Spring Ai Alibaba-1.1.0.0-M5-SequentialAgent
在顺序执行模式中，多个Agent按预定义的顺序依次执行。每个Agent的输出成为下一个Agent的输入。
流程：
- Agent A处理初始输入
- Agent A 的输出传递给Agent B
- Agent B 处理并传递给Agent C
- 最后一个Agent返回最终结果
一、🏗️ 架构概览
二、✨ 项目概述
这是一个基于 Spring AI Alibaba 的多 Agent 协作系统示例项目。该项目演示了如何构建和运行多个 AI Agent 之间的协作流程，特别展示了顺序执行的 Agent 工作流。项目还包含了完善的监控和性能跟踪功能。
三、🛠️ 技术栈
- Spring Boot 3.4.x
- Spring AI Alibaba
- Alibaba Cloud DashScope (通义千问 API)
- micrometer 1.16.0
- Java 17+
四、📁 项目结构
SpringAiAlibabaMultiAgent/
├── src/
│ ├── main/
│ │ ├── java/
│ │ │ └── com/alibaba/springaialibabamultiagent/
│ │ │ ├── SpringAiAlibabaMultiAgentApplication.java # 应用启动类
│ │ │ ├── config/ # 配置类目录
│ │ │ │ ├── AgentConfig.java # Agent 配置类
│ │ │ │ ├── MonitoringConfig.java # 监控配置类
│ │ │ │ └── StorageConfig.java # 存储配置类
│ │ │ ├── controller/ # 控制器目录
│ │ │ │ └── SequentialAgentController.java # 顺序 Agent 控制器
│ │ └── resources/
│ │ └── application.properties # 应用配置文件
├── pom.xml # Maven 配置文件
└── README.md # 项目说明文件五、⚙️ 配置pom.xml
项目集成了 Micrometer 和 Prometheus 监控，可以通过以下端点访问监控信息：
-
http://localhost:8020/actuator/health: 健康检查 -
http://localhost:8020/actuator/metrics: 指标信息 -
http://localhost:8020/actuator/prometheus: Prometheus 格式的监控数据dependency>
<dependency> <groupId>com.alibaba.cloud.ai</groupId> <artifactId>spring-ai-alibaba-starter-dashscope</artifactId> <version>1.1.0.0-M5</version> </dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-web</artifactId>
<version>3.4.12</version>
</dependency>
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-actuator</artifactId>
<version>3.4.12</version>
</dependency>
<dependency>
<groupId>io.micrometer</groupId>
<artifactId>micrometer-registry-prometheus</artifactId>
<version>1.16.0</version>
</dependency>
<dependency>
<groupId>com.alibaba.cloud.ai</groupId>
<artifactId>spring-ai-alibaba-agent-framework</artifactId>
<version>1.1.0.0-M5</version>
</dependency>
六、📝 配置application.properties
server.port=8020
spring.application.name=SpringAiAlibabaMultiAgent
# ====SpringAIAlibaba Config=============
spring.ai.dashscope.api-key=${qwen-api}
spring.ai.dashscope.model=qwen-turbo
# 配置温度参数
spring.ai.dashscope.chat.options.temperature=0.5
# 配置最大token数
spring.ai.dashscope.chat.options.max-tokens=1048
# 启用字符编码过滤器
server.servlet.encoding.enabled=true
# 强制对所有请求和响应使用指定的字符编码
server.servlet.encoding.force=true
# 设置字符编码
server.servlet.encoding.charset=UTF-8
# Micrometer 监控配置
management.endpoints.web.exposure.include=health,info,metrics,prometheus
management.prometheus.metrics.export.enabled=true七、🧰 配置类
🔗存储配置类
@Configuration
public class StorageConfig {
@Bean
public MemoryStore memoryStore() {
return new MemoryStore();
}
@Bean
public MemorySaver memorySaver() {
return new MemorySaver();
}
}🔗注册表配置类
@Configuration
public class MonitoringConfig {
/**
* 创建ObservationRegistry Bean
* 用于配置和初始化ObservationRegistry实例
* @return ObservationRegistry实例
*/
@Bean
public ObservationRegistry observationRegistry() {
return ObservationRegistry.create();
}
}🔗agent 配置类
@Configuration
public class AgentConfig {
/**
* 创建写作Agent Bean
* @param dashScopeChatModel DashScope聊天模型
* @return ReactAgent实例
*/
@Bean
public ReactAgent writerAgent(DashScopeChatModel dashScopeChatModel, MemorySaver memorySaver) {
return ReactAgent.builder()
.name("writer_agent")
.model(dashScopeChatModel)
.description("专业写作Agent")
.instruction("你是一个知名的作家，擅长写作和创作。请根据用户的提问进行回答。")
.outputKey("article")
.saver(memorySaver)
.build();
}
/**
* 创建评审Agent Bean
* @param dashScopeChatModel DashScope聊天模型
* @return ReactAgent实例
*/
@Bean
