fix：修复上下文消息构建顺序，确保AI正确理解对话上下文

消息顺序调整为：历史消息 → 知识库内容 → 当前用户消息

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

README.md

@@ -32,7 +32,7 @@
 |     模块     | 现有能力 
 |:----------:|---
 |  **模型管理**  | 多模型接入(OpenAI/DeepSeek/通义/智谱)、多模态理解、Coze/DIFY/FastGPT平台集成 
-|  **知识管理**  | 本地RAG + 向量库(Milvus/Weaviate)  + 文档解析        
+|  **知识管理**  | 本地RAG + 向量库(Milvus/Weaviate/Qdrant)  + 文档解析        
 |  **工具管理**  | Mcp协议集成、Skills能力 + 可扩展工具生态                             
 |  **流程编排**  | 可视化工作流设计器、节点拖拽编排、SSE流式执行,目前已经支持模型调用,邮件发送,人工审核等节点  
 |  **多智能体**  | 基于Langchain4j的Agent框架、Supervisor模式编排,支持多种决策模型          
@@ -62,12 +62,16 @@
 ## 🛠️ 技术架构
 
 ### 核心框架
-- **后端架构**：Spring Boot 4.0 + Spring ai 2.0 + Langchain4j
+- **后端架构**：Spring Boot 3.5.8 + Langchain4j
-- **数据存储**：MySQL 8.0 + Redis + 向量数据库（Milvus/Weaviate）
+- **数据存储**：MySQL 8.0 + Redis + 向量数据库（Milvus/Weaviate/Qdrant）
 - **前端技术**：Vue 3 + Vben Admin + element-plus-x
 - **安全认证**：Sa-Token + JWT 双重保障
 
 
+- **文档处理**：PDF、Word、Excel 解析，图像智能分析
+- **实时通信**：WebSocket 实时通信，SSE 流式响应
+- **系统监控**：完善的日志体系、性能监控、服务健康检查
+
 ## 🐳 Docker 部署
 
 本项目提供两种 Docker 部署方式：
@@ -218,9 +222,6 @@ docker-compose -f docker-compose-all.yaml restart [服务名]
   算力和模型 API 服务
 - [优云智算](https://www.compshare.cn/?ytag=GPU_YY-gh_ruoyi) - 万卡RTX40系GPU+海内外主流模型API服务，秒级响应，按量计费，新客免费用。
 
-## 优秀开源项目及社区推荐
-- [imaiwork](https://gitee.com/tsinghua-open/imaiwork) - AI手机开源版，AI获客手机项目，基于无障碍模式，RPA，比豆包AI手机更强大。
-
 ## 💬 社区交流
 
 <div align="center">

README_EN.md

@@ -32,14 +32,13 @@
 
 ## ✨ Core Features
 
-| Module | Current Capabilities | Extension Direction |
+| Module | Current Capabilities |
-|:---:|---|---|
+|:---:|---|
-| **Model Management** | Multi-model integration (OpenAI/DeepSeek/Tongyi/Zhipu), multi-modal understanding, Coze/DIFY/FastGPT platform integration | Auto mode, fault tolerance |
+| **Model Management** | Multi-model integration (OpenAI/DeepSeek/Tongyi/Zhipu), multi-modal understanding, Coze/DIFY/FastGPT platform integration |
-| **Knowledge Base** | Local RAG + Vector DB (Milvus/Weaviate) + Knowledge Graph + Document parsing + Reranking | Audio/video parsing, knowledge source |
+| **Knowledge Base** | Local RAG + Vector DB (Milvus/Weaviate/Qdrant) + Document parsing |
-| **Tool Management** | MCP protocol integration, Skills capability + Extensible tool ecosystem | Tool plugin marketplace, toolAgent auto-loading |
+| **Tool Management** | MCP protocol integration, Skills capability + Extensible tool ecosystem |
-| **Workflow Orchestration** | Visual workflow designer, drag-and-drop node orchestration, SSE streaming execution, currently supports model (with RAG) calls, email sending, manual review nodes | More node types |
+| **Workflow Orchestration** | Visual workflow designer, drag-and-drop node orchestration, SSE streaming execution, currently supports model calls, email sending, manual review nodes |
-| **Multi-Agent** | Agent framework based on Langchain4j, Supervisor mode orchestration, supports multiple decision models | Configurable agents |
+| **Multi-Agent** | Agent framework based on Langchain4j, Supervisor mode orchestration, supports multiple decision models |
-| **AI Coding** | Intelligent code analysis, project scaffolding generation, Copilot assistant | Code generation optimization |
 
 ## 🚀 Quick Start
 
@@ -59,19 +58,134 @@
 | 🛠️ Admin Panel | [ruoyi-admin](https://github.com/ageerle/ruoyi-admin) | [ruoyi-admin](https://gitee.com/ageerle/ruoyi-admin) | [ruoyi-admin](https://gitcode.com/ageerle/ruoyi-admin) |
 
