react-expert

The react-expert skill provides deep, evidence-based technical research into the React library. Designed for core contributors, library authors, and advanced engineers, this tool bypasses potential inaccuracies in general LLM training data by grounding all answers directly in the official React repository. It is the primary utility for resolving ambiguity regarding React internals, experimental features, or conflicting API documentation patterns.

• Performs parallelized research across six distinct domains: unit tests, source code implementation, git commit history, PR discussions, GitHub issues, and type definitions (Flow/TypeScript).

• Automatically maintains a local clone of the Facebook/React repository to ensure research is performed against the most current or specific commit hashes.

• Extracts concrete usage examples directly from test suites, ensuring that all code snippets are verified, functional, and representative of canonical patterns.

• Identifies triggers for console warnings, error messages, and deprecation notices by parsing core implementation files.

• Reconstructs design rationale and historical context by traversing commit logs and GitHub PR threads.

• Provides a skeptical, evidence-first framework that prioritizes source-level findings over conventional knowledge or secondary documentation.

• To use this skill, invoke it via the command line followed by the React API or topic (e.g., /react-expert useTransition).

• The tool expects no prior assumptions; it will highlight discrepancies between current documentation and actual framework behavior discovered in source files.

• Inputs should be specific React hooks, components, or conceptual topics; outputs consist of comprehensive, source-linked reports formatted as structured documentation.

• Constraints include a strict ban on external web searches; all information must be derived from the local React repo via standard git and gh CLI interactions.

• This tool is essential for teams standardizing on React, developers debugging framework-level behavior, or authors creating documentation that requires absolute accuracy regarding implementation details and edge-case error handling.