react-advanced
React Advanced: Core Patterns & Conventions (Cross-Platform)
This skill defines the rules, conventions, and architectural decisions for building modern React applications with the TanStack ecosystem and XState. It is intentionally opinionated to prevent common pitfalls and enforce patterns that scale.
These patterns work identically on web and React Native. For platform-specific patterns:
- Web: see
react-web-advanced(TanStack Router, Start, Virtual) - React Native: see
react-native-advanced(Expo Router, FlashList, MMKV)
For detailed API documentation of any library mentioned here, use other appropriate tools (documentation lookup, web search, etc.) — this skill focuses on how and why to use these tools, not their full API surface.
The useEffect Ban
Do not use useEffect for:
- Data fetching — use React Query (
useSuspenseQuery/useQuery) or route loaders - Derived state — compute during render or use
useMemo - Syncing state with props — use the prop directly, or reset with React
key - Responding to user events — put logic in event handlers
- Subscribing to external stores — use
useSyncExternalStore - Complex async flows — use XState machines with
invoke/fromPromise
Acceptable uses of useEffect
Legitimate cases:
- Synchronizing with non-React external systems — DOM APIs, third-party widgets (maps, charts), imperative libraries that need mount/unmount lifecycle
- Browser/native API subscriptions with cleanup — when
useSyncExternalStoreis too low-level for a one-off case (WebSocket connections, resize observers) - Analytics/logging on mount — fire-and-forget side effects with no state updates
- Bridging React Query data into XState — the
useEffectbridge pattern for pushing server state into a machine via events (seereferences/xstate.md)
useMountEffect for mount-only effects
When you have a legitimate mount-only effect (cases 1–3 above), use a useMountEffect
helper instead of raw useEffect(fn, []):
// utils/useMountEffect.ts
import { useEffect, type EffectCallback } from "react";
// eslint-disable-next-line react-hooks/exhaustive-deps
const useMountEffect = (effect: EffectCallback) => useEffect(effect, []);
export default useMountEffect;
Usage:
useMountEffect(() => {
const plugin = $.myPlugin(ref.current);
return () => {
plugin.destroy();
};
});
Why: raw useEffect(fn, []) triggers the react-hooks/exhaustive-deps lint rule and
makes the developer prove the empty array is intentional. useMountEffect makes the
"run once on mount" intent explicit in code and silences the warning correctly.
What to use instead
| Instead of useEffect for... | Use |
|---|---|
| Fetching data | useSuspenseQuery + route loader prefetch |
| Consuming a Promise | use() hook (React 19+) + <Suspense> |
| Derived/computed values | Direct computation or useMemo |
| External store subscription | useSyncExternalStore |
| Deferring expensive renders | useDeferredValue / useTransition |
| Complex async orchestration | XState invoke with fromPromise |
| Resetting state on prop change | React key prop on the component |
| User-triggered side effects | Event handlers directly |
State Management Philosophy
Server state vs client state — never mix them
Server state (data from APIs/databases) and client state (UI state existing only in the client) are fundamentally different concerns. Mixing them causes stale data bugs, duplication, and synchronization nightmares.
| Concern | Owner | Examples |
|---|---|---|
| Server data | React Query | Users, posts, products, orders |
| URL/route state | Router | Path params, search params (TanStack Router or Expo Router) |
| Complex UI flows | XState | Multi-step wizards, auth flows, drag-and-drop |
| Shared client UI | Zustand | Theme, sidebar, selected items, global filters, preferences |
| Form fields | TanStack Form | Input values, validation errors, submission |
| Schema validation | Zod | Search params, form validators, API contracts |
| Simple local UI | useState |
Toggle, accordion expanded, input focus |
Decision flowchart
Is the data from a server / API?
YES -> React Query (queryOptions + useSuspenseQuery)
NO -> Is it in the URL / route params?
YES -> Router (platform-specific: TanStack Router or Expo Router)
NO -> Is it a complex multi-state flow (3+ states, async, guards)?
YES -> XState (useMachine / createActorContext)
NO -> Is it a form field?
YES -> TanStack Form (with Zod validators)
NO -> Is it shared across components / trees?
YES -> Zustand (create store + selectors)
NO -> useState / useReducer
When to reach for XState over useState/useReducer
Use XState when:
- There are 3+ mutually exclusive states with defined transitions
- Async side effects must be cancelled on state change (race conditions)
- The logic has guards (conditions that gate transitions)
- You need parallel states for independent concerns
- The flow needs to be tested in isolation from React
- Multiple steps with back/forward navigation (wizards)
Do not use XState for simple toggles, single boolean flags, or counter state. That is
useState territory.
