React Native Performance Review

Use this skill to review React Native code for avoidable render work, list inefficiencies, memory leaks, jank risks, and mobile UI performance regressions.

Workflow

1. Inspect the target repository guidance first, especially local AGENTS.md, package manifests, navigation structure, and relevant React Native screens or packages.
2. Read the changed files before scanning. Use rg for quick pattern checks, then confirm each candidate by reading the surrounding code.
3. Check render stability: inline JSX callbacks, inline object/array props, missing React.memo, missing useCallback for handlers passed to children, and missing useMemo for expensive derived values.
4. Check list behavior: stable keys, FlatList/FlashList usage, pagination, item memoization, renderItem stability, and large unvirtualized collections.
5. Check context, state, and effects: broad context values, duplicated or derived state, unbatched updates, listener cleanup, timers, AppState, keyboard, network, and navigation subscriptions.
6. Classify findings by impact and provide concrete fixes. Implement fixes only when the user asks for code changes or the task is explicitly implementation-oriented.

Review Focus

• Inline functions or inline objects in hot render paths that force child re-renders.
• Large lists without virtualization, pagination, stable keys, or memoized item components.
• Context providers carrying large or frequently changing values across broad trees.
• Effects that subscribe to AppState, keyboard, network, timers, or navigation without cleanup.
• Derived state stored redundantly instead of computed from source state.
• Deep wrapper-heavy component trees where flattening would reduce layout and reconciliation cost.
• Image or asset optimization only when size, memory, bundle impact, or repeated rendering is a known concern.

Output

• Lead with findings ordered by severity, with file and line references when possible.
• Separate blocking issues from high, medium, and low priority recommendations.
• Explain impact in mobile-user terms: security exposure, broken rollout, jank, inaccessible flow, analytics drift, build/runtime failure, or maintainability risk.
• Include specific remediation guidance. Keep code snippets small unless the user asks for an implementation.
• If no issues are found, say so clearly and note any meaningful test or verification gaps.

References

• references/review-guide.md: Migrated detailed review guide from the original mobile-ai-skills agent definition. Read it when you need the full checklist, example report shape, or grep patterns.