SpringTutorials
    Spring AI Tutorials
    Tutorial 10

    Exploring Model Context Protocol (MCP) With Spring AI

    Build interoperable AI tools and services using Anthropic's open standard for connecting LLMs to external data and capabilities

    1
    What is the Model Context Protocol?

    The Model Context Protocol (MCP) is an open standard developed by Anthropic that provides a universal way to connect AI models to external data sources, tools, and services. Think of it as a "USB-C for AI" — a standardized interface that works across different AI applications.

    Standardized

    One protocol works across all MCP-compatible clients and servers

    Client-Server

    MCP Clients connect to MCP Servers that expose tools and resources

    Extensible

    Add new capabilities without modifying the core AI application

    MCP vs Direct Tool Calling

    With direct tools, each application must implement its own integrations. With MCP, you build an integration once and use it everywhere — in Claude Desktop, your Spring AI app, or any MCP-compatible client.

    2
    MCP Architecture

    MCP Communication Flow

    AI Application

    (Host)

    →

    MCP Client

    (Spring AI)

    ↔

    MCP Server

    (Tools/Resources)

    Key Concepts

    MCP Server

    Exposes tools (functions the AI can call), resources (data it can read), and prompts (templates for interactions).

    MCP Client

    Connects to MCP servers, discovers available capabilities, and invokes them on behalf of the AI model.

    Transport Layer

    MCP supports multiple transports: stdio (standard I/O for local processes) and SSE (Server-Sent Events over HTTP).

    3
    Using MCP Client in Spring AI

    Step 1: Add Dependencies

    Xml Example
    <dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-mcp-client-spring-boot-starter</artifactId></dependency><dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-openai-spring-boot-starter</artifactId></dependency>

    Step 2: Configure MCP Servers

    Configure MCP servers in application.yml:

    Yaml Example
    spring:ai:mcp:client:stdio:# File system server - provides file operationsfilesystem:command: npx
args:-"-y"-"@modelcontextprotocol/server-filesystem"-"/path/to/allowed/directory"# Brave search server - web search capabilitiesbrave-search:command: npx
args:-"-y"-"@anthropic/mcp-server-brave-search"env:BRAVE_API_KEY: ${BRAVE_API_KEY}

    Step 3: Use MCP Tools with ChatClient

    Java Example
    importorg.springframework.ai.chat.client.ChatClient;importorg.springframework.ai.mcp.SyncMcpToolCallbackProvider;importorg.springframework.stereotype.Service;@ServicepublicclassMCPAgentService{privatefinalChatClient chatClient;publicMCPAgentService(ChatClient.Builder chatClientBuilder,SyncMcpToolCallbackProvider mcpToolProvider){// Register all MCP tools with the ChatClientthis.chatClient = chatClientBuilder
.defaultSystem("""
You are a helpful assistant with access to file system 
operations and web search capabilities.
Use the available tools to help users with their requests.
""").defaultTools(mcpToolProvider)// All MCP server tools available.build();}publicStringprocess(String userMessage){return chatClient.prompt().user(userMessage).call().content();}}

    Auto-Discovery

    Spring AI automatically discovers all tools exposed by configured MCP servers and makes them available to your ChatClient.

    4
    Building Your Own MCP Server

    Spring AI also lets you create MCP servers to expose your Spring services as tools:

    Step 1: Add Server Dependency

    Xml Example
    <dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-mcp-server-spring-boot-starter</artifactId></dependency>

    Step 2: Define Tools as Spring Beans

    Java Example
    importorg.springframework.ai.tool.annotation.Tool;importorg.springframework.ai.tool.annotation.ToolParam;importorg.springframework.stereotype.Component;@ComponentpublicclassWeatherService{@Tool(description ="Get current weather for a city")publicWeatherResponsegetCurrentWeather(@ToolParam(description ="City name")String city,@ToolParam(description ="Temperature unit: celsius or fahrenheit")String unit){// Implement weather API callreturnnewWeatherResponse(city,22.5, unit,"Sunny");}@Tool(description ="Get 5-day weather forecast")publicList<ForecastDay>getWeatherForecast(@ToolParam(description ="City name")String city){// Return forecast datareturnList.of(newForecastDay("Monday",20,"Cloudy"),newForecastDay("Tuesday",23,"Sunny"));}}recordWeatherResponse(String city,double temperature,String unit,String condition){}recordForecastDay(String day,int tempHigh,String condition){}

    Step 3: Configure Server Transport

    Yaml Example
    spring:ai:mcp:server:# Use Server-Sent Events for HTTP-based communicationtransport: sse
sse-endpoint: /mcp/sse
# Or use stdio for local process communication# transport: stdio

    Step 4: Run and Connect

    Your MCP server is now available at http://localhost:8080/mcp/sse and can be connected to by any MCP client, including Claude Desktop or another Spring AI application.

    5
    Best Practices & Use Cases

    Common MCP Server Types

    Database Access

    Query PostgreSQL, MySQL, or MongoDB and let AI analyze results

    API Integrations

    Wrap REST APIs (Slack, GitHub, Jira) as MCP tools

    Code Execution

    Run Python scripts, execute shell commands safely

    Business Logic

    Expose domain-specific operations as AI-callable tools

    Security Considerations

    Validate all inputs before executing operations
    Use authentication for production MCP servers (OAuth2 supported)
    Limit file system access to specific directories
    Implement rate limiting for public-facing servers

    Pre-built MCP Servers

    Check out github.com/modelcontextprotocol/servers for ready-to-use servers for GitHub, Slack, PostgreSQL, Puppeteer, and more.

    What You've Learned

    MCP Fundamentals

    Standardized AI-tool connectivity

    Architecture

    Clients, servers, and transports

    MCP Client

    Connect to external MCP servers

    MCP Server

    Build your own tool servers

    Use Cases

    Database, APIs, and business logic

    Security

    Authentication and validation

    💬 Comments & Discussion