Test MCP Server

A Model Context Protocol (MCP) server for managing user data with support for resources, tools, and prompts.

Features

Resources

The server provides two resource endpoints for accessing user data:

• All Users (users://all) - Retrieves all users from the database

• User Profile (users://{userId}/profile) - Retrieves a specific user's profile by ID

Tools

Three tools are available for user management:

1. create-user - Create a new user with specified details

   • Parameters: name, email, address, phone

2. create-random-user - Automatically generate and create a user with fake data

   • Uses AI sampling to generate realistic user information

3. generate-fake-user (Prompt) - Generate fake user data based on a given name

   • Parameter: name

Installation

npm install

Requirements

• Node.js (with ES modules support)

• Dependencies:

  • @modelcontextprotocol/sdk

  • zod

Usage

Starting the Server

npm start

The server uses stdio transport for communication.

Resource Access

Get all users:

users://all

Get specific user profile:

users://123/profile

Tool Usage

Create a user:

{
  "name": "John Doe",
  "email": "john@example.com",
  "address": "123 Main St",
  "phone": "555-0123"
}

Create a random user: No parameters required - automatically generates fake user data using AI sampling.

Data Storage

User data is stored in ./src/data/users.json as a JSON array. Each user object contains:

• id (number) - Auto-incremented user ID

• name (string) - User's full name

• email (string) - Email address

• address (string) - Physical address

• phone (string) - Phone number

Server Capabilities

• Resources: Query user data via URI schemes

• Tools: Perform user management operations

• Prompts: Generate templated prompts for user creation

Development

The server is built using the Model Context Protocol SDK and implements:

• Resource templates with dynamic URI parameters

• Tool definitions with Zod schema validation

• AI sampling for generating fake data

• File-based persistence

License

MIT

Notes

• User IDs are auto-incremented starting from the current user count + 1

• All operations return JSON responses

• Error handling is implemented for user not found scenarios

• The server uses the sampling API to generate realistic fake user data