CDS GRC MCP Server

A Model Context Protocol (MCP) server that simulates e-commerce marketplace operations for GRC (Governance, Risk, and Compliance) testing and development purposes.

🚀 Features

Seller Management

• GetInfoSeller: Retrieve detailed seller information including ratings, sales statistics, policies, and verification status

Product Management

• ProductList: List products with advanced filtering, sorting, and pagination
• ProductGet: Get detailed product information by GTIN
• ProductCount: Count products matching specified filters
• ProductGetVariants: Retrieve product variants with pagination

Order Management

• OrderList: List orders with filtering by status, sales channel, date range, and more
• OrderGet: Get detailed order information
• OrderAcknowledge: Acknowledge orders (global acceptance, not per line)
• OrderShip: Ship orders with tracking information
• OrderCancel: Cancel orders according to business rules

📋 Requirements

• Node.js 18+
• npm or yarn

🛠️ Installation

npm install @ludovicl33/cds-grc-mcp

🔧 Usage

As a Library

import { 
  getInfoSeller, 
  productList, 
  orderList,
  orderAcknowledge,
  generateSeller
} from '@ludovicl33/cds-grc-mcp';

// Get seller information
const sellerInfo = await getInfoSeller('oauth-token');

// List products
const products = await productList({ 
  limit: 10, 
  sortBy: 'updatedAt' 
});

// List orders
const orders = await orderList({ 
  filters: { status: 'pending' } 
});

// Acknowledge an order
const result = await orderAcknowledge({ orderId: 'ORD-001' });

// Generate a random seller (for testing)
const randomSeller = generateSeller();

As an MCP Server

# Start the MCP server
npm start

# Or run directly
node build/index.js

# Or use the CLI command (after installation)
cds-grc-mcp

As an MCP Server in LLMs

This package is designed to work with LLMs that support the Model Context Protocol (MCP). Here's how to integrate it:

Claude Desktop/Anthropic

1. Install the package globally: npm install -g @ludovicl33/cds-grc-mcp
2. In Claude, use the MCP integration to connect to the server
3. The server will be available as cds-grc-mcp command

Other MCP Clients

# Start the server
cds-grc-mcp

# Or with specific configuration
cds-grc-mcp --config mcp-config.json

MCP Configuration Example

{
  "mcpServers": {
    "cds-grc-mcp": {
      "command": "cds-grc-mcp",
      "args": [],
      "env": {}
    }
  }
}

🎯 MCP Tools Available

| Tool | Description | Parameters | |------|-------------|------------| | GetInfoSeller | Get seller information | oauthToken | | ProductList | List products with filters | filters, cursor, limit, sortBy, sortDir | | ProductGet | Get product by GTIN | gtin | | ProductCount | Count products | filters | | ProductGetVariants | Get product variants | groupReference, cursor, limit | | OrderList | List orders with filters | filters, cursor, limit, sortBy, sortDir | | OrderGet | Get order details | orderId | | OrderAcknowledge | Acknowledge order | orderId | | OrderShip | Ship order | orderId, trackingNumber, carrier | | OrderCancel | Cancel order | orderId, reason | | CustomerComplaintsList | List customer complaints | filters, cursor, limit, sortBy, sortDir | | RespondToComplaint | Respond to customer complaint | complaintId, responseContent, responderType |

🔍 Data Structure

Mock Data Included

• Products: 3 sample products with different visibility rules
• Orders: 4 sample orders in various states (pending, acknowledged, shipped, cancelled)
• Customer Complaints: 4 sample complaints with various statuses, priorities, and categories
• Consistent relationships between products, orders, and complaints

Seller Distribution

• Cdiscount: 50% probability (marketplace owner)
• Seller1, Seller2, Seller3, Seller4, Seller5: 10% probability each (third-party sellers)
• Distribution function: generateSeller() provides realistic seller assignment

Business Rules

• Field visibility: Products have different field access based on createdBySeller flag
• Order workflow: Strict state transitions (pending → acknowledged → shipped)
• Complaint management: Seller responses automatically close complaints, client responses keep them open
• Validation: Comprehensive input validation with meaningful error messages

Customer Complaints Structure

Each complaint includes:

• Basic info: ID, discussion ID, order details, product information
• Status: ouvert (open) or fermé (closed)
• Priority: basse, moyenne, haute, or critique
• Category: livraison, qualité, facturation, or autre
• Origin: vendeur (seller) or client (customer)
• Platform: Cdiscount, Amazon, Fnac, or Rakuten
• Timestamps: Creation, update, and closure dates

🧪 Testing

# Run all tests
npm test

# Run tests in watch mode
npm run test:watch

# Build and test
npm run build && npm test

📦 Build

# Build the project
npm run build

# Build and start
npm run dev

🔐 Authentication

This is a mock/simulation server. All authentication is simulated and returns consistent test data. No real API keys or credentials are required.

📄 License

MIT License - see LICENSE file for details.

👨‍💻 Author

Ludovic Lefebvre - @ludovicl33

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📝 Changelog

1.4.0

• Added RespondToComplaint MCP tool for responding to customer complaints
• Automatic complaint closure when sellers respond
• Enhanced complaint management workflow
• Comprehensive test coverage for complaint responses

1.3.0

• Added customer complaints management system
• New CustomerComplaintsList MCP tool with filtering, sorting, and pagination
• Comprehensive complaint data structure including status, priority, category, and origin
• Mock data for 4 sample complaints with realistic scenarios
• Full test coverage for complaints functionality

1.2.0

• Added seller field to products and orders
• Implemented generateSeller() function with realistic probability distribution
• Cdiscount: 50% probability, other sellers: 10% each
• Enhanced data consistency between products and orders

1.0.0

• Initial release
• Complete MCP server implementation
• Mock e-commerce marketplace operations
• Comprehensive test coverage
• TypeScript support with ES modules
• CLI executable support for LLM integration

🔗 Related Links

• Model Context Protocol (MCP)
• MCP SDK
• CDiscount API Documentation

Note: This package is designed for testing, development, and educational purposes. It provides realistic mock data and behavior for e-commerce operations without requiring actual marketplace integrations.