agentleFS
waynesuttonwaynesutton/markdown-site/.claude/react-effect-decision/SKILL.md

react-effect-decision

Skill628 starsChanged 7 months agoMITView on GitHub

Combine React's official "You Might Not Need an Effect" guidance with this project's stricter no direct useEffect stance. Use when writing, reviewing, or refactoring React components that might reach for useEffect, derived state, event relays, reset logic, subscriptions, or client fetching.

The file

Show its rules and text

We split this skill into the rules it gives an agent

25 rules. About 0 are common, 18 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈18

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Default rule for this project: do not reach for direct useEffect first.

    line 8
  2. Distinctive:

    Before writing any effect, ask these in order:

    line 19
  3. Very specific:

    If you answer yes to any of 1 through 5, do not write a direct useEffect.

    line 28
  4. Very specific:

    Avoid direct useEffect in components whenever a clearer pattern exists

    line 34
  5. Very specific:

    Prefer derived values, event handlers, key remounts, memoization, query libraries, Convex hooks, and focused custom hooks

    line 35
  6. Very specific:

    If mount only external setup is truly needed, use an explicit useMountEffect wrapper instead of sprinkling raw useEffect through feature …

    line 36
  7. Distinctive:

    Use that only for one time external setup like DOM focus, third party widgets, or browser API listeners that belong to the component …

    line 48
  8. Distinctive:

    If it can be calculated from existing inputs, keep it out of state.

    line 70
  9. Distinctive:

    If the calculation is pure and expensive, memoize it instead of syncing it into state:

    line 74
  10. Distinctive:

    Use memoization only when there is real repeated work to skip.

    line 83
  11. Distinctive:

    Wait for an effect to notice the flag

    line 90
  12. Distinctive:

    If the work happens because the user clicked, submitted, dragged, or selected, keep it in the handler.

    line 106
  13. Very specific:

    If a different entity should feel like a fresh component instance, give the inner component a key.

    line 122
  14. Distinctive:

    If you cannot derive it and do not want a full remount, adjust the state during render of the same component before children render stale …

    line 126
  15. Distinctive:

    Prefer these, in order:

    line 128
  16. Distinctive:

    Only then use a guarded render time adjustment pattern

    line 133
  17. Distinctive:

    Use Convex React hooks where the data already lives in Convex

    line 144
  18. Distinctive:

    Use the project's existing data abstraction or a query library for remote APIs

    line 145
  19. Distinctive:

    If raw client fetching is unavoidable, isolate it inside a custom hook and handle stale response cleanup

    line 146
  20. Distinctive:

    When you use one, document the external system in a comment or nearby helper.

    line 201
  21. Very specific:

    If you are about to write useEffect, check:

    line 205
  22. Distinctive:

    If you cannot name the external system, you probably do not need an effect.

    line 215
  23. Distinctive:

    Use these during review:

    line 219
  24. Distinctive:

    Should this be a custom hook instead of inline effect code

    line 225
  25. Distinctive:

    Use dedicated hooks for fetching and subscriptions

    line 236

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

React effect decision

Default rule for this project: do not reach for direct useEffect first.

Most effect usage in React code is a sign that the component is syncing state that should be derived during render, routing an event through state, or resetting local state in a way React already models better.

This skill merges:

  • React's official decision tree from "You Might Not Need an Effect"
  • The project's stricter no direct useEffect preference

Start here

Before writing any effect, ask these in order:

  1. Can I calculate this from props or state during render
  2. Is this triggered by a user action and better placed in the event handler
  3. Am I trying to reset component state when identity changed and should use key
  4. Is this an expensive pure calculation that should use useMemo
  5. Is this an external subscription that should use useSyncExternalStore or a focused custom hook
  6. Is this real external synchronization that must happen on mount or on dependency changes

If you answer yes to any of 1 through 5, do not write a direct useEffect.

Project stance

React allows effects for real external synchronization. This project is stricter:

  • Avoid direct useEffect in components whenever a clearer pattern exists
  • Prefer derived values, event handlers, key remounts, memoization, query libraries, Convex hooks, and focused custom hooks
  • If mount only external setup is truly needed, use an explicit useMountEffect wrapper instead of sprinkling raw useEffect through feature code

Example wrapper:

import { useEffect } from "react";

export function useMountEffect(effect: () => void | (() => void)) {
  useEffect(effect, []);
}

Use that only for one time external setup like DOM focus, third party widgets, or browser API listeners that belong to the component lifecycle.