 ### Partner Projects
-| Project Name           | GitHub Repository                                             | Gitee Repository
+| Project Name           | GitHub Repository                                             | Gitee Repository |
 |----------------|-------------------------------------------------------|------------------------------------------------------|
 | element-plus-x | [element-plus-x](https://github.com/element-plus-x/Element-Plus-X)       | [element-plus-x](https://gitee.com/he-jiayue/element-plus-x)       |
 
 ## 🛠️ Technical Architecture
 
 ### Core Framework
-- **Backend**: Spring Boot 4.0 + Spring AI 2.0 + Langchain4j
+- **Backend**: Spring Boot 3.5.8 + Langchain4j
-- **Data Storage**: MySQL 8.0 + Redis + Vector Databases (Milvus/Weaviate)
+- **Data Storage**: MySQL 8.0 + Redis + Vector Databases (Milvus/Weaviate/Qdrant)
 - **Frontend**: Vue 3 + Vben Admin + element-plus-x
 - **Security**: Sa-Token + JWT dual-layer security
 
 
+- **Document Processing**: PDF, Word, Excel parsing, intelligent image analysis
+- **Real-time Communication**: WebSocket real-time communication, SSE streaming response
+- **System Monitoring**: Comprehensive logging system, performance monitoring, service health checks
+
+## 🐳 Docker Deployment
+
+This project provides two Docker deployment methods:
+
+### Method 1: One-click Start All Services (Recommended)
+
+Use `docker-compose-all.yaml` to start all services at once (including backend, admin panel, user frontend, and dependencies):
+
+```bash
+# Clone the repository
+git clone https://github.com/ageerle/ruoyi-ai.git
+cd ruoyi-ai
+
+# Start all services (pull pre-built images from registry)
+docker-compose -f docker-compose-all.yaml up -d
+
+# Check service status
+docker-compose -f docker-compose-all.yaml ps
+
+# Access services
+# Admin Panel: http://localhost:25666 (admin / admin123)
+# User Frontend: http://localhost:25137
+# Backend API: http://localhost:26039
+```
+
+### Method 2: Step-by-step Deployment (Source Build)
+
+If you need to build backend services from source, follow these steps:
+
+#### Step 1: Deploy Backend Service
+
+```bash
+# Enter backend project directory
+cd ruoyi-ai
+
+# Start backend service (build from source)
+docker-compose up -d --build
+
+# Wait for backend service to start
+docker-compose logs -f backend
+```
+
+#### Step 2: Deploy Admin Panel
+
+```bash
+# Enter admin panel project directory
+cd ruoyi-admin
+
+# Build and start admin panel
+docker-compose up -d --build
+
+# Access admin panel
+# URL: http://localhost:5666
+```
+
+#### Step 3: Deploy User Frontend (Optional)
+
+```bash
+# Enter user frontend project directory
+cd ruoyi-web
+
+# Build and start user frontend
+docker-compose up -d --build
+
+# Access user frontend
+# URL: http://localhost:5137
+```
+
+### Service Ports
+
+| Service | One-click Port | Step-by-step Port | Description |
+|------|-------------|-------------|------|
+| Admin Panel | 25666 | 5666 | Admin backend access |
+| User Frontend | 25137 | 5137 | User frontend access |
+| Backend Service | 26039 | 6039 | Backend API service |
+| MySQL | 23306 | 23306 | Database service |
+| Redis | 26379 | 6379 | Cache service |
+| Weaviate | 28080 | 28080 | Vector database |
+| MinIO API | 29000 | 9000 | Object storage API |
+| MinIO Console | 29090 | 9090 | Object storage console |
+
+### Image Registry
+
+All images are hosted on Alibaba Cloud Container Registry:
+
+```
+crpi-31mraxd99y2gqdgr.cn-beijing.personal.cr.aliyuncs.com/ruoyi_ai
+```
+
+Available images:
+- `mysql:v3` - MySQL database (includes initialization SQL)
+- `redis:6.2` - Redis cache
+- `weaviate:1.30.0` - Vector database
+- `minio:latest` - Object storage
+- `ruoyi-ai-backend:latest` - Backend service
+- `ruoyi-ai-admin:latest` - Admin frontend
+- `ruoyi-ai-web:latest` - User frontend
+
+### Common Commands
+
+```bash
+# Stop all services
+docker-compose -f docker-compose-all.yaml down
+
+# View service logs
+docker-compose -f docker-compose-all.yaml logs -f [service-name]
+
+# Restart a service
+docker-compose -f docker-compose-all.yaml restart [service-name]
+```
+
 ## 📚 Documentation
 
 Want to learn more about installation, deployment, configuration, and secondary development?
@@ -109,14 +223,13 @@ Thanks to the following excellent open-source projects for their support:
 - [PPIO Cloud](https://ppinfra.com/user/register?invited_by=P8QTUY&utm_source=github_ruoyi-ai) - Provides cost-effective GPU computing and model API services
 - [Youyun Intelligent Computing](https://www.compshare.cn/?ytag=GPU_YY-gh_ruoyi) - Thousands of RTX40 series GPUs + mainstream models API services, second-level response, pay-per-use, free for new customers.
 
