spring-ai-mcp-client

What is spring-ai-mcp-client?

spring-ai-mcp-client is a Java-based Model Context Protocol (MCP) client built with Spring AI. It connects to MCP protocol–supported servers and enables AI-powered chat interactions using Anthropic Claude or OpenAI models. It handles protocol version negotiation, capability negotiation, tool discovery and execution, resource access, prompt system interactions, and optional features like roots management and sampling.

How to use spring-ai-mcp-client?

Clone the repository, configure MCP server details in an mcp-servers.json file, set your Anthropic or OpenAI API key, and provide Spring Boot configuration in application.yml. Then run mvn spring-boot:run and access the chat endpoint at http://localhost:8081/ai with a GET request containing a message parameter.

Key features of spring-ai-mcp-client

• Client-side MCP protocol implementation including version negotiation
• Supports Anthropic Claude and OpenAI models
• Manages communication via JSON‑RPC over stdio transport
• Enables tool discovery, execution, resource access, and prompt interactions
• Optional support for roots management and sampling
• Configurable request timeout, server name, and sync type

Use cases of spring-ai-mcp-client

• Building AI‑powered chatbots that leverage external MCP‑compatible servers
• Integrating model context tools and resources into Spring‑based applications
• Creating a unified gateway to query multiple MCP servers with a single API call
• Prototyping conversational AI features with either Anthropic or OpenAI models

FAQ from spring-ai-mcp-client

What AI models does spring-ai-mcp-client support?

It supports both Anthropic Claude models and OpenAI models. You configure the API key in application.yml for either provider.

What are the runtime requirements?

Java 17 or later, Maven or Gradle for building, and optionally Docker. You also need either an Anthropic API key or an OpenAI API key.

How do I configure MCP servers?

Create an mcp-servers.json file (placed on the classpath) listing each server’s command and arguments. Then reference that file in spring.ai.mcp.client.stdio.servers-configuration in your application.yml.

What transport and protocol does spring-ai-mcp-client use?

It uses stdio transport with JSON‑RPC message format. The client negotiates protocol version and capabilities with the server automatically.

Where is the chat endpoint and how do I invoke it?

The application runs on port 8081 by default (configurable via server.port). Send a GET request to http://localhost:8081/ai?message=YourQuestion to start a chat.