2. Types (exported for reuse)
export interface UserCardProps {
user: User
onEdit?: (id: string) => void
variant?: 'compact' | 'full'
className?: string
}
// 3. Component (named export, not default)
export const UserCard = memo(function UserCard({
user,
onEdit,
variant = 'full',
className,
}: UserCardProps) {
// 4. Hooks first
const [isExpanded, setIsExpanded] = useState(false)
// 5. Derived state (no useEffect for derived!)
const displayName = `${user.firstName} ${user.lastName}`
// 6. Handlers (useCallback for passed-down refs)
const handleEdit = useCallback(() => {
onEdit?.(user.id)
}, [onEdit, user.id])
// 7. Early returns for edge cases
if (!user) return null
// 8. JSX (max 50 lines)
return (
<div className={clsx('rounded-lg border p-4', className)}>
<h3>{displayName}</h3>
{variant === 'full' && <p>{user.bio}</p>}
{onEdit && <Button onClick={handleEdit}>Edit</Button>}
</div>
)
})
Component Composition Patterns
1. Compound Components (for related UI groups)
// Usage: <Tabs><Tabs.List><Tabs.Tab>A</Tabs.Tab></Tabs.List><Tabs.Panel>...</Tabs.Panel></Tabs>
const TabsContext = createContext<TabsContextType | null>(null)
export function Tabs({ children, defaultValue }: TabsProps) {
const [activeTab, setActiveTab] = useState(defaultValue)
return (
<TabsContext.Provider value={{ activeTab, setActiveTab }}>
{children}
</TabsContext.Provider>
)
}
Tabs.List = TabsList
Tabs.Tab = TabsTab
Tabs.Panel = TabsPanel
2. Render Props (for flexible rendering logic)
export function DataList<T>({ items, renderItem, renderEmpty }: DataListProps<T>) {
if (items.length === 0) return renderEmpty?.() ?? <EmptyState />
return <ul>{items.map((item, i) => <li key={i}>{renderItem(item)}</li>)}</ul>
}
3. Higher-Order Components (for cross-cutting concerns — use sparingly)
export function withAuth<P>(Component: ComponentType<P>) {
return function AuthenticatedComponent(props: P) {
const { user, isLoading } = useAuth()
if (isLoading) return <Spinner />
if (!user) return <Navigate to="/login" />
return <Component {...props} />
}
}
10 Component Rules
- One component per file — always
- Named exports — never default exports (refactoring safety)
- Props interface — always explicit, always exported
- No business logic in components — extract to hooks
- No inline styles — use Tailwind classes or CSS modules
- No string refs — useRef only
- No index as key — use stable identifiers
- Memo strategically — not everywhere, only for expensive renders
- Children over props — prefer composition over configuration
- Accessible by default — semantic HTML, ARIA when needed
Phase 4: State Management Decision Framework
State Type Decision Tree
Is it server data (from API)?
├─ YES → TanStack Query (or SWR) — NEVER Redux/Zustand for server state
│
└─ NO → Is it shared across features?
├─ YES → Is it complex with many actions?
│ ├─ YES → Zustand (or Redux Toolkit if team knows it)
│ └─ NO → Jotai (atomic) or Zustand (simple store)
│
└─ NO → Is it shared within a feature?
├─ YES → Context + useReducer (or Zustand feature store)
└─ NO → useState / useReducer (component-local)
State Management Comparison
| Tool | Best For | Bundle | Learning | Team Size |
|---|---|---|---|---|
| useState | Component-local | 0 KB | None | Any |
| useReducer | Complex local state | 0 KB | Low | Any |
| Context | Feature-scoped, low-frequency | 0 KB | Low | Any |
| Zustand | Global client state | 1.1 KB | Low | Any |
| Jotai | Atomic derived state | 3.4 KB | Medium | Small-Med |
| TanStack Query | Server state | 12 KB | Medium | Any |
| Redux Toolkit | Complex global + middleware | 11 KB | High | Large |
Server State with TanStack Query
// api/users.ts — query key factory pattern
export const userKeys = {
all: ['users'] as const,
lists: () => [...userKeys.all, 'list'] as const,
list: (filters: Filters) => [...userKeys.lists(), filters] as const,
details: () => [...userKeys.all, 'detail'] as const,
detail: (id: string) => [...userKeys.details(), id] as const,
}
// hooks/useUsers.ts
export function useUsers(filters: Filters) {
return useQuery({
queryKey: userKeys.list(filters),
queryFn: () => fetchUsers(filters),
staleTime: 5 * 60 * 1000, // 5 min
placeholderData: keepPreviousData,
})
}
export function useUpdateUser() {
const queryClient = useQueryClient()
return useMutation({
mutationFn: updateUser,
onMutate: async (newUser) => {
// Optimistic update
await queryClient.cancelQueries({ queryKey: userKeys.detail(newUser.id) })
const previous = queryClient.getQueryData(userKeys.detail(newUser.id))
queryClient.setQueryData(userKeys.detail(newUser.id), newUser)
return { previous }
},
onError: (err, newUser, context) => {
queryClient.setQueryData(userKeys.detail(newUser.id), context?.previous)
},
onSettled: (data, err, variables) => {
queryClient.invalidateQueries({ queryKey: userKeys.detail(variables.id) })
queryClient.invalidateQueries({ queryKey: userKeys.lists() })
},
})
}
Client State with Zustand
// stores/useUIStore.ts — thin, focused stores
interface UIStore {
sidebarOpen: boolean
theme: 'light' | 'dark' | 'system'
toggleSidebar: () => void
setTheme: (theme: UIStore['theme']) => void
}
export const useUIStore = create<UIStore>()(
persist(
(set) => ({
sidebarOpen: true,
theme: 'system',
toggleSidebar: () => set((s) => ({ sidebarOpen: !s.sidebarOpen })),
setTheme: (theme) => set({ theme }),
}),
{ name: 'ui-preferences' }
)
)
// Usage: const theme = useUIStore((s) => s.theme) — always use selectors!