public ReactAgent reviewerAgent(DashScopeChatModel dashScopeChatModel, MemorySaver memorySaver) {
return ReactAgent.builder()
.name("reviewer_agent")
.model(dashScopeChatModel)
.description("专业评审Agent")
.instruction("你是一个知名的评论家，擅长对文章进行评论和修改。" +
"对于散文类文章，请确保文章中必须包含对于西湖风景的描述。" +
"最终只返回修改后的文章，不要包含任何评论信息。")
.outputKey("reviewed_article")
.saver(memorySaver)
.build();
}
}八、🎯 控制器
package com.alibaba.springaialibabamultiagent.controller;
import com.alibaba.cloud.ai.graph.CompileConfig;
import com.alibaba.cloud.ai.graph.OverAllState;
import com.alibaba.cloud.ai.graph.RunnableConfig;
import com.alibaba.cloud.ai.graph.agent.ReactAgent;
import com.alibaba.cloud.ai.graph.agent.flow.agent.SequentialAgent;
import com.alibaba.cloud.ai.graph.checkpoint.config.SaverConfig;
import com.alibaba.cloud.ai.graph.checkpoint.savers.MemorySaver;
import com.alibaba.cloud.ai.graph.store.stores.MemoryStore;
import com.alibaba.springaialibabamultiagent.listener.ComprehensiveGraphListener;
import io.micrometer.observation.ObservationRegistry;
import jakarta.annotation.Resource;
import org.springframework.ai.chat.messages.AssistantMessage;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestParam;
import org.springframework.web.bind.annotation.RestController;
import java.util.List;
import java.util.Optional;
@RestController
public class SequentialAgentController {
@Resource
private ReactAgent writerAgent;
@Resource
private ReactAgent reviewerAgent;
@Resource
private ObservationRegistry observationRegistry;
@Resource
private MemoryStore memoryStore;
/**
* 创建一个顺序执行的博客撰写和评审Agent
*
* @param topic 博客主题
* @return 执行结果字符串
*/
@GetMapping("/ai/blog")
public String generateBlog(@RequestParam(name = "topic", defaultValue = "西湖风景") String topic,
@RequestParam(name = "user_id", defaultValue = "西湖风景") String user_id) {
try {
// 使用基于用户ID的稳定threadId，确保同一用户的对话可以保持记忆
String threadId = "user_" + user_id;
SequentialAgent blogAgent = SequentialAgent.builder()
.name("blog_agent")
.subAgents(List.of(writerAgent, reviewerAgent))
.description("一个顺序执行的博客撰写和评审Agent")
.compileConfig(CompileConfig.builder()
.recursionLimit(20) // 限制最多递归20次
//http://localhost:8020/actuator/prometheus
.observationRegistry(observationRegistry)
//.withLifecycleListener(new ComprehensiveGraphListener())
.store(memoryStore)
.build())
.build();
RunnableConfig runnableConfig = RunnableConfig.builder()
.threadId(threadId)
.addMetadata("user_id", user_id)
.store(memoryStore)
.build();
Optional<OverAllState> result = blogAgent.invoke("请写一篇关于'" + topic + "'的300字左右的散文",runnableConfig);
StringBuilder response = new StringBuilder();
if (result.isPresent()) {
OverAllState state = result.get();
// 获取第一个Agent（writer_agent）的输出
Optional<Object> articleObj = state.value("article");
if (articleObj.isPresent() && articleObj.get() instanceof AssistantMessage) {
AssistantMessage article = (AssistantMessage) articleObj.get();
response.append("===原始文章===: ").append(article.getText()).append("\n\n");
}
// 获取第二个Agent（reviewer_agent）的输出
Optional<Object> reviewedArticleObj = state.value("reviewed_article");
if (reviewedArticleObj.isPresent() && reviewedArticleObj.get() instanceof AssistantMessage) {
AssistantMessage reviewedArticle = (AssistantMessage) reviewedArticleObj.get();
response.append("===评审后文章===: ").append(reviewedArticle.getText());
}
}
return response.toString();
} catch (Exception e) {
// 更好的错误处理，记录详细错误信息
return "发生错误: " + e.getMessage() + "\n请稍后重试或检查网络连接";
}
}
}九、🌐 测试场景
http://localhost:8020/ai/blog?topic=天降祥瑞&user_id=1十、🚀 扩展能力
-
统一监控与治理：耗时统计、成功率分析等。
-
prometheus 实时监控
-
引入redisStore实现存储