React Doctor is a TypeScript-based static analysis tool designed to identify code quality issues in React applications before they reach production. Built by Million, the tool performs deterministic scanning of codebases to catch problems spanning state management, effects, performance, architecture, security, and accessibility concerns. The project explicitly targets scenarios where AI agents or developers write suboptimal React code, positioning itself as an automated code review mechanism that can catch these issues systematically.
The tool's scope extends across the entire React ecosystem. According to its documentation, React Doctor works with all major React frameworks and libraries including Next.js, Vite, TanStack, React Native, and Expo, making it broadly applicable rather than framework-specific. This universal compatibility reflects a design philosophy centered on catching common React anti-patterns regardless of the specific tooling or framework choice.
React Doctor offers multiple integration pathways to fit different development workflows. Users can run quick audits at their project root for immediate feedback. For teams using AI coding assistants, the tool functions as an installable skill that agents like Claude Code, Cursor, Codex, and OpenCode can learn from, allowing these systems to understand and avoid the issues React Doctor identifies in future code generation. For continuous integration pipelines, React Doctor integrates with GitHub Actions and provides scaffolding for GitLab CI, scanning pull requests and reporting only newly introduced issues rather than flagging the entire existing codebase backlog. Configuration is handled through a doctor.config.ts file, allowing teams to customize which rules run and how they execute.
The project maintains active development with strong contributor engagement.
React Doctor implements telemetry collection through Sentry to track crashes, basic run traces, and anonymous usage metrics. The collected data includes environment information like CLI version and Node version, invocation context distinguishing between local runs and CI environments, project shape metadata such as framework and React version without file contents, and rule firing statistics showing which checks triggered and their frequency. Users retain full control over this telemetry and can disable it entirely with the --no-telemetry flag.
The project is MIT-licensed and actively welcomes community contributions through its GitHub issues. The tool's positioning as an agent-friendly code review system reflects the growing intersection of AI-assisted development and automated quality assurance, where static analysis tools must integrate seamlessly with both human developers and machine learning-based code generation systems.
