Reactotron

Install in VS Code Install in Cursor

An MCP (Model Context Protocol) server for Reactotron, enabling AI assistants to read logs, inspect app state, monitor network requests, and trigger custom commands in your React or React Native app.

How it works

The MCP server runs a WebSocket proxy between your app and the Reactotron desktop app:

App ──► MCP Proxy (port 9091) ──► Reactotron (port 9090)
               │
          captures everything
               │
          MCP tools (Claude/Cursor etc.)

Your app connects to the proxy instead of directly to Reactotron. The proxy forwards all traffic to Reactotron (so the desktop UI works normally) and captures every message for the MCP tools.

The proxy works in standalone mode — if Reactotron is not open, the app connection is kept alive and all messages are still captured. When Reactotron is opened later, the proxy reconnects automatically (with exponential backoff). This means you don't need to have Reactotron running before starting your app.

Requirements

• Your app pointed at the proxy port (default 9091) instead of Reactotron directly
• Reactotron desktop app (optional — for the visual UI)

Installation

Claude Code

claude mcp add reactotron-mcp npx reactotron-mcp

Or with Bun:

claude mcp add reactotron-mcp bunx reactotron-mcp

npx (Node)

{
  "mcpServers": {
    "reactotron": {
      "command": "npx",
      "args": ["-y", "reactotron-mcp"]
    }
  }
}

bunx (Bun)

{
  "mcpServers": {
    "reactotron": {
      "command": "bunx",
      "args": ["reactotron-mcp"]
    }
  }
}

Custom port

If Reactotron is running on a non-default port, set REACTOTRON_PORT:

{
  "mcpServers": {
    "reactotron": {
      "command": "npx",
      "args": ["-y", "reactotron-mcp"],
      "env": {
        "REACTOTRON_PORT": "9090"
      }
    }
  }
}

App setup

Point your app's Reactotron config at the proxy port instead of the default:

// Before
Reactotron.configure({ host: 'localhost' }).connect()

// After
Reactotron.configure({ host: 'localhost', port: 9091 }).connect()

Configuration

Tools

Resources

Resources expose the same data as tools but as readable URIs, useful for attaching live context directly to a conversation.

Prompts

Prompts are reusable templates that guide the AI through common debugging workflows. Invoke them via your AI assistant's prompt/slash-command interface.

Development

bun install
bun run dev       # run from source
bun run build     # bundle for distribution
bun run typecheck # type-check without building