What problem does it solve? React codebases accumulate subtle bugs, performance regressions, and accessibility gaps that static scanners list without judgment. This Skill turns React Doctor's raw scan output into a senior-engineer audit: it decides which findings actually matter, prioritizes them by user impact, and writes plans precise enough for a cheaper model to execute without context. ## Core Features & Use Cases - Evidence-based audit: Runs React Doctor read-only as JSON, then vets every finding at its file:line against five categories — bugs, performance, accessibility, security, and maintainability. - Leverage-driven prioritization: Severity is judged by hot-path impact (per-keystroke components, provider fan-out, untrusted-input sinks), not the rule's raw severity. - Self-contained plans: Each plan inlines the exact current code, the canonical fix recipe from react.doctor, repo conventions, scope boundaries, and mechanical plus behavioral verification steps. - Use Case: Ask it to audit a Next.js app before a performance push; it scans, confirms the unstable context value re-rendering the whole tree, and writes a plan another agent executes in an isolated worktree. ## Quick Start Ask the agent to audit this React codebase and produce prioritized improvement plans using the improve-react skill.
