React Analyzer MCP

React Analyzer MCP does one thing: it walks your React codebase, extracts every component and prop definition it can find, and hands that structure back to your LLM in a single pass. No API calls, no cloud dependencies. It's local analysis for local codebases. I'd reach for this when I'm onboarding to a large React project or planning a migration where I need to understand prop usage across dozens of components. Instead of asking Claude to grep through files one at a time, you get a structured snapshot of the entire component tree. It generates llm.txt files and documentation that Claude can actually parse without hallucinating prop names or missing optional fields. The workflow is straightforward: point it at your React project, let it scan, then ask Claude questions like "which components accept a theme prop" or "show me all components that take children but don't validate the type". The answers come back fast because the analysis already happened. It's particularly useful for prop audits before a TypeScript migration or when you're trying to enforce consistent prop patterns across a team. Quirks: it's a build-it-yourself setup. You clone the repo, run npm install, build it, then wire it into your Claude Desktop config manually. There's no published npm package, so updates mean pulling from main and rebuilding. The README is minimal. If you're not comfortable editing JSON config files or don't know where your Claude Desktop config lives, you'll spend time Googling. It also won't catch dynamic prop spreading or runtime prop injection. If your codebase uses a lot of {...rest} patterns or composes props through higher-order components, the output will be incomplete. It's a static analyser, so it sees what TypeScript or PropTypes declare, not what actually runs. Who shouldn't bother: if your React project is small enough to hold in your head, this is overkill. If you're not already using Claude Desktop for development, the setup friction isn't worth it for a one-off question. But for teams working on sprawling React apps who want to ask structural questions without waiting for manual audits, it's a solid addition to the toolkit.