Spring AI MCP Server and Client

A demonstration project showcasing the integration of Spring AI's Model Completion Provider (MCP) protocol for building AI-powered applications. This project consists of a server component that provides AI model completion services with function-calling capabilities, and a client component that consumes these services.

Overview

This project demonstrates how to use Spring AI's MCP protocol to create a server that exposes AI model completion services and a client that consumes these services. The server component provides mathematical tools (sum and multiply) that can be called by AI models, while the client component exposes a REST endpoint for users to submit questions.

Features

• MCP Server: Provides AI model completion services with function-calling capabilities
• MCP Client: Consumes AI model completion services from the server
• Tool Integration: Demonstrates how to expose Java methods as callable tools for AI models
• REST API: Exposes a REST endpoint for users to submit questions
• Server-Sent Events (SSE): Uses SSE for communication between server and client

Prerequisites

• Java 17 or higher
• Maven 3.6 or higher
• OpenAI API key (for the client component)

Installation and Setup

1. Clone the repository:

   git clone https://github.com/stylepatrick/spring-ai-mcp-server-client.git
   cd spring-ai-mcp-server-client
   

2. Build the server component:

   cd mcp-server
   mvn clean install
   

3. Build the client component:

   cd ../mcp-client
   mvn clean install
   

4. Update the OpenAI API key in mcp-client/src/main/resources/application.yml with your own API key.

Usage

1. Start the server component:

   cd mcp-server
   mvn spring-boot:run
   

2. Start the client component:

   cd mcp-client
   mvn spring-boot:run
   

3. Send a question to the client component:

   curl -X POST "http://localhost:8081/question?question=What%20is%202%20plus%203?"
   

Configuration

Server Configuration

The server component is configured in mcp-server/src/main/resources/application.yaml:

spring:
  application:
    name: mcp-server
  ai:
    mcp:
      server:
        sse-endpoint: /api/v1/sse
        sse-message-endpoint: /api/v1/mcp
        type: sync

logging:
  level:
    io.modelcontextprotocol: TRACE
    org.springframework.ai.mcp: TRACE

Client Configuration

The client component is configured in mcp-client/src/main/resources/application.yml:

spring:
  application:
    name: mcp-client
  ai:
    openai:
      api-key: your-api-key
    mcp:
      client:
        toolcallback:
          enabled: true
        sse:
          connections:
            server1:
              url: http://localhost:8080
              sse-endpoint: /api/v1/sse

server:
  port: 8081

Project Structure

• mcp-server: Server component that provides AI model completion services

  • McpServerApplication.java: Main application class
  • tools/MathTools.java: Mathematical tools that can be called by AI models
  • tools/ToolConfig.java: Configuration for tool callback provider

• mcp-client: Client component that consumes AI model completion services

  • McpClientApplication.java: Main application class
  • resource/MyController.java: REST controller for handling user questions
  • service/MyService.java: Service for processing questions using ChatClient

License

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

Copyright (c) 2025 stylepatrick