-## Outstanding Open-Source Projects and Community Recommendations
-- [imaiwork](https://gitee.com/tsinghua-open/imaiwork) - Open-source AI phone, AI customer acquisition phone project, based on accessibility mode and RPA, more powerful than Doubao AI phone.
 
 ## 💬 Community Chat
 
 <div align="center">
 
-**[📱 Join Telegram Group](https://t.me/+LqooQAc5HxRmYmE1)**
+**[📱 Join Telegram Group](
+https://t.me/+LqooQAc5HxRmYmE1)**
 
 </div>
 

docs/docker/qdrant/docker-compose.yml

@@ -0,0 +1,12 @@
+---
+services:
+  qdrant:
+    image: qdrant/qdrant:latest
+    ports:
+      - 6333:6333
+      - 6334:6334
+    volumes:
+      - qdrant_data:/qdrant/storage
+volumes:
+  qdrant_data:
+...

ruoyi-admin/src/main/java/org/ruoyi/RuoYiAIApplication.java

@@ -4,6 +4,9 @@ import org.springframework.boot.SpringApplication;
 import org.springframework.boot.autoconfigure.SpringBootApplication;
 import org.springframework.boot.context.metrics.buffering.BufferingApplicationStartup;
 
+import java.net.InetSocketAddress;
+import java.net.ServerSocket;
+
 /**
  * 启动程序
  *
@@ -13,10 +16,66 @@ import org.springframework.boot.context.metrics.buffering.BufferingApplicationSt
 public class RuoYiAIApplication {
 
     public static void main(String[] args) {
+        // killPortProcess(6039);
         SpringApplication application = new SpringApplication(RuoYiAIApplication.class);
         application.setApplicationStartup(new BufferingApplicationStartup(2048));
         application.run(args);
-        System.out.println("(♥◠‿◠)ﾉﾞ     RuoYi-AI启动成功   ლ(´ڡ`ლ)ﾞ");
+        System.out.println("(♥◠‿◠)ﾉﾞ     RuoYi-AI启动成功   ლ(´ڡ`ლ)冢");
+    }
+
+    /**
+     * 检查并终止占用指定端口的进程
+     *
+     * @param port 端口号
+     */
+    private static void killPortProcess(int port) {
+        try {
+            if (!isPortInUse(port)) {
+                return;
+            }
+            System.out.println("端口 " + port + " 已被占用，正在查找并终止进程...");
+
+            ProcessBuilder pb = new ProcessBuilder("netstat", "-ano");
+            Process process = pb.start();
+            java.io.BufferedReader reader = new java.io.BufferedReader(
+                    new java.io.InputStreamReader(process.getInputStream()));
+
+            String line;
+            while ((line = reader.readLine()) != null) {
+                if (line.contains(":" + port + " ") && line.contains("LISTENING")) {
+                    String[] parts = line.trim().split("\\s+");
+                    String pid = parts[parts.length - 1];
+                    System.out.println("找到占用端口 " + port + " 的进程 PID: " + pid + "，正在终止...");
+
+                    ProcessBuilder killPb = new ProcessBuilder("taskkill", "/F", "/PID", pid);
+                    Process killProcess = killPb.start();
+                    int exitCode = killProcess.waitFor();
+                    if (exitCode == 0) {
+                        System.out.println("进程 " + pid + " 已成功终止");
+                    } else {
+                        System.out.println("终止进程 " + pid + " 失败，exitCode: " + exitCode);
+                    }
+                    break;
+                }
+            }
+
+            // 等待一小段时间确保端口释放
+            Thread.sleep(500);
+        } catch (Exception e) {
+            System.out.println("检查/终止端口进程时发生异常: " + e.getMessage());
+        }
+    }
+
+    /**
+     * 检查端口是否被占用
+     */
+    private static boolean isPortInUse(int port) {
+        try (ServerSocket socket = new ServerSocket()) {
+            socket.bind(new InetSocketAddress(port));
+            return false;
+        } catch (Exception e) {
+            return true;
+        }
     }
 
 }

ruoyi-admin/src/main/resources/application.yml

@@ -275,7 +275,7 @@ warm-flow:
 
 # 向量库配置
 vector-store:
-  # 向量存储类型 可选(weaviate/milvus)
+  # 向量存储类型 可选(weaviate/milvus/qdrant)
   # 如需修改向量库类型，请修改此配置值!
   type: milvus
   # Weaviate配置
@@ -287,3 +287,10 @@ vector-store:
   milvus:
     url: http://localhost:19530
     collectionname: LocalKnowledge
+  # Qdrant配置
+  qdrant:
+    host: localhost
+    port: 6334
+    collectionname: LocalKnowledge
+    api-key:
+    use-tls: false

ruoyi-modules/ruoyi-aiflow/src/main/java/org/ruoyi/workflow/service/WorkflowComponentService.java

