Chat Assistant with MCP Client

Project Overview

This project implements a Model Context Protocol (MCP) client using Spring Boot and the Spring AI MCP Client starter. The client connects to an MCP server to interact with various LLM models and provides a chat interface for user interactions.

Demo

!img.png

Overview

The MCP client is built using:

• Java 21
• Spring Boot 3.5.0
• Spring AI MCP Client Starter 1.0.0: Documentation

Prerequisites

• Java 21 or higher
• Gradle 8.x or higher
• Access to an MCP Server instance
• Docker (optional, for containerized deployment)

Project Structure

src/
├── main/
│   ├── java/        # Java source files
│   └── resources/   # Configuration files
└── test/           # Test files

Dependencies

• spring-ai-starter-mcp-client: Core MCP client implementation
• stater-web: Web framework for building RESTful APIs
• spring-boot-starter-test: Testing framework
• starter-thymeleaf: Template engine for rendering chat HTML view

Configuration

The client can be configured through application.properties or application.yml files in the src/main/resources directory to connect to the MCP server.

Implementation Steps

1. Create an MCP Client to connect to the any Java SSE MCP Server like our https://github.com/Ntsal/ntsal-ecom-java-mcp-server/tree/sse

2. Add ChatClient with an paid LLM Model like Anthropic Claude, Google Gemini, OpenAI GPT to test its tool calling with MCP

3. Facilitate an Interactive Chat Interface

4. Try with free Ollama models that support tool calling like llama3.2:3b, qwen3:8b,...