[9]. SpringAI Alibaba MCP

大纲

模型上下文协议

function calling 每次都要重复开发工具，MCP相当于把这些工具共用

入门

大模型和大模型之间的对话

MCP架构

通信方式

两者对比

Server

一个配置即可：

java 复制代码

package com.atguigu.study.config;

import com.atguigu.study.service.WeatherService;
import org.springframework.ai.tool.ToolCallbackProvider;
import org.springframework.ai.tool.method.MethodToolCallbackProvider;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

@Configuration
public class McpServerConfig {
    /**
     * 将工具方法暴露给外部 mcp client 调用
     *
     * @param weatherService
     * @return
     */
    @Bean
    public ToolCallbackProvider weatherTools(WeatherService weatherService) {
        return MethodToolCallbackProvider.builder()
                .toolObjects(weatherService)
                .build();
    }
}

相应的tool

java 复制代码

package com.atguigu.study.service;

import org.springframework.ai.tool.annotation.Tool;
import org.springframework.stereotype.Service;

import java.util.Map;



@Service
public class WeatherService {
    @Tool(description = "根据城市名称获取天气预报")
    public String getWeatherByCity(String city) {
        Map<String, String> map = Map.of(
                "北京", "11111降雨频繁，其中今天和后天雨势较强，部分地区有暴雨并伴强对流天气，需注意",
                "上海", "22222多云,15℃~27℃,南风3级，当前温度27℃。",
                "深圳", "333333多云40天，阴16天，雨30天，晴3天"
        );
        return map.getOrDefault(city, "抱歉：未查询到对应城市！");
    }
}

POM server启动的是netty服务器

xml 复制代码

 <!--注意事项（重要）
            spring-ai-starter-mcp-server-webflux不能和<artifactId>spring-boot-starter-web</artifactId>依赖并存，
            否则会使用tomcat启动,而不是netty启动，从而导致mcpserver启动失败，但程序运行是正常的，mcp客户端连接不上。
        -->
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter</artifactId>
        </dependency>
        <!--mcp-server-webflux-->
        <dependency>
            <groupId>org.springframework.ai</groupId>
            <artifactId>spring-ai-starter-mcp-server-webflux</artifactId>
        </dependency>

properties配置

properties 复制代码

# ====mcp-server Config=============
spring.ai.mcp.server.type=async
spring.ai.mcp.server.name=customer-define-mcp-server
spring.ai.mcp.server.version=1.0.0

Client

Map类型的yaml文件配置

properties 复制代码

# ====mcp-client Config=============
spring.ai.mcp.client.type=async
spring.ai.mcp.client.request-timeout=60s
spring.ai.mcp.client.toolcallback.enabled=true
spring.ai.mcp.client.sse.connections.mcp-server1.url=http://localhost:8014

java 复制代码

package com.atguigu.study.controller;

import jakarta.annotation.Resource;
import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.model.ChatModel;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RequestParam;
import org.springframework.web.bind.annotation.RestController;
import reactor.core.publisher.Flux;

@RestController
public class McpClientController
{
    @Resource
    private ChatClient chatClient;//使用mcp支持

    @Resource
    private ChatModel chatModel;//没有纳入tool支持，普通调用

    // http://localhost:8015/mcpclient/chat?msg=上海
    @GetMapping("/mcpclient/chat")
    public Flux<String> chat(@RequestParam(name = "msg",defaultValue = "北京") String msg)
    {
        System.out.println("使用了mcp");
        return chatClient.prompt(msg).stream().content();
    }

    /**
     * http://localhost:8015/mcpclient/chat2?msg=上海
     * @param msg
     * @return
     */
    @RequestMapping("/mcpclient/chat2")
    public Flux<String> chat2(@RequestParam(name = "msg",defaultValue = "北京") String msg)
    {
        System.out.println("未使用mcp");
        return chatModel.stream(msg);
    }
}

Client添加tool calling的使用

tools就是server对应的服务

java 复制代码

package com.atguigu.study.config;

import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.model.ChatModel;
import org.springframework.ai.tool.ToolCallbackProvider;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;


@Configuration
public class SaaLLMConfig {
    @Bean
    public ChatClient chatClient(ChatModel chatModel, ToolCallbackProvider tools) {
        return ChatClient.builder(chatModel)
                .defaultToolCallbacks(tools.getToolCallbacks())  //mcp协议，配置见yml文件
                .build();
    }
}

baidu MCP

java 复制代码

package com.atguigu.study.controller;

import jakarta.annotation.Resource;
import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.model.ChatModel;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
import reactor.core.publisher.Flux;

/**
 * @auther zzyybs@126.com
 * @create 2025-08-01 15:57
 * @Description TODO
 */
@RestController
public class McpClientCallBaiDuMcpController {
    @Resource
    private ChatClient chatClient; //添加了MCP调用能力

    @Resource
    private ChatModel chatModel; //没有添加MCP调用能力

    /**
     * 添加了MCP调用能力
     * http://localhost:8016/mcp/chat?msg=查询北纬39.9042东经116.4074天气
     * http://localhost:8016/mcp/chat?msg=查询61.149.121.66归属地
     * http://localhost:8016/mcp/chat?msg=查询昌平到天安门路线规划
     *
     * @param msg
     * @return
     */
    @GetMapping("/mcp/chat")
    public Flux<String> chat(String msg) {
        return chatClient.prompt(msg).stream().content();
    }

    /**
     * 没有添加MCP调用能力
     * http://localhost:8016/mcp/chat2?msg=查询北纬39.9042东经116.4074天气
     *
     * @param msg
     * @return
     */
    @RequestMapping("/mcp/chat2")
    public Flux<String> chat2(String msg) {
        return chatModel.stream(msg);
    }
}

java 复制代码

package com.atguigu.study.config;

import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.model.ChatModel;
import org.springframework.ai.tool.ToolCallbackProvider;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

/**
 * @auther zzyybs@126.com
 * @create 2025-07-31 20:47
 * @Description TODO
 */
@Configuration
public class SaaLLMConfig
{
    @Bean
    public ChatClient chatClient(ChatModel chatModel, ToolCallbackProvider tools)
    {
        return ChatClient.builder(chatModel)
                //mcp协议，配置见yml文件，此处只赋能给ChatClient对象
                .defaultToolCallbacks(tools.getToolCallbacks())
                .build();
    }
}

properties 复制代码

server.port=8016

# \u8BBE\u7F6E\u5168\u5C40\u7F16\u7801\u683C\u5F0F
server.servlet.encoding.enabled=true
server.servlet.encoding.force=true
server.servlet.encoding.charset=UTF-8

spring.application.name=SAA-16ClientCallBaiduMcpServer

# ====SpringAIAlibaba Config=============
spring.ai.dashscope.api-key=${aliQwen}

# ====mcp-client Config=============
spring.ai.mcp.client.request-timeout=20s
spring.ai.mcp.client.toolcallback.enabled=true
spring.ai.mcp.client.stdio.servers-configuration=classpath:/mcp-server.json5

mac的配置

json 复制代码

{
  "mcpServers": {
    "baidu-map": {
      "command": "npx",
      "args": [
        "-y",
        "@baidumap/mcp-server-baidu-map"
      ],
      "env": {
        "BAIDU_MAP_API_KEY": "API key"
      }
    }
  }
}

MCP底层源码

其实就是调用baidu的地址，这里就相当于是function calling