@@ -98,7 +98,7 @@ public class WorkflowComponentService extends ServiceImpl<WorkflowComponentMappe
         return baseMapper.selectPage(new Page<>(currentPage, pageSize), wrapper);
     }
 
-    @Cacheable(cacheNames = WORKFLOW_COMPONENTS)
+    // @Cacheable(cacheNames = WORKFLOW_COMPONENTS)
     public List<WorkflowComponent> getAllEnable() {
         return ChainWrappers.lambdaQueryChain(baseMapper)
                 .eq(WorkflowComponent::getIsEnable, true)

ruoyi-modules/ruoyi-chat/pom.xml

@@ -91,6 +91,12 @@
             <version>${langchain4j.community.version}</version>
         </dependency>
 
+        <dependency>
+            <groupId>dev.langchain4j</groupId>
+            <artifactId>langchain4j-qdrant</artifactId>
+            <version>${langchain4j.community.version}</version>
+        </dependency>
+
         <dependency>
             <groupId>dev.langchain4j</groupId>
             <artifactId>langchain4j-mcp</artifactId>

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/agent/ChartGenerationAgent.java

@@ -6,7 +6,7 @@ import dev.langchain4j.service.UserMessage;
 import dev.langchain4j.service.V;
 
 
-public interface ChartGenerationAgent extends Agent {
+public interface ChartGenerationAgent {
 
     @SystemMessage("""
             You are a chart generation specialist. Your only task is to generate Apache ECharts

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/agent/GetNameInfo.java

@@ -1,38 +0,0 @@
-package org.ruoyi.agent;
-
-import dev.langchain4j.agentic.Agent;
-import dev.langchain4j.service.SystemMessage;
-import dev.langchain4j.service.UserMessage;
-import dev.langchain4j.service.V;
-
-/**
- * User Name Retrieval Agent
- * A simple assistant that retrieves user names using the get_name tool.
- */
-public interface GetNameInfo {
-
-    @SystemMessage("""
-        You are a user identity assistant. You MUST always use tools to get information.
-
-        MANDATORY REQUIREMENTS:
-        - You MUST call the get_user_name_by_id tool for ANY question about names or identity
-        - NEVER respond without calling the get_user_name_by_id tool first
-        - Return ONLY the exact string returned by the get_user_name_by_id tool
-        - Do not make up names like "John Doe" or any other default names
-        - Do not use your knowledge to answer - ALWAYS use the tool
-
-        Your workflow:
-        1. Extract userId from the query (if mentioned), or use "1" as default
-        2. ALWAYS call the get_user_name_by_id tool with the userId parameter
-        3. Return the exact result as plain text with no additions
-
-        CRITICAL: If you don't call the get_user_name_by_id tool, your response is wrong.
-        """)
-    @UserMessage("""
-        Get the user name using the get_user_name_by_id tool. Query: {{query}}
-
-        IMPORTANT: Return only the exact result from the tool.
-        """)
-    @Agent("User identity assistant that returns user name from get_name tool")
-    String search(@V("query") String query);
-}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/agent/McpAgent.java

@@ -1,42 +0,0 @@
-package org.ruoyi.agent;
-
-import dev.langchain4j.agentic.Agent;
-import dev.langchain4j.service.SystemMessage;
-import dev.langchain4j.service.UserMessage;
-import dev.langchain4j.service.V;
-
-public interface McpAgent extends Agent {
-
-    /**
-     * 系统提示词：通用工具调用智能体
-     * 不限定具体工具类型，让 LangChain4j 自动传递工具描述给 LLM
-     */
-    @SystemMessage("""
-        你是一个AI助手，可以通过调用各种工具来帮助用户完成不同的任务。
-
-        【工具使用规则】
-        1. 根据用户的请求，判断需要使用哪些工具
-        2. 仔细阅读每个工具的描述，确保理解工具的功能和参数要求
-        3. 使用正确的参数调用工具
-        4. 如果工具执行失败，向用户友好地说明错误原因，并尝试提供替代方案
-        5. 对于复杂任务，可以分步骤使用多个工具完成
-        6. 将工具执行结果以清晰易懂的方式呈现给用户
-
-        【响应格式】
-        - 直接回答用户的问题
-        - 如果使用了工具，说明使用了什么工具以及结果
-        - 如果遇到错误，提供友好的错误信息和解决建议
-        """)
-
-    @UserMessage("""
-        {{query}}
-        """)
-
-    @Agent("通用工具调用智能体")
-    /**
-     * 智能体对外调用入口
-     * @param query 用户的自然语言请求
-     * @return 处理结果
-     */
-    String callMcpTool(@V("query") String query);
-}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/agent/SqlAgent.java

@@ -11,7 +11,7 @@ import dev.langchain4j.service.V;
  * and returning relevant data and analysis results.
  *
  */
-public interface SqlAgent extends Agent {
+public interface SqlAgent {
 
     @SystemMessage("""
         This agent is designed for MySQL 5.7

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/agent/WebSearchAgent.java

@@ -10,7 +10,7 @@ import dev.langchain4j.service.V;
  * A web search assistant that answers natural language questions by searching the internet
  * and returning relevant information from web pages.
  */
-public interface WebSearchAgent extends Agent {
+public interface WebSearchAgent {
 
     @SystemMessage("""
         You are a web search assistant. Answer questions by searching and retrieving web content.

