Spring AI Example

What is Spring AI Example?

This project demonstrates implementation patterns and best practices for using Spring AI tools, integrating with the Model Context Protocol (MCP) through a server module (with WebFlux and WebMvc SSE support) and a client module for building AI-powered proposals. It is designed for developers working with Spring AI, MCP, and large language models.

How to use Spring AI Example?

Clone the repository and run the Spring Boot application. Configure the Ollama AI model, PGVector vector store, and H2 database in the application properties. The mcp-server module can be invoked via SSE endpoints, while the proposal-agent module acts as an MCP client.

Key features of Spring AI Example

• Methods as tools via @Tool annotation
• Custom tool result converters
• Tool context injection in methods
• Tool parameter descriptions with @ToolParam
• MCP server with SSE support
• MCP client for AI proposal generation

Use cases of Spring AI Example

• Building chatbots that retrieve current date/time or customer information
• Creating proposal generation systems with tool-based data access
• Demonstrating Spring AI tool patterns in an MCP architecture
• Testing real-time SSE updates from MCP servers

FAQ from Spring AI Example

What is the project structure?

It contains two main modules: mcp-server (MCP server with SSE support) and proposal-agent (MCP client for AI proposals), plus a common source folder.

What AI model and database are used?

The configuration uses the Ollama AI model with PGVector for vector storage and H2 database for development.

What transport does the MCP server use?

It uses Server-Sent Events (SSE) for real‑time communication, implemented with Spring’s SseEmitter.

How are tools defined in Spring AI?

Methods are annotated with @Tool to become tools, optionally using @ToolParam for parameter descriptions and custom result converters via resultConverter.

Does the project include client and server examples?

Yes, the mcp-server module implements the MCP server, while the proposal-agent module implements the MCP client.