Architecture: Which Library Owns What
| Layer | Library | Responsibility |
|---|---|---|
| Server state | React Query | Fetching, caching, invalidation, background refetch |
| Complex UI state | XState | State machines, actor model, flow orchestration |
| Shared client UI | Zustand | Cross-component UI state, preferences, selections |
| Form lifecycle | TanStack Form | Field values, validation, submission |
| Schema validation | Zod | Search params, form validators, API contracts |
| Data display | TanStack Table | Headless sorting, filtering, pagination, grouping |
| Testing | Vitest + TL | Unit, component, integration, machine testing |
| Simple local state | useState | Toggles, local inputs, component-scoped values |
The golden rule: queryOptions as single source of truth
Define query options once, import everywhere — loaders, components, invalidation:
export const postsQueryOptions = queryOptions({
queryKey: ["posts"],
queryFn: fetchPosts,
staleTime: 30_000,
});
export const postQueryOptions = (postId: string) =>
queryOptions({
queryKey: ["posts", postId],
queryFn: () => fetchPost(postId),
staleTime: 30_000,
});
Component Composition
Compound components
Use Context-based compound components when a group of components shares implicit state. The parent manages state; children consume it through context. Memoize the context value.
Slots pattern
Use named props for slot-like composition (header, footer, actions). Avoid deeply
nested render-prop trees.
Inversion of Control
When adding boolean props or branching logic to handle caller-specific behavior, push that logic back to the caller via callbacks, reducers, or render functions. Three similar if-statements is a signal to invert control.
Common Pitfalls
-
Derived state in useEffect — computing values in an effect and storing in useState causes double renders. Compute during render or use
useMemo. -
Storing server data in client state — putting API responses in Zustand/Redux means you own caching and invalidation. Use React Query instead.
-
Duplicating URL state in useState — use your router's search/param hooks directly.
-
Using React Query AND XState for the same data — React Query owns fetching/caching. XState receives data via events and handles orchestration only.
-
Calling React hooks inside XState machines — hooks only work in React components. Use
fromPromisein machines, bridge data viauseEffectevents. -
Array indices as keys — use stable IDs (
item.id). Index keys cause incorrect state association on reorder/insert/delete. -
Defining components inside components — creates new component types each render, forcing React to unmount/remount. Define at module level.
-
Context for high-frequency state — Context re-renders all consumers on every change. Use Zustand with selectors for shared rapidly-changing values (see
references/zustand.md), or local state if component-scoped. -
Not using
.catch()on Zod search param schemas —.default()only handles missing keys;.catch()also handles invalid values from malformed URLs.
Reference Files
Read the relevant reference file when working with a specific library:
| File | When to read |
|---|---|
references/react-query.md |
Query patterns, mutations, cache, Suspense integration |
references/table.md |
Column defs, sorting, filtering, pagination, server-side ops |
references/form.md |
Field validation, arrays, schema validation, performance |
references/xstate.md |
State machines, actors, auth flows, wizards, React integration |
references/zustand.md |
Shared client UI state, selectors, slices, middleware, vanilla stores |
references/zod.md |
Schema validation, v4 API, Form integration, error handling |
references/testing.md |
Vitest setup, Testing Library, MSW, testing Query/Form/XState |
references/integration.md |
Combining libraries: Zustand+XState, Query+XState bridge |
More from trancong12102/agentskills
deps-dev
Look up the latest stable version of any open-source package across npm, PyPI, Go, Cargo, Maven, and NuGet. Use when the user asks 'what's the latest version of X', 'what version should I use', 'is X deprecated', 'how outdated is my package.json/requirements.txt/Cargo.toml', or needs version numbers for adding or updating dependencies. Also covers pinning versions, checking if packages are maintained, or comparing installed vs latest versions. Do NOT use for private/internal packages or for looking up documentation (use context7).
151council-review
Multi-perspective code review that synthesizes findings from multiple reviewers into a unified report. Use when the user asks to review code changes, audit a diff, check code quality, review a PR, review commits, or review uncommitted changes. Also covers 'code review', 'review my changes', 'check this before I merge', or wanting multiple perspectives on code. Do not use for documentation/markdown review or trivial single-line changes.
94oracle
Deep analysis and expert reasoning. Use when the user asks for 'oracle', 'second opinion', architecture analysis, elusive bug debugging, impact assessment, security reasoning, refactoring strategy, or trade-off evaluation — problems that benefit from deep, independent reasoning. Do not use for simple factual questions, code generation, code review (use council-review), or tasks needing file modifications.
92context7
Fetch up-to-date documentation for any open-source library or framework. Use when the user asks to look up docs, check an API, find code examples, or verify how a feature works — especially with a specific library name, version migration, or phrases like 'what's the current way to...' or 'the API might have changed'. Also covers setup and configuration docs. Do NOT use for general programming concepts, internal project code, or version lookups (use deps-dev).
86conventional-commit
Generates git commit messages following Conventional Commits 1.0.0 specification with semantic types (feat, fix, etc.), optional scope, and breaking change annotations. Use when committing code changes or creating commit messages.
58react-web-advanced
Web-specific React patterns for type-safe file-based routing, route-level data loading, server-side rendering, search param validation, code splitting, and list virtualization. Use when building React web apps with route loaders, SSR streaming, validated search params, lazy route splitting, or virtualizing large DOM lists. Do not use for React Native apps — use react-native-advanced instead.
45