MCP Employee API Server

A Model Context Protocol (MCP) server that provides tools for managing employee data through a REST API. This server exposes employee management operations as MCP tools that can be used by AI assistants and other MCP clients.

Features

• Employee Management: Full CRUD operations for employee data

• REST API Integration: Connects to a local employee API server

• MCP Protocol: Exposes functionality through the Model Context Protocol

• Async Operations: Built with async/await for optimal performance

• Error Handling: Robust error handling for API requests

Available Tools

The server provides the following MCP tools:

• get_employees() - Retrieve all employees

• get_employee(id) - Get a specific employee by ID

• add_employee(name, age) - Create a new employee

• update_employee(id, name, age) - Update an existing employee

• delete_employee(id) - Delete an employee by ID

Prerequisites

• Python 3.13 or higher

• A running employee API server at http://localhost:8000

Installation

1. Clone the repository:

   git clone https://github.com/JoseGarayar/mcp_test.git
   cd mcp_test

2. Clone the api employee repository:

   git clone https://github.com/JoseGarayar/api_employees.git

3. Install dependencies using uv:

   uv sync

Usage

Running the MCP Server

Start the MCP server using stdio transport:

uv run python main.py

The server will run and listen for MCP protocol messages via stdin/stdout.

API Configuration

The server is configured to connect to a local API server at http://localhost:8000. You can modify the URL_BASE constant in main.py to point to a different API endpoint.

Example API Endpoints

The server expects the following API endpoints to be available:

• GET /employees - List all employees

• GET /employees/{id} - Get employee by ID

• POST /employees - Create new employee

• PUT /employees/{id} - Update employee

• DELETE /employees/{id} - Delete employee

Development

Project Structure

mcp_test/
├── main.py          # Main MCP server implementation
├── pyproject.toml   # Project configuration and dependencies
├── README.md        # This file
└── uv.lock         # Lock file for dependencies

Dependencies

• httpx - Async HTTP client for API requests

• mcp[cli] - Model Context Protocol implementation

Development Dependencies

• ruff - Python linter and formatter

Error Handling

The server includes comprehensive error handling:

• Network timeouts (30 seconds)

• HTTP error status codes

• Invalid HTTP methods

• Connection failures

All errors are gracefully handled and return None for failed operations.

License

This project is part of a test implementation for MCP server development.