README — MCP Server and Client with Spring AI & Mistral

🧠 Overview

This repository provides a minimal working example of implementing the Model Context Protocol (MCP) using Spring AI.
It demonstrates how to integrate an MCP server exposing tools over SSE/WebMVC and an MCP client orchestrating a Mistral LLM to call those tools dynamically.

Goal: Quickly understand, configure, test, and deploy a simple yet complete MCP setup using Spring Boot and Mistral.

⚙️ Prerequisites

Before running the project, make sure you have:

• Java 25
• Maven 3.8+

🚀 Build & Run

You’ll need to start the MCP Server, then the MCP Client, and finally the frontend.

Run the following commands for each module (server, client, frontend):

cd project-name
mvn clean install -DskipTests
mvn spring-boot:run

Default Ports & Endpoints

Component	Default Port	Description
MCP Server	8082	Exposes MCP tools and SSE stream on /mcp/sse
MCP Client	8080	Orchestrates the LLM and exposes an HTTP endpoint for user queries
Frontend	8081	Provides a simple web UI to interact with the MCP setup

🧩 Architecture Overview

Frontend (8081)
   │
   ▼
MCP Client (8080) ──> Mistral LLM
   │
   ▼
MCP Server (8082)

• MCP Server: Defines and exposes domain-specific tools (via Spring MVC + SSE).
• MCP Client: Connects to the server, processes user input, and delegates reasoning and tool usage to Mistral.
• Frontend: Offers a basic interface to send queries and visualize responses.

🧰 Useful Links

• Spring AI Documentation
• Mistral AI
• Model Context Protocol (MCP)