React Advanced logo

React Advanced

Community
bobmatnyc
react-advanced

Modern React 19 platform features and rendering architecture - the React Compiler (auto-memoization) and how it changes memoization guidance, concurrent rendering (useTransition/useDeferredValue/Suspense), the use() hook, Actions (useActionState/useOptimistic/useFormStatus), ref-as-prop and Context-as-provider changes, compound components, context selectors, error boundaries, portals, virtualization, code-splitting, and the RSC/'use client' boundary.

Overview

Publisherbobmatnyc
Repositoryclaude-mpm-skills
Skill namereact-advanced
Stars
75
Forks
19
Bundled files
7
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 7 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by bobmatnyc on GitHub. Read the source before you install it.

Installation

Install the React Advanced AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/bobmatnyc/claude-mpm-skills.git /tmp/claude-mpm-skills
mkdir -p .claude/skills
cp -r /tmp/claude-mpm-skills/toolchains/javascript/frameworks/react/react-advanced .claude/skills/react-advanced
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable React Advanced in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads React Advanced on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so React Advanced is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

React Advanced: React 19 Platform & Rendering Architecture

Overview

The reference for React-19-era platform capabilities and rendering architecture. This skill covers the features that reshape how modern React applications are built and optimized: the React Compiler and the end of reflexive manual memoization, concurrent rendering primitives, the use() hook, Actions and the form hooks, the React 19 ergonomic changes (ref as a plain prop, <Context> as provider, native metadata), advanced component-architecture patterns, rendering-performance engineering, and a conceptual treatment of the React Server Component boundary.

This guidance treats React 19 (Dec 2024) and React Compiler 1.0 (stable Oct 2025) as current. Pre-19 memoization advice — wrapping values in useMemo/useCallback/React.memo by reflex — is now an anti-pattern for new code. The compiler memoizes automatically. Authoritative documentation lives at react.dev; this skill distills and applies it rather than restating it.

Boundary Map: What This Skill Does NOT Cover

This skill stays focused on the React-19 platform and rendering architecture. Defer to the owning skill for everything else:

TopicOwning skill
Core hooks tutorial (useState/useEffect/useRef/useMemo/useCallback/useContext), JSX, props/state, lists, controlled inputsreact-core
Advanced custom-hook composition recipes — SWR-backed data hooks, debounced search, memoized-provider value pattern, discriminated-union hook statesreact-hooks-composition
Explicit finite-state-machine modeling with XState v5 (setup(), actors, guards, parallel states)react-state-machine
Server-state caching, query invalidation, optimistic updates via React Querytanstack-query
Global client state via a store with no providerszustand
Framework wiring of Server Components, Server Actions (revalidateTag/revalidatePath), App Router data fetching, caching, TurbopackNext.js skills (nextjs-core, nextjs-v16)
Docking layout UI (drag-drop panels, splitters)flexlayout-react

This skill explains the React-level concepts (concurrent rendering, the RSC boundary as a React feature, Actions as a React primitive). Framework-specific wiring of those concepts belongs to the Next.js skills.

When to Use This Skill

Reach for this skill when:

  • Adopting the React Compiler and untangling the old memoization habits
  • Using concurrent features (useTransition, useDeferredValue, Suspense for data) to keep a UI responsive
  • Building with React 19 Actions, forms, or the use() hook
  • Optimizing render performance: context over-rendering, large lists, code-splitting
  • Architecting compound components, context selectors, error boundaries, or portals
  • Reasoning about the RSC / 'use client' boundary at the React level

The React Compiler — and the End of Manual Memoization

The React Compiler reached 1.0 stable on 2025-10-07. It is a build-time tool that auto-memoizes components and hooks at fine granularity, including conditional memoization that hand-written useMemo cannot express. It works with React 17, 18, and 19, and performs best on 19.

The change vs React 17/18: the long-standing advice to wrap values in useMemo/useCallback and components in React.memo is obsolete for new code. The compiler handles memoization. Scattering manual memo by reflex now adds noise and can be net-negative.

tsx
// BEFORE (pre-compiler reflex): hand-memoize everything
const sorted = useMemo(() => items.slice().sort(compare), [items]);
const onPick = useCallback((id: string) => onSelect(id), [onSelect]);
const Row = React.memo(function Row({ item }: { item: Item }) { /* ... */ });
tsx
// AFTER (compiler enabled): write plain code; the compiler memoizes
const sorted = items.slice().sort(compare);
const onPick = (id: string) => onSelect(id);
function Row({ item }: { item: Item }) { /* ... */ }

When manual memo still matters (the escape hatch):

  • A computed value feeds an effect dependency array and must hold a stable reference for correctness, not just performance.
  • Code the compiler cannot statically analyze — it safely skips such components rather than miscompiling them, so manual memo remains a valid local optimization.

Existing code: the compiler preserves manual memoization on purpose; the preserve-manual-memoization lint flags memo that was load-bearing. Do not strip useMemo/useCallback blindly after enabling the compiler.

Adoption path: install eslint-plugin-react-hooks v6+ (it absorbed the former eslint-plugin-react-compiler) → fix the reported Rules of React violations → enable the compiler in the build (Babel / Vite / Metro / Rsbuild). React DevTools shows a "Memo ✨" badge on compiler-optimized components.

Full adoption guide, per-bundler setup, and before/after analysis: react-compiler.md.

Concurrent Rendering

Concurrent features let React interrupt and prioritize rendering so urgent updates (typing) stay responsive while expensive updates (filtering a large list) yield.

  • useTransition marks a state update as non-urgent and exposes isPending. React keeps the old UI interactive while the transition renders in the background.
  • React 19 async transitions: startTransition accepts async functions. The gotcha: state updates issued after an await fall outside the transition unless re-wrapped in startTransition.
  • useDeferredValue produces a lagging copy of a value and is interruptible and device-adaptive. Prefer it over a fixed setTimeout debounce for expensive derived renders; React 19 adds an initialValue.
  • Suspense for data suspends on a thrown promise (or one read via use()) and shows the nearest <Suspense fallback>. A changing key resets a boundary.
tsx
function ProductSearch({ allProducts }: { allProducts: Product[] }) {
  const [query, setQuery] = useState('');
  const deferredQuery = useDeferredValue(query); // lags behind input under load

  // Filtering reads the deferred value, so typing stays responsive.
  const results = allProducts.filter((p) => p.name.includes(deferredQuery));

  return (
    <>
      <input value={query} onChange={(e) => setQuery(e.target.value)} />
      <ResultsList results={results} stale={query !== deferredQuery} />
    </>
  );
}

Streaming SSR at the React level (renderToPipeableStream, shell design, Suspense reveal batching in 19.2) is covered in concurrent-rendering.md; framework streaming specifics defer to the Next.js skills.

React 19 Platform Hooks & APIs

The use() hook

use() reads a promise or a context value and may be called conditionally — after early returns — a deliberate exception to the Rules of Hooks. It cannot sit inside try/catch; surface rejected promises through an Error Boundary instead. Prefer creating the promise high in the tree (or on the server) and passing it down, since a promise created during render is recreated each render.

Actions

Actions are the React-level async-mutation primitive, independent of any framework:

  • useActionState (renamed from the canary useFormState) runs an action, tracking pending and error state and threading the previous result forward.
  • useOptimistic renders an optimistic value while the action is in flight, reverting automatically on completion.
  • useFormStatus reads the pending state of the enclosing <form> from a child, with no prop drilling.
  • <form action={fn}> wires a function directly to submission, auto-resetting on success.
tsx
function NameForm({ save }: { save: (name: string) => Promise<string> }) {
  const [error, submit, isPending] = useActionState(
    async (_prev: string | null, formData: FormData) => {
      const result = await save(formData.get('name') as string);
      return result.startsWith('error') ? result : null;
    },
    null,
  );

  return (
    <form action={submit}>
      <input name="name" disabled={isPending} />
      {error && <p role="alert">{error}</p>}
    </form>
  );
}

Ergonomic changes

  • ref is a plain prop. forwardRef is deprecated (a codemod migrates existing usage); a function component receives ref like any other prop.
  • <Context value> replaces <Context.Provider>, which is now legacy.
  • Native document metadata: <title>, <meta>, and <link> hoist to <head> automatically, replacing react-helmet. preload/preinit/preconnect handle resource hints.

Complete examples and migration notes from 17/18: react19-actions-and-apis.md.

Advanced Component Patterns

  • Compound components share implicit state through context so a parent and its sub-components coordinate without prop drilling (<Tabs> / <Tabs.Tab>).
  • Context optimization: native context has no selector. Any change to a provider value re-renders all consumers; React.memo does not block context-driven re-renders, and use(Context) adds conditional reading but not selective subscription. The compiler does not fix large-context over-rendering. The native fixes are memoizing the provider value and splitting contexts by change frequency; the userland use-context-selector adds slice subscriptions (with tearing/legacy caveats).
  • Render props vs custom hooks: hooks win for logic reuse; render props remain useful for headless "what to render" components.
  • Error boundaries are still class-based; react-error-boundary adds a hook (useErrorBoundary) for async and event-handler errors that native boundaries miss. React 19 logs caught errors once and adds onCaughtError/onUncaughtError.
  • Portals (createPortal) change DOM placement only — context and errors still flow through the React tree, not the DOM tree.

Decision guides and full examples: component-patterns.md.

Performance Architecture

  • Virtualization renders only visible rows. @tanstack/react-virtual is the modern headless choice with the strongest TypeScript story; react-window is mature but stalled; react-virtuoso is batteries-included. After virtualizing, the per-item renderer cost becomes the bottleneck — keep rows light.
  • Code-splitting with React.lazy + Suspense splits by route first. Avoid over-splitting tiny components into many sub-10KB chunks, which adds request overhead without payoff.
  • Profiling uses the React DevTools Profiler against a production build to see what actually re-rendered and why; Why-Did-You-Render helps trace avoidable renders.

Library comparison, code-splitting strategy, and the full anti-pattern → detection → fix table: performance.md.

RSC / 'use client' Boundary (Conceptual)

React Server Components render ahead of time, cannot hold state or run effects, and produce serializable output. The boundary is defined on the module dependency graph (a 'use client' directive at the top of a module), not on the render tree. Props crossing into a Client Component must be serializable, and a Client Component may still render a Server Component passed to it as children. The 'use server' directive marks Server Functions whose arguments are untrusted and must be validated and authorized.

Explicit defer: caching, route loaders, and App Router data fetching belong to the Next.js skills. Pin the React version, since RSC bundler APIs are not semver-stable within 19.x. Conceptual model and the framework hand-off: rsc-boundary.md.

Anti-Patterns to Avoid

❌ Don't: Hand-memoize everything with the compiler enabled

tsx
// BAD: redundant manual memo once the compiler is on; adds noise and cost
const value = useMemo(() => ({ a, b }), [a, b]);
const onClick = useCallback(() => doThing(a), [a]);
tsx
// GOOD: write plain code; let the compiler memoize
const value = { a, b };
const onClick = () => doThing(a);

❌ Don't: Strip manual memo blindly after enabling the compiler

tsx
// BAD: removing memo that stabilized an effect dependency changes behavior
const config = { endpoint, token }; // recreated each render → effect re-fires
useEffect(() => subscribe(config), [config]);
tsx
// GOOD: keep the deliberate escape hatch the preserve-manual-memoization lint protects
const config = useMemo(() => ({ endpoint, token }), [endpoint, token]);
useEffect(() => subscribe(config), [config]);

❌ Don't: Put high-frequency state in one large context

tsx
// BAD: every consumer re-renders when any field changes; memo cannot block it
const AppContext = createContext<{ theme: Theme; cursor: Point } | null>(null);
tsx
// GOOD: split by change frequency so fast-changing state is isolated
const ThemeContext = createContext<Theme | null>(null);   // changes rarely
const CursorContext = createContext<Point | null>(null);  // changes often

❌ Don't: Update state after await inside an async transition

tsx
// BAD: the post-await update escapes the transition; isPending ends early
startTransition(async () => {
  const data = await load();
  setData(data); // not part of the transition
});
tsx
// GOOD: re-wrap the post-await update
startTransition(async () => {
  const data = await load();
  startTransition(() => setData(data));
});

❌ Don't: Keep forwardRef / <Context.Provider> / react-helmet in new React 19 code

tsx
// BAD: superseded idioms in new code
const Input = forwardRef<HTMLInputElement, Props>((props, ref) => <input ref={ref} {...props} />);
<ThemeContext.Provider value={theme}>{children}</ThemeContext.Provider>;
tsx
// GOOD: ref as a plain prop, <Context> as provider, native metadata
function Input({ ref, ...props }: Props & { ref?: React.Ref<HTMLInputElement> }) {
  return <input ref={ref} {...props} />;
}
<ThemeContext value={theme}>{children}</ThemeContext>;

The & { ref?: ... } intersection is only needed when Props is a custom interface that must expose ref. For wrappers over an intrinsic element, deriving Props from React.ComponentProps<'input'> already includes a correctly-typed ref, so the intersection is redundant.

When to Reach for Sibling Skills

  • react-core: fundamentals — components, JSX, props/state, core hooks, lists, controlled inputs.
  • react-hooks-composition: custom-hook composition recipes — SWR data hooks, debounced search, memoized providers, async-state hooks.
  • react-state-machine: explicit XState v5 finite-state machines for complex flows where impossible states must be unrepresentable.
  • tanstack-query / zustand: server-state caching and global client state, respectively.
  • Next.js skills: framework wiring of Server Components, Server Actions, routing, and caching.

Best Practices Summary

  1. Compiler first: enable eslint-plugin-react-hooks v6+, fix Rules of React, then turn on the React Compiler.
  2. Stop reflexive memoization: reach for useMemo/useCallback/React.memo only as a deliberate escape hatch (effect-dependency stability, or unanalyzable code).
  3. Stay responsive with concurrency: prefer useTransition/useDeferredValue over manual debounce for expensive renders; re-wrap post-await updates.
  4. Use Actions for mutations: useActionState + useOptimistic + useFormStatus cover pending, error, and optimistic states at the React level.
  5. Read use() correctly: call it conditionally, never in try/catch, and create promises high in the tree.
  6. Adopt React 19 ergonomics: ref as a prop, <Context value>, native metadata — drop forwardRef and react-helmet.
  7. Fix context over-rendering structurally: memoize the value and split contexts; the compiler does not solve this.
  8. Engineer list and bundle performance: virtualize large lists, code-split by route, and profile against production builds.
  9. Respect the RSC boundary: it is module-graph-based; keep crossing props serializable, validate server-function inputs, and pin the React version.

Navigation

  • React Compiler: adoption per bundler, ESLint/Rules of React, what the compiler skips, preserve-manual-memoization, before/after examples.
  • Concurrent Rendering: useTransition (sync/async + post-await gotcha), useDeferredValue, Suspense-for-data, streaming SSR shells, reveal batching, boundary key resets.
  • React 19 Actions & APIs: use(), useActionState, useOptimistic, useFormStatus, <form action>, ref-as-prop, useImperativeHandle, <Context value>, metadata/preloading, with migration notes.
  • Component Patterns: compound components, context selectors/splitting + use-context-selector, render-props-vs-hooks, error boundaries, portals.
  • Performance: virtualization library comparison, code-splitting strategy, Profiler workflow, anti-pattern → detection → fix catalog.
  • RSC Boundary: conceptual 'use client'/'use server' model, serialization rules, server-function security, Next.js hand-off, version pinning.

References

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the React Advanced AI skill do?

Modern React 19 platform features and rendering architecture - the React Compiler (auto-memoization) and how it changes memoization guidance, concurrent rendering (useTransition/useDeferredValue/Suspense), the use() hook, Actions (useActionState/useOptimistic/useFormStatus), ref-as-prop and Context-as-provider changes, compound components, context selectors, error boundaries, portals, virtualization, code-splitting, and the RSC/'use client' boundary.

Why use React Advanced on TypingMind?

Because you install it once and use it with any model. React Advanced is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install React Advanced in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/bobmatnyc/claude-mpm-skills/tree/main/toolchains/javascript/frameworks/react/react-advanced. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use React Advanced?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with React Advanced?

As many as you like. As long as a model supports skills, you can use React Advanced with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the React Advanced AI skill free?

Yes. It is published on GitHub by bobmatnyc under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