---

name: "react-native-code-review" description: "Review a React Native or Expo change for platform APIs, navigation, and lifecycle safety. Use for a React Native code review. Do not use for a web-only React bug or an Expo upgrade nobody asked for."

React Native Code Review

Shared skill react-native-code-review for React Native. Codex and Antigravity CLI both read .agents/skills/react-native-code-review/SKILL.md. Install this one file. Codex calls it with $react-native-code-review or /skills. Cursor, GitHub Copilot, and Devin also read .agents/skills. A second copy with the same name collides. Do not install another tool's file there.

When to run

Run this review for a React Native or Expo change: a screen, a native permission, navigation, or "will this work on both phones".

When to leave it

Stop for a web-only React repository. Do not upgrade Expo or React Native unless the user asked.

Steps

1. Read the React Native version and the Expo SDK field if it exists. A mismatch with the SDK's supported React Native is a high finding. A DOM API (window, document, View Transitions) in native code is a high finding.
2. Review effects that set state after await. A stack screen that is covered is still mounted. Work that should stop on blur belongs in useFocusEffect, not in an unmount cleanup. Missing that split is a finding.
3. Review navigation params and cache keys. A secret in a param, or a private cache with no user id, is a high finding.
4. Review permissions. A request with no denial path is a finding.
5. Note how the change was run: device, simulator, or not run. Do not call a release ready from a simulator-only look.

Output

Findings with severity, file, line, the platform assumption, and a fix. Include the versions you read.

## Findings
- high | path:line | what is wrong | what to do instead

## No findings
Say "No findings" and list the files you opened. A search you skipped is not a clean result.

## Example
Input: `HomeScreen.tsx` does `const user = await load()` and then `setUser(user)` with no cancelled flag. The screen can unmount first.
Output: high | HomeScreen.tsx:22 | the state update runs after unmount | return when the effect cleanup has run. A one-time load in `useEffect` that checks that flag is not a finding.

Do not flag the valid pattern named in the standards. Flag the mistake next to it.

Stop

Stop when the user asks you to ship a store build or to change the SDK as part of a review.

Severity

high is a bug a caller can hit or a secret that leaks. medium is a broken contract that needs a specific input. low is a cleanup that does not change behaviour.