This example demonstrates how to use the Spring AI Model Context Protocol (MCP) with the Brave Search MCP Server. The application enables natural language interactions with Brave Search, allowing you to perform internet searches through a conversational interface.

Prerequisites

Java 17 or higher
Maven 3.6+
npx package manager
OpenAI API key
Brave Search API key

Setup

Install npx (Node Package eXecute): First, make sure to install npm and then run:
```
npm install -g npx
```

Clone the repository:

git clone https://github.com/spring-projects/spring-ai-examples.git
cd model-context-protocol/brave-starter

Set up your API keys:

export OPENAI_API_KEY='your-openai-api-key-here'
export BRAVE_API_KEY='your-brave-api-key-here'

Build the application:
```
./mvnw clean install
```

Running the Application

Run the application using Maven:

./mvnw spring-boot:run

The application will demonstrate the integration by asking a sample question about Spring AI and Model Context Protocol, utilizing Brave Search to gather information.

How it Works

The application integrates Spring AI with the Brave Search MCP server through several components:

MCP Client Autoconfiguration

Add the Spring AI MCP boot-starter to your POM:

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

Application properties:

spring.ai.mcp.client.stdio.enabled=true
spring.ai.mcp.client.stdio.servers-configuration=classpath:/mcp-servers-config.json

The servers-configuration should point to Anthropic Desktop-compatible JSON configuration like:

{
  "mcpServers": {
    "brave-search": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-brave-search"
      ],
      "env": {
      }
    }
  }
}

The MCP auto-configuration will use the servers-configuration to initialize MCP client connection to the provided servers and create an McpToolAdapter bean, providing access to the MCP tools exposed by the configured MCP servers.

Chat Integration

The ChatClient is configured with the Brave Search function callbacks:

var chatClient = chatClientBuilder
        .defaultTools(mcpToolAdapter.toolCallbacks())
        .build();

This setup allows the AI model to do the following:

Understand when to use Brave Search
Format queries appropriately
Process and incorporate search results into responses