LLM SSE MCP Demo 2025

This project demonstrates the integration between LLM clients and MCP (Model Context Protocol) servers using Server-Sent Events (SSE) for real-time communication. It consists of two Spring Boot applications that showcase tool calling capabilities with various LLM models.

Project Structure

1. SSE MCP Server Demo (sse-mcp-server-demo)

A Spring Boot application that serves as an MCP server, exposing mathematical and date/time tools via SSE endpoints.

Features:

• Math Tools: Addition and multiplication operations
• DateTime Tools: Current time retrieval and alarm setting
• SSE Support: Real-time communication via Server-Sent Events
• MCP Protocol: Implements Model Context Protocol for tool discovery and execution

Available Tools:

• sumNumbers(int, int) - Adds two numbers
• multiplyNumbers(int, int) - Multiplies two numbers
• getCurrentDateTime() - Gets current date/time in user's timezone
• setAlarm(String) - Sets an alarm for specified ISO-8601 time

Port: 8080 (default)

2. LLM MCP Client Demo (llm-mcp-client-demo)

A Spring Boot web application that connects to the MCP server and provides an interactive chat interface with multiple LLM providers.

Features:

• Multiple LLM Support: Anthropic Claude, OpenAI GPT, Google Gemini, Ollama
• Interactive Web Chat: Real-time chat interface using Thymeleaf
• MCP Integration: Connects to SSE MCP server for tool calling
• Tool Discovery: Automatically discovers and uses available tools from MCP server

Supported LLM Models:

• Anthropic Claude: claude-sonnet-4-20250514
• OpenAI: o4-mini-2025-04-16
• Google Gemini: gemini-2.5-flash-preview-05-20
• Ollama: qwen3:8b (local)

Port: 9090

Prerequisites

• Java 17 or higher
• Gradle
• API keys for cloud LLM providers (optional)
• Ollama installed locally (for local models)

Environment Variables

For the LLM client, set the following environment variables for cloud providers:

export ANTHROPIC_API_KEY=your_anthropic_key
export OPENAI_API_KEY=your_openai_key
export GOOGLE_PROJECT_ID=your_google_project_id
export GOOGLE_ZONE=your_google_zone

Quick Start

1. Start the MCP Server

cd sse-mcp-server-demo
./gradlew bootRun

2. Start the LLM Client

cd llm-mcp-client-demo
./gradlew bootRun

3. Access the Chat Interface

Open your browser and navigate to: http://localhost:9090

Usage

1. Start both services in the order specified above
2. Open the web interface at http://localhost:9090
3. Send messages that require mathematical operations or time queries
4. Watch the LLM automatically discover and use the available tools

Example queries:

• "What's 5 times 4 plus 7?"
• "What time is it?"
• "Set an alarm for 2025-01-01T10:00:00"

Technology Stack

• Spring Boot 3.x
• Spring AI - LLM integration framework
• Model Context Protocol (MCP) - Tool discovery and execution
• Server-Sent Events (SSE) - Real-time communication
• Thymeleaf - Web templating
• Gradle - Build system

Configuration

Both applications use YAML configuration files:

• MCP server configuration in sse-mcp-server-demo/src/main/resources/application.yml
• LLM client configuration in llm-mcp-client-demo/src/main/resources/application.yml

Development

The project demonstrates:

• MCP Protocol Implementation using Spring AI
• SSE-based Real-time Communication
• Multi-provider LLM Integration
• Tool Calling and Discovery
• Interactive Web Interfaces

This demo serves as a foundation for building more complex LLM-powered applications with external tool capabilities.