What problem does it solve? React performance reviews often devolve into adding useMemo, useCallback, and React.memo everywhere without evidence. This Skill provides a structured review workflow that identifies real render costs, context invalidation, unstable keys, and unnecessary client JavaScript, while explicitly avoiding memoization recommendations that lack measured justification. ## Core Features & Use Cases - Evidence-Based Review Workflow: Six-step process from identifying the slow path through measurement, bottleneck classification, structural fixes, and before/after validation. - Decision Gates for Memoization: Explicit criteria for when useMemo, useCallback, React.memo, virtualization, and context splitting are justified, including React Compiler considerations. - Severity-Ranked Findings: High/Medium/Low severity rubric with a required output format covering verdict, evidence, findings, do-not-change items, and suggested patches. - Use Case: When reviewing a Next.js pull request where a page feels sluggish, use this Skill to trace whether the cause is a broad 'use client' boundary, a context provider invalidating a large subtree, or an unvirtualized list, and report findings with file locations and validation methods. ## Quick Start Ask the agent to review your React components or current working changes for render performance issues and report findings with severity and evidence without editing code.