5 State Management Rules
- Server state ≠ client state — never mix them in the same store
- Smallest scope possible — useState > Context > Zustand > Redux
- No useEffect for derived state — use useMemo or compute inline
- Selectors always —
useStore(s => s.field)notuseStore() - URL is state — search params, filters, pagination → URL, not React state
Phase 5: Hooks Engineering
Custom Hook Template
// hooks/useDebounce.ts
export function useDebounce<T>(value: T, delayMs: number = 300): T {
const [debouncedValue, setDebouncedValue] = useState(value)
useEffect(() => {
const timer = setTimeout(() => setDebouncedValue(value), delayMs)
return () => clearTimeout(timer)
}, [value, delayMs])
return debouncedValue
}
Essential Custom Hooks Library
| Hook | Purpose | When to Use |
|---|---|---|
useDebounce | Debounce value changes | Search inputs, resize |
useMediaQuery | Responsive breakpoints | Conditional rendering |
useLocalStorage | Persistent local state | Preferences, drafts |
useIntersection | Viewport detection | Lazy load, infinite scroll |
usePrevious | Track previous value | Animations, comparisons |
useClickOutside | Detect outside clicks | Dropdowns, modals |
useEventListener | Safe event binding | Keyboard, scroll, resize |
useToggle | Boolean state toggle | Modals, accordions |
Hook Rules (beyond React's rules)
- One concern per hook —
useUserSearchnotuseEverything - Return tuple or object — tuple for 1-2 values, object for 3+
- Accept options object —
useDebounce(value, { delay: 300 })scales better - Handle cleanup — every subscription/timer needs cleanup in useEffect return
- No hooks in conditions — extract conditional logic into the hook body
- Test hooks independently — use
renderHookfrom testing-library
Phase 6: TypeScript Integration
Strict Configuration
{
"compilerOptions": {
"strict": true,
"noUncheckedIndexedAccess": true,
"noImplicitOverride": true,
"exactOptionalPropertyTypes": true,
"forceConsistentCasingInFileNames": true,
"paths": {
"@/*": ["./src/*"]
}
}
}
Essential Type Patterns
// 1. Discriminated unions for state machines
type AsyncState<T> =
| { status: 'idle' }
| { status: 'loading' }
| { status: 'success'; data: T }
| { status: 'error'; error: Error }
// 2. Polymorphic components
type ButtonProps<C extends ElementType = 'button'> = {
as?: C
variant?: 'primary' | 'secondary'
} & ComponentPropsWithoutRef<C>
export function Button<C extends ElementType = 'button'>({
as,
variant = 'primary',
...props
}: ButtonProps<C>) {
const Component = as || 'button'
return <Component {...props} />
}
// 3. Branded types for IDs
type UserId = string & { __brand: 'UserId' }
type PostId = string & { __brand: 'PostId' }
// 4. Zod for runtime validation
const userSchema = z.object({
id: z.string().uuid(),
email: z.string().email(),
role: z.enum(['admin', 'user', 'viewer']),
})
type User = z.infer<typeof userSchema>
5 TypeScript Rules
- Zero
any— useunknownand narrow, or generics - Zod at boundaries — validate all external data (API, forms, URL params)
- Discriminated unions over optional fields —
{ status: 'success'; data: T }not{ data?: T; error?: Error } - Branded types for IDs — prevent
userIdbeing passed wherepostIdexpected - Satisfies over as —
config satisfies Configpreserves inference;as Configlies
Phase 7: Performance Optimization
Performance Budget
| Metric | Target | Measurement |
|---|---|---|
| First Contentful Paint | < 1.8s | Lighthouse |