Decision guide

1. Derived state belongs in render

Bad smell:

  • useEffect(() => setX(deriveFromY(y)), [y])
  • State that only mirrors props or other state

Do this instead:

function TodoList({ todos, showActive }: Props) {
  const activeTodos = todos.filter((todo) => !todo.completed);
  const visibleTodos = showActive ? activeTodos : todos;

  return <List items={visibleTodos} />;
}

If it can be calculated from existing inputs, keep it out of state.

2. Expensive pure calculations use useMemo

If the calculation is pure and expensive, memoize it instead of syncing it into state:

const visibleTodos = useMemo(
  () => getVisibleTodos(todos, showActive),
  [todos, showActive],
);

Use memoization only when there is real repeated work to skip.

3. Event caused work belongs in the event handler

Bad smell:

  • Set a flag in state
  • Wait for an effect to notice the flag
  • Reset the flag after the side effect runs

Do this instead:

function PurchaseButton({ product }: Props) {
  async function handleClick() {
    await addToCart(product);
    showNotification(`Added ${product.name}`);
  }

  return <button onClick={handleClick}>Buy</button>;
}

If the work happens because the user clicked, submitted, dragged, or selected, keep it in the handler.

4. Reset with key, not effect choreography

Bad smell:

  • useEffect(() => setFormState(initialFromProps), [id])

Do this instead:

function EditContact({ contact, onSave }: Props) {
  return <EditContactForm key={contact.id} contact={contact} onSave={onSave} />;
}

If a different entity should feel like a fresh component instance, give the inner component a key.

5. Adjusting state from prop changes is a last resort

If you cannot derive it and do not want a full remount, adjust the state during render of the same component before children render stale data.

Prefer these, in order:

  1. Derive it
  2. Reset with key
  3. Store an ID and derive the selected object from current inputs
  4. Only then use a guarded render time adjustment pattern

6. Data fetching should use the app's data layer or a focused hook

Bad smell:

  • Raw fetch(...).then(setState) inside a component effect
  • Manual loading, retry, cancellation, and stale response handling in feature code

Preferred options:

  • Use Convex React hooks where the data already lives in Convex
  • Use the project's existing data abstraction or a query library for remote APIs
  • If raw client fetching is unavoidable, isolate it inside a custom hook and handle stale response cleanup

Example custom hook pattern:

function useSearchResults(url: string) {
  const [data, setData] = useState<Result[] | null>(null);

  useEffect(() => {
    let ignore = false;

    fetch(url)
      .then((response) => response.json())
      .then((json) => {
        if (!ignore) {
          setData(json);
        }
      });

    return () => {
      ignore = true;
    };
  }, [url]);

  return data;
}

That is still less preferred than a dedicated query layer, but better than scattering raw fetch effects across components.

7. External subscriptions should use useSyncExternalStore

For browser or third party stores, prefer the React built in subscription model:

function useOnlineStatus() {
  return useSyncExternalStore(
    subscribe,
    () => navigator.onLine,
    () => true,
  );
}

This is clearer and safer than mirroring an external mutable source through ad hoc effect code.

8. True external synchronization is the narrow allowed case

An effect or useMountEffect is appropriate when the component must synchronize with something outside React:

  • DOM focus or scroll positioning
  • Third party widget setup and cleanup
  • Browser API subscriptions
  • Network synchronization that should happen because the component is on screen, not because a user event fired

When you use one, document the external system in a comment or nearby helper.

Quick smell test

If you are about to write useEffect, check:

  • Is this just derived state
  • Is this event specific logic
  • Is this a reset that should use key
  • Is this a pure expensive calculation
  • Is this a subscription better modeled with useSyncExternalStore
  • Is this data loading that belongs in Convex, a query library, or a custom hook
  • What external system am I synchronizing with

If you cannot name the external system, you probably do not need an effect.

Code review prompts

Use these during review:

  • What breaks if this effect is removed and the value is derived during render
  • Is this component storing data that can be recomputed from props or state
  • Why is this work not in the click, submit, or change handler that caused it
  • Would key={entityId} remove the reset logic entirely
  • Should this be a custom hook instead of inline effect code
  • Is this component talking to an external system, or only to React state

Summary

The safest path in this repo is simple:

  • Derive during render when possible
  • Handle user caused work in handlers
  • Reset with key
  • Memoize expensive pure work
  • Use dedicated hooks for fetching and subscriptions
  • Reserve effect style lifecycle code for real external synchronization

Sources:

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.