Curupira - MCP Frontend Debugging Tool

Overview

Curupira is a Model Context Protocol (MCP) debugging tool that provides AI assistants with direct access to browser DevTools, React state, and runtime debugging capabilities. It bridges the gap between AI assistants and complex frontend applications through Chrome DevTools Protocol.

Key Features

• 🔍 Chrome DevTools Protocol Integration: Full access to CDP domains (Runtime, Debugger, Network, DOM, etc.)
• ⚛️ React Application Debugging: Component inspection, props/state analysis, hooks debugging
• 🎯 State Management: Apollo GraphQL, Zustand stores, XState machines, TanStack Query
• 🛠️ Framework Support: React Router, Framer Motion, React Hook Form, Vite, Panda CSS
• 🚀 Multiple Transports: WebSocket, HTTP/SSE, and stdio for different deployment scenarios
• 📦 Dependency Injection: Clean architecture with comprehensive DI container
• 🤖 AI-Optimized: Built specifically for AI assistant interaction via MCP

📚 Documentation

• API Documentation - Complete API reference for all resources, tools, and prompts
• CLI Reference - Command-line interface documentation
• Development Guide - Development setup and architecture
• Usage Examples - Common debugging scenarios and integration examples
• Troubleshooting Guide - Solutions for common issues
• Architecture - System architecture and design

Quick Start

Using Docker with Browserless

# Start Browserless Chrome service
docker run -d -p 3000:3000 --name browserless browserless/chrome

# Start Curupira MCP server
docker run -p 8080:8080 -e CHROME_SERVICE_URL=http://host.docker.internal:3000 drzln/curupira:latest

Using npm

# Install globally
npm install -g curupira

# Start with default configuration
curupira start

# Or specify a custom config
curupira start --config ./config/production.yaml

From Source

# Clone the repository
git clone https://github.com/pleme-io/nexus.git
cd nexus/pkgs/services/typescript/curupira

# Install dependencies
npm install

# Build the project
npm run build

# Start the MCP server
npm run start

# Or run in development mode with hot reload
npm run dev

Claude Code Integration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "curupira": {
      "command": "node",
      "args": ["/path/to/curupira/mcp-server/dist/main.js"],
      "env": {
        "CHROME_SERVICE_URL": "http://localhost:3000",
        "LOGGING_LEVEL": "info"
      }
    }
  }
}

Architecture

Curupira follows a clean architecture with dependency injection:

graph TD
    A[Chrome Browser] -->|CDP WebSocket| B[Browserless Service]
    B -->|CDP Protocol| C[Chrome Service Layer]
    C -->|Events| D[Tool/Resource Providers]
    D -->|MCP Protocol| E[MCP Server]
    E -->|stdio/HTTP/WS| F[Claude/AI Assistant]
    
    G[DI Container] -.->|Injects| C
    G -.->|Injects| D
    G -.->|Injects| E

Architectural Layers

1. Foundation Layer - Branded types, interfaces, error definitions
2. Core Services - Chrome service, buffer services, validators
3. MCP Layer - Tool/resource providers, protocol handlers
4. Application Layer - Server bootstrap, DI container, transport management

Key Components

• Chrome Service: Manages CDP connections with event-driven architecture
• Tool Registry: Dynamic tool registration with factory pattern
• Resource Registry: Provides browser state as MCP resources
• DI Container: Manages dependencies and service lifecycle

Installation

Prerequisites

• Node.js 20+ LTS
• npm 10+
• Chrome/Chromium browser
• Docker (optional)

Chrome Extension Setup

1. Clone the repository
2. Build the extension: npm run build:extension
3. Open Chrome and navigate to chrome://extensions/
4. Enable "Developer mode"
5. Click "Load unpacked" and select chrome-extension/dist

Server Setup

# Using npm scripts
npm run dev:server

# Using Docker
docker-compose up -d

# Using Kubernetes
kubectl apply -f k8s/

Usage Examples

Debug Lazy-Loaded Icons

// AI Assistant can request:
await mcp.runTool('inspect', {
  component: 'Icon',
  props: { name: 'pigeon', lazy: true }
})

Analyze GraphQL Errors

// AI Assistant can request:
await mcp.runTool('analyzeGraphQL', {
  operation: 'GetProducts',
  error: "cannot query field 'products' on type 'Product'"
})

Time Travel Debugging

// AI Assistant can request:
const recording = await mcp.runTool('timeTravel', {
  store: 'cart',
  duration: 30000
})

MCP Resources & Tools

Available Resources

• console - Browser console logs and errors
• network - HTTP requests and responses
• dom - DOM tree and element inspection
• react - React component tree and state
• state - Application state management

Available Tools

• inspect - Inspect components and elements
• evaluate - Execute code in page context
• debug - Set breakpoints and step through code
• profile - Performance profiling
• timeTravel - Record and replay state changes

Configuration

Create a .env file based on .env.example:

# Server Configuration
CURUPIRA_PORT=8080
CURUPIRA_HOST=localhost
CURUPIRA_LOG_LEVEL=debug

# WebSocket Configuration
CURUPIRA_WS_URL=ws://localhost:8080/mcp

# Authentication (optional)
CURUPIRA_AUTH_ENABLED=false
CURUPIRA_JWT_SECRET=your-secret-here

Docker Support

Build Image

docker build -t curupira:latest .

Run Container

docker run -p 8080:8080 \
  -e CURUPIRA_LOG_LEVEL=info \
  curupira:latest

Docker Compose

docker-compose up -d

Development

Project Structure

curupira/
├── chrome-extension/   # Browser extension
├── mcp-server/        # MCP protocol server
├── shared/            # Shared utilities
├── docs/              # Documentation
├── k8s/               # Kubernetes configs
└── scripts/           # Build scripts

NPM Scripts

# Development
npm run dev              # Start all services
npm run dev:server       # Start MCP server only
npm run dev:extension    # Build extension (watch)

# Building
npm run build           # Build all packages
npm run build:docker    # Build Docker image

# Testing
npm run test           # Run all tests
npm run test:watch     # Run tests (watch)
npm run test:coverage  # Generate coverage

# Code Quality
npm run lint          # Lint and fix
npm run format        # Format code
npm run type-check    # Check types

Testing

# Unit tests
npm run test:unit

# Integration tests
npm run test:integration

# E2E tests
npm run test:e2e

Contributing

We welcome contributions! Please see our Contributing Guide for details.

Development Setup

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Security

For security issues, please email security@pleme.io instead of using the issue tracker.

Publishing

For Maintainers

To release a new version:

# Bump version (patch, minor, or major)
./scripts/version.sh patch

# Push changes and tag
git push origin main
git push origin v1.0.1

# Publish to npm and Docker Hub
./scripts/publish.sh

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Built on the Model Context Protocol specification
• Inspired by Chrome DevTools and React DevTools
• Named after the Curupira from Brazilian folklore

Links

• Documentation
• npm Package
• Docker Hub
• Issue Tracker