| Largest Contentful Paint | < 2.5s | Lighthouse |
| Interaction to Next Paint | < 200ms | Lighthouse |
| Cumulative Layout Shift | < 0.1 | Lighthouse |
| Bundle size (gzipped) | < 200 KB | webpack-bundle-analyzer |
| JS execution (main thread) | < 3s | Chrome DevTools |
Optimization Priority Stack
| Priority | Technique | Impact | Effort |
|---|---|---|---|
| P0 | Code splitting (route-based) | 🔴 High | Low |
| P0 | Image optimization (next/image, srcset) | 🔴 High | Low |
| P1 | Tree shaking (named imports) | 🟡 Medium | Low |
| P1 | Virtualization for long lists | 🟡 Medium | Medium |
| P1 | Debounce expensive operations | 🟡 Medium | Low |
| P2 | React.memo on expensive components | 🟢 Low-Med | Low |
| P2 | useMemo/useCallback for expensive calculations | 🟢 Low-Med | Low |
| P3 | Web Workers for heavy computation | 🟢 Low | High |
Code Splitting Patterns
// 1. Route-based (automatic with Next.js, manual with React Router)
const Dashboard = lazy(() => import('./features/dashboard'))
const Settings = lazy(() => import('./features/settings'))
// 2. Component-based (heavy components)
const Chart = lazy(() => import('./components/Chart'))
const MarkdownEditor = lazy(() =>
import('./components/MarkdownEditor').then(m => ({ default: m.MarkdownEditor }))
)
// 3. Library-based (heavy third-party)
const { PDFViewer } = await import('@react-pdf/renderer')
React Compiler (React 19+)
// With React Compiler enabled, manual memo/useMemo/useCallback become unnecessary
// The compiler auto-memoizes. Remove manual optimizations:
// ❌ const memoized = useMemo(() => expensiveCalc(data), [data])
// ✅ const memoized = expensiveCalc(data) // compiler handles it
// Enable in babel config:
// plugins: [['babel-plugin-react-compiler', {}]]
Rendering Performance Rules
- Never create components inside components — define at module level
- Never create objects/arrays in JSX —
style={{ color: 'red' }}rerenders always - Children as props prevent rerender —
<Layout><ExpensiveChild /></Layout> - Key must be stable and unique — not index, not
Math.random() - Avoid context value churn — memoize provider value or split contexts
- Profile before optimizing — React DevTools Profiler, not guesswork
Phase 8: Error Handling & Resilience
Error Boundary Architecture
// Three levels of error boundaries:
// 1. App-level (catches everything, shows full-page error)
// 2. Feature-level (isolates feature failures)
// 3. Component-level (for risky widgets — charts, third-party)
// Modern error boundary with react-error-boundary
import { ErrorBoundary, FallbackProps } from 'react-error-boundary'
function FeatureErrorFallback({ error, resetErrorBoundary }: FallbackProps) {
return (
<div role="alert" className="rounded-lg border-red-200 bg-red-50 p-4">
<h3>Something went wrong</h3>
<pre className="text-sm text-red-600">{error.message}</pre>
<button onClick={resetErrorBoundary}>Try again</button>
</div>
)
}
// Usage:
<ErrorBoundary FallbackComponent={FeatureErrorFallback} onReset={() => queryClient.clear()}>
<DashboardFeature />
</ErrorBoundary>
Error Handling Checklist
- App-level error boundary wrapping entire app
- Feature-level boundaries for each major feature
- API errors handled in TanStack Query's
onError/ error states - Form validation errors shown inline (not alerts)
- 404 page for unknown routes
- Offline detection and graceful degradation
- Error reporting to monitoring (Sentry, etc.)
- User-friendly error messages (no stack traces in production)
Phase 9: Forms & Validation
Form Library Decision
| Library | Best For | Bundle | Renders |
|---|---|---|---|
| React Hook Form | Most forms | 9 KB | Minimal (uncontrolled) |
| Formik | Simple forms | 13 KB | Every keystroke |
| TanStack Form | Type-safe complex | 5 KB | Controlled |
| Native | 1-2 field forms | 0 KB | You control |
Default recommendation: React Hook Form + Zod
Form Pattern
const schema = z.object({
email: z.string().email('Invalid email'),
password: z.string().min(8, 'Min 8 characters'),
role: z.enum(['admin', 'user']),
})
type FormData = z.infer<typeof schema>
export function LoginForm({ onSubmit }: { onSubmit: (data: FormData) => void }) {
const form = useForm<FormData>({
resolver: zodResolver(schema),
defaultValues: { email: '', password: '', role: 'user' },
})
return (
<form onSubmit={form.handleSubmit(onSubmit)} noValidate>
<label htmlFor="email">Email</label>
<input id="email" type="email" {...form.register('email')} aria-invalid={!!form.formState.errors.email} />
{form.formState.errors.email && (
<p role="alert">{form.formState.errors.email.message}</p>
)}
{/* ... more fields */}
<button type="submit" disabled={form.formState.isSubmitting}>
{form.formState.isSubmitting ? 'Signing in...' : 'Sign in'}
</button>
</form>
)
}
Phase 10: Testing Strategy
Test Pyramid for React
| Level | Tool | Coverage Target | What to Test |
|---|---|---|---|
| Unit | Vitest | 80% business logic | Hooks, utilities, reducers |
| Component | Testing Library | Key user flows | Rendering, interactions, a11y |
| Integration | Testing Library | Feature flows | Multi-component workflows |
| E2E | Playwright | Critical paths | Auth, checkout, core flows |
| Visual | Chromatic/Percy | UI components | Regression detection |
Testing Patterns
// Component test (Testing Library philosophy: test behavior, not implementation)
import { render, screen } from '@testing-library/react'
import userEvent from '@testing-library/user-event'
describe('UserCard', () => {
it('calls onEdit when edit button clicked', async () => {
const user = userEvent.setup()
const onEdit = vi.fn()
render(<UserCard user={mockUser} onEdit={onEdit} />)
await user.click(screen.getByRole('button', { name: /edit/i }))
expect(onEdit).toHaveBeenCalledWith(mockUser.id)
})
it('does not render edit button when onEdit not provided', () => {
render(<UserCard user={mockUser} />)
expect(screen.queryByRole('button', { name: /edit/i })).not.toBeInTheDocument()
})
})
7 Testing Rules
- Test behavior, not implementation — never test state directly or useEffect
- Use accessible queries —
getByRole>getByTestId>getByText - User events over fireEvent —
userEvent.clicksimulates real interaction - One assertion per concept — not one per test, but focused assertions
- Mock at boundaries — API calls, not internal functions
- No snapshot tests — they break on every change and test nothing meaningful
- Arrange-Act-Assert — clear structure in every test
Phase 11: Accessibility (a11y)
10-Point Accessibility Checklist
- Semantic HTML —
<button>not<div onClick>,<nav>not<div class="nav"> - Keyboard navigation — every interactive element reachable via Tab, operable via Enter/Space
- Focus management — visible focus indicator, logical tab order, focus trap in modals
- Alt text — every
<img>has descriptive alt (oralt=""if decorative) - Color contrast — 4.5:1 for normal text, 3:1 for large text (WCAG AA)
- ARIA labels —
aria-labelfor icon-only buttons,aria-describedbyfor hints - Live regions —
aria-live="polite"for dynamic content (toasts, form errors) - Reduced motion — respect
prefers-reduced-motionfor animations - Screen reader testing — test with VoiceOver (Mac) or NVDA (Windows)
- Automated scanning — axe-core in CI (
vitest-axeor@axe-core/playwright)
Phase 12: Production Deployment Checklist
Mandatory (P0)
- TypeScript strict mode, zero errors
- All tests passing
- Bundle analyzed, no unexpected large dependencies
- Error boundaries at app and feature level
- Environment variables validated at build time
- Security headers configured (CSP, HSTS, X-Frame-Options)
- SEO meta tags (title, description, OG tags)
- Analytics/error monitoring integrated
- Performance budget met (LCP < 2.5s)
Recommended (P1)
- Storybook for component library
- Visual regression tests
- a11y automated checks in CI
- Feature flags for risky features
- Preview deployments for PRs
- Bundle size CI check (fail if +10%)
Recommended Stack (2025+)
| Layer | Recommendation | Alternative |
|---|---|---|
| Framework | Next.js 15 | Remix, Vite SPA |
| Language | TypeScript (strict) | — |
| Styling | Tailwind CSS v4 | CSS Modules |
| Components | shadcn/ui | Radix, Headless UI |
| State (server) | TanStack Query v5 | SWR |
| State (client) | Zustand | Jotai |
| Forms | React Hook Form + Zod | TanStack Form |
| Testing | Vitest + Testing Library | Jest |
| E2E | Playwright | Cypress |
| Linting | Biome | ESLint + Prettier |
| Auth | Auth.js (NextAuth) | Clerk, Lucia |
| Database | Drizzle ORM | Prisma |
| Deployment | Vercel | Cloudflare, Fly.io |
| Monitoring | Sentry | Datadog |
Quality Scoring (0-100)
Shortened here. Read the whole file on GitHub.
Signals
- GitHub stars
- 2k
- Forks
- 324
- Last commit
- Jul 2026
Advanced
- Catalog kind
- skill
- Gateway key
afrexai-react-production- Source
- github.com/leoyeai/openclaw-master-skills