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/agent/tool/ExecuteSqlQueryTool.java

@@ -43,7 +43,7 @@ public class ExecuteSqlQueryTool implements BuiltinToolProvider {
     @Tool("Execute a SELECT SQL query and return the results. Example: SELECT * FROM sys_user")
     public String executeSql(String sql) {
         // 2. 手动推入数据源上下文
-        DynamicDataSourceContextHolder.push("agent");
+//        DynamicDataSourceContextHolder.push("agent");
         if (sql == null || sql.trim().isEmpty()) {
             return "Error: SQL query cannot be empty";
         }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/config/VectorStoreProperties.java

@@ -59,4 +59,37 @@ public class VectorStoreProperties {
          */
         private String collectionname;
     }
+
+    /**
+     * Qdrant配置
+     */
+    private Qdrant qdrant = new Qdrant();
+
+    @Data
+    public static class Qdrant {
+        /**
+         * 主机地址
+         */
+        private String host = "localhost";
+
+        /**
+         * gRPC端口
+         */
+        private int port = 6334;
+
+        /**
+         * 集合名称
+         */
+        private String collectionname = "LocalKnowledge";
+
+        /**
+         * API密钥（可选）
+         */
+        private String apiKey;
+
+        /**
+         * 是否启用TLS
+         */
+        private boolean useTls = false;
+    }
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/factory/EmbeddingModelFactory.java

@@ -2,8 +2,9 @@ package org.ruoyi.factory;
 
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.common.chat.service.chat.IChatModelService;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
+import org.ruoyi.common.chat.service.chat.IChatModelService;
+import org.ruoyi.observability.EmbeddingModelListenerProvider;
 import org.ruoyi.service.embed.BaseEmbedModelService;
 import org.ruoyi.service.embed.MultiModalEmbedModelService;
 import org.springframework.beans.factory.NoSuchBeanDefinitionException;
@@ -27,6 +28,7 @@ public class EmbeddingModelFactory {
     private final ApplicationContext applicationContext;
 
     private final IChatModelService chatModelService;
+    private final EmbeddingModelListenerProvider embeddingModelListenerProvider;
 
     // 模型缓存，使用ConcurrentHashMap保证线程安全
     private final Map<String, BaseEmbedModelService> modelCache = new ConcurrentHashMap<>();
@@ -109,6 +111,8 @@ public class EmbeddingModelFactory {
             BaseEmbedModelService model = applicationContext.getBean(factory, BaseEmbedModelService.class);
             // 配置模型参数
             model.configure(config);
+            // 增加嵌入模型监听器
+            model.addListeners(embeddingModelListenerProvider.getEmbeddingModelListeners());
             log.info("成功创建嵌入模型: factory={}, modelId={}", config.getProviderCode(), config.getId());
             return model;
         } catch (NoSuchBeanDefinitionException e) {

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/factory/ResourceLoaderFactory.java

@@ -1,6 +1,7 @@
 package org.ruoyi.factory;
 
 import lombok.AllArgsConstructor;
+import org.apache.commons.lang3.StringUtils;
 import org.ruoyi.constant.FileTypeConstants;
 import org.ruoyi.service.knowledge.ResourceLoader;
 import org.ruoyi.service.knowledge.impl.loader.*;
@@ -16,6 +17,7 @@ public class ResourceLoaderFactory {
     private final ExcelTextSplitter excelTextSplitter;
 
     public ResourceLoader getLoaderByFileType(String fileType) {
+        fileType = StringUtils.removeStart(fileType, ".");
         if (FileTypeConstants.isTextFile(fileType)) {
             return new TextFileLoader(characterTextSplitter);
         } else if (FileTypeConstants.isWord(fileType)) {

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/factory/VectorStoreStrategyFactory.java

@@ -7,6 +7,7 @@ import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.config.VectorStoreProperties;
 import org.ruoyi.service.vector.VectorStoreService;
 import org.ruoyi.service.vector.impl.MilvusVectorStoreStrategy;
+import org.ruoyi.service.vector.impl.QdrantVectorStoreStrategy;
 import org.ruoyi.service.vector.impl.WeaviateVectorStoreStrategy;
 import org.springframework.stereotype.Component;
 
@@ -27,6 +28,7 @@ public class VectorStoreStrategyFactory {
     private final VectorStoreProperties vectorStoreProperties;
     private final WeaviateVectorStoreStrategy weaviateStrategy;
     private final MilvusVectorStoreStrategy milvusStrategy;
+    private final QdrantVectorStoreStrategy qdrantStrategy;
 
     private Map<String, VectorStoreService> strategies;
 
@@ -35,6 +37,7 @@ public class VectorStoreStrategyFactory {
         strategies = new HashMap<>();
         strategies.put("weaviate", weaviateStrategy);
         strategies.put("milvus", milvusStrategy);
+        strategies.put("qdrant", qdrantStrategy);
         log.info("向量库策略工厂初始化完成，支持的策略: {}", strategies.keySet());
     }
 

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/ChatModelListenerProvider.java

@@ -0,0 +1,41 @@
+package org.ruoyi.observability;
+
+import cn.hutool.core.collection.CollUtil;
+import dev.langchain4j.model.chat.listener.ChatModelListener;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelListener;
+import lombok.Getter;
+import org.springframework.context.annotation.Lazy;
+import org.springframework.lang.Nullable;
+import org.springframework.stereotype.Component;
+
+import java.util.Collections;
+import java.util.List;
+
+/**
+ * LangChain4j 监听器共享提供者。
+ * <p>
+ * 供所有 {@link dev.langchain4j.model.chat.StreamingChatModel} 构建器使用，
+ * 将可观测性监听器注入到模型实例中。
+ *
+ * @author evo
+ */
+@Component
+@Getter
+@Lazy
+public class ChatModelListenerProvider {
+
+    private final List<ChatModelListener> chatModelListeners;
+    private final List<EmbeddingModelListener> embeddingModelListeners;
+
+    public ChatModelListenerProvider(@Nullable List<ChatModelListener> chatModelListeners,
+                                     @Nullable List<EmbeddingModelListener> embeddingModelListeners) {
+        if (CollUtil.isEmpty(chatModelListeners)) {
+            chatModelListeners = Collections.emptyList();
+        }
+        if (CollUtil.isEmpty(embeddingModelListeners)) {
+            embeddingModelListeners = Collections.emptyList();
+        }
+        this.chatModelListeners = chatModelListeners;
+        this.embeddingModelListeners = embeddingModelListeners;
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/EmbeddingModelListenerProvider.java

@@ -0,0 +1,34 @@
+package org.ruoyi.observability;
+
+import cn.hutool.core.collection.CollUtil;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelListener;
+import lombok.Getter;
+import org.springframework.context.annotation.Lazy;
+import org.springframework.lang.Nullable;
+import org.springframework.stereotype.Component;
+
+import java.util.Collections;
+import java.util.List;
+
+/**
+ * EmbeddingModel 监听器共享提供者。
+ * <p>
+ * 供所有 {@link dev.langchain4j.model.embedding.EmbeddingModel} 构建器使用，
+ * 将可观测性监听器注入到模型实例中。
+ *
+ * @author evo
+ */
+@Component
+@Getter
+@Lazy
+public class EmbeddingModelListenerProvider {
+
+    private final List<EmbeddingModelListener> embeddingModelListeners;
+
+    public EmbeddingModelListenerProvider(@Nullable List<EmbeddingModelListener> embeddingModelListeners) {
+        if (CollUtil.isEmpty(embeddingModelListeners)) {
+            embeddingModelListeners = Collections.emptyList();
+        }
+        this.embeddingModelListeners = embeddingModelListeners;
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/LangChain4jObservabilityConfig.java

@@ -0,0 +1,129 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.Experimental;
+import dev.langchain4j.mcp.client.McpClientListener;
+import dev.langchain4j.model.chat.listener.ChatModelListener;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelListener;
+import dev.langchain4j.observability.api.AiServiceListenerRegistrar;
+import dev.langchain4j.observability.api.listener.*;
+import jakarta.annotation.PostConstruct;
+import lombok.RequiredArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.context.annotation.Bean;
+import org.springframework.context.annotation.Configuration;
+
+import java.util.List;
+
+/**
+ * LangChain4j 可观测性配置类。
+ * <p>
+ * 负责注册所有 langchain4j 的监听器：
+ * <ul>
+ *   <li>{@link AiServiceListener} - AI服务级别的事件监听器（通过 AiServiceListenerRegistrar 注册）</li>
+ *   <li>{@link ChatModelListener} - ChatModel 级别的监听器（注入到模型构建器）</li>
+ *   <li>{@link EmbeddingModelListener} - EmbeddingModel 级别的监听器（注入到模型构建器）</li>
+ * </ul>
+ *
+ * @author evo
+ */
+@Configuration
+@RequiredArgsConstructor
+@Slf4j
+public class LangChain4jObservabilityConfig {
+
+    private final AiServiceListenerRegistrar registrar = AiServiceListenerRegistrar.newInstance();
+
+    /**
+     * 注册 AI 服务级别的事件监听器
+     */
+    @PostConstruct
+    public void registerAiServiceListeners() {
+        log.info("正在注册 LangChain4j AI Service 事件监听器...");
+        registrar.register(
+            new MyAiServiceStartedListener(),
+            new MyAiServiceRequestIssuedListener(),
+            new MyAiServiceResponseReceivedListener(),
+            new MyAiServiceCompletedListener(),
+            new MyAiServiceErrorListener(),
+            new MyInputGuardrailExecutedListener(),
+            new MyOutputGuardrailExecutedListener(),
+            new MyToolExecutedEventListener()
+        );
+        log.info("LangChain4j AI Service 事件监听器注册完成");
+    }
+
+    // ==================== AI Service 监听器 Beans ====================
+
+    @Bean
+    public AiServiceStartedListener aiServiceStartedListener() {
+        return new MyAiServiceStartedListener();
+    }
+
+    @Bean
+    public AiServiceRequestIssuedListener aiServiceRequestIssuedListener() {
+        return new MyAiServiceRequestIssuedListener();
+    }
+
+    @Bean
+    public AiServiceResponseReceivedListener aiServiceResponseReceivedListener() {
+        return new MyAiServiceResponseReceivedListener();
+    }
+
+    @Bean
+    public AiServiceCompletedListener aiServiceCompletedListener() {
+        return new MyAiServiceCompletedListener();
+    }
+
+    @Bean
+    public AiServiceErrorListener aiServiceErrorListener() {
+        return new MyAiServiceErrorListener();
+    }
+
+    @Bean
+    public InputGuardrailExecutedListener inputGuardrailExecutedListener() {
+        return new MyInputGuardrailExecutedListener();
+    }
+
+    @Bean
+    public OutputGuardrailExecutedListener outputGuardrailExecutedListener() {
+        return new MyOutputGuardrailExecutedListener();
+    }
+
+    @Bean
+    public ToolExecutedEventListener toolExecutedEventListener() {
+        return new MyToolExecutedEventListener();
+    }
+
+    // ==================== ChatModel 监听器 ====================
+
+    @Bean
+    public ChatModelListener chatModelListener() {
+        return new MyChatModelListener();
+    }
+
+    @Bean
+    public List<ChatModelListener> chatModelListeners() {
+        return List.of(new MyChatModelListener());
+    }
+
+    // ==================== EmbeddingModel 监听器 ====================
+
+    @Bean
+    @Experimental
+    public EmbeddingModelListener embeddingModelListener() {
+        return new MyEmbeddingModelListener();
+    }
+
+    @Bean
+    @Experimental
+    public List<EmbeddingModelListener> embeddingModelListeners() {
+        return List.of(new MyEmbeddingModelListener());
+    }
+
+    // ==================== MCP Client 监听器 ====================
+
+    @Bean
+    public McpClientListener mcpClientListener() {
+        return new MyMcpClientListener();
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyAgentListener.java

@@ -0,0 +1,145 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.agentic.observability.AgentInvocationError;
+import dev.langchain4j.agentic.observability.AgentRequest;
+import dev.langchain4j.agentic.observability.AgentResponse;
+import dev.langchain4j.agentic.planner.AgentInstance;
+import dev.langchain4j.agentic.scope.AgenticScope;
+import dev.langchain4j.service.tool.BeforeToolExecution;
+import dev.langchain4j.service.tool.ToolExecution;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.Map;
+import java.util.concurrent.atomic.AtomicReference;
+
+/**
+ * 自定义的 AgentListener 的监听器。
+ * 监听 Agent 相关的所有可观测性事件，包括：
+ * <ul>
+ *   <li>Agent 调用前/后的生命周期事件</li>
+ *   <li>Agent 执行错误事件</li>
+ *   <li>AgenticScope 的创建/销毁事件</li>
+ *   <li>工具执行前/后的生命周期事件</li>
+ * </ul>
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyAgentListener implements dev.langchain4j.agentic.observability.AgentListener {
+
+    /** 最终捕获到的思考结果（主 Agent 完成后写入，供外部获取） */
+    private final AtomicReference<String> sharedOutputRef = new AtomicReference<>();
+
+    public String getCapturedResult() {
+        return sharedOutputRef.get();
+    }
+
+    // ==================== Agent 调用生命周期 ====================
+
+    @Override
+    public void beforeAgentInvocation(AgentRequest agentRequest) {
+        AgentInstance agent = agentRequest.agent();
+        AgenticScope scope = agentRequest.agenticScope();
+        Map<String, Object> inputs = agentRequest.inputs();
+
+        log.info("【Agent调用前】Agent名称: {}", agent.name());
+        log.info("【Agent调用前】Agent ID: {}", agent.agentId());
+        log.info("【Agent调用前】Agent类型: {}", agent.type().getName());
+        log.info("【Agent调用前】Agent描述: {}", agent.description());
+        log.info("【Agent调用前】Planner类型: {}", agent.plannerType());
+        log.info("【Agent调用前】输出类型: {}", agent.outputType());
+        log.info("【Agent调用前】输出Key: {}", agent.outputKey());
+        log.info("【Agent调用前】是否为异步: {}", agent.async());
+        log.info("【Agent调用前】是否为叶子节点: {}", agent.leaf());
+        log.info("【Agent调用前】Agent参数列表:");
+        for (var arg : agent.arguments()) {
+            log.info("  - 参数名: {}, 类型: {}, 默认值: {}",
+                    arg.name(), arg.rawType().getName(), arg.defaultValue());
+        }
+        log.info("【Agent调用前】Agent输入参数: {}", inputs);
+        log.info("【Agent调用前】AgenticScope memoryId: {}", scope.memoryId());
+        log.info("【Agent调用前】AgenticScope当前状态: {}", scope.state());
+        log.info("【Agent调用前】Agent调用历史记录数: {}", scope.agentInvocations().size());
+
+        // 打印嵌套的子Agent信息
+        if (!agent.subagents().isEmpty()) {
+            log.info("【Agent调用前】子Agent列表:");
+            for (AgentInstance sub : agent.subagents()) {
+                log.info("  - 子Agent: {} ({})", sub.name(), sub.type().getName());
+            }
+        }
+
+        // 打印父Agent信息
+        if (agent.parent() != null) {
+            log.info("【Agent调用前】父Agent: {}", agent.parent().name());
+        }
+    }
+
+    @Override
+    public void afterAgentInvocation(AgentResponse agentResponse) {
+        AgentInstance agent = agentResponse.agent();
+        Map<String, Object> inputs = agentResponse.inputs();
+        Object output = agentResponse.output();
+        String outputStr = output != null ? output.toString() : "";
+
+        log.info("【Agent调用后】Agent名称: {}", agent.name());
+        log.info("【Agent调用后】Agent ID: {}", agent.agentId());
+        log.info("【Agent调用后】Agent输入参数: {}", inputs);
+        log.info("【Agent调用后】Agent输出结果: {}", output);
+        log.info("【Agent调用后】是否为叶子节点: {}", agent.leaf());
+
+        // 捕获主 Agent 的最终输出，供外部获取
+        if ("invoke".equals(agent.agentId()) && !outputStr.isEmpty()) {
+            sharedOutputRef.set(outputStr);
+            log.info("【Agent调用后】已捕获主Agent输出: {}", outputStr);
+        }
+    }
+
+    @Override
+    public void onAgentInvocationError(AgentInvocationError error) {
+        AgentInstance agent = error.agent();
+        Map<String, Object> inputs = error.inputs();
+        Throwable throwable = error.error();
+
+        log.error("【Agent执行错误】Agent名称: {}", agent.name());
+        log.error("【Agent执行错误】Agent ID: {}", agent.agentId());
+        log.error("【Agent执行错误】Agent类型: {}", agent.type().getName());
+        log.error("【Agent执行错误】Agent输入参数: {}", inputs);
+        log.error("【Agent执行错误】错误类型: {}", throwable.getClass().getName());
+        log.error("【Agent执行错误】错误信息: {}", throwable.getMessage(), throwable);
+    }
+
+    // ==================== AgenticScope 生命周期 ====================
+
+    @Override
+    public void afterAgenticScopeCreated(AgenticScope agenticScope) {
+        log.info("【AgenticScope已创建】memoryId: {}", agenticScope.memoryId());
+        log.info("【AgenticScope已创建】初始状态: {}", agenticScope.state());
+    }
+
+    @Override
+    public void beforeAgenticScopeDestroyed(AgenticScope agenticScope) {
+        log.info("【AgenticScope即将销毁】memoryId: {}", agenticScope.memoryId());
+        log.info("【AgenticScope即将销毁】最终状态: {}", agenticScope.state());
+        log.info("【AgenticScope即将销毁】总调用次数: {}", agenticScope.agentInvocations().size());
+    }
+
+    // ==================== 工具执行生命周期 ====================
+
+    @Override
+    public void beforeToolExecution(BeforeToolExecution beforeToolExecution) {
+        var toolRequest = beforeToolExecution.request();
+        log.info("【工具执行前】工具请求ID: {}", toolRequest.id());
+        log.info("【工具执行前】工具名称: {}", toolRequest.name());
+        log.info("【工具执行前】工具参数: {}", toolRequest.arguments());
+    }
+
+    @Override
+    public void afterToolExecution(ToolExecution toolExecution) {
+        var toolRequest = toolExecution.request();
+        log.info("【工具执行后】工具请求ID: {}", toolRequest.id());
+        log.info("【工具执行后】工具名称: {}", toolRequest.name());
+        log.info("【工具执行后】工具执行结果: {}", toolExecution.result());
+        log.info("【工具执行后】工具执行是否失败: {}", toolExecution.hasFailed());
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyAiServiceCompletedListener.java

@@ -0,0 +1,41 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.observability.api.event.AiServiceCompletedEvent;
+import dev.langchain4j.observability.api.listener.AiServiceCompletedListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.time.Instant;
+import java.util.List;
+import java.util.Optional;
+import java.util.UUID;
+
+/**
+ * 自定义的 AiServiceCompletedEvent 的监听器。
+ * 它表示在 AI 服务调用完成时发生的事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyAiServiceCompletedListener implements AiServiceCompletedListener {
+
+    @Override
+    public void onEvent(AiServiceCompletedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        Optional<Object> result = event.result();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        List<Object> aiServiceMethodArgs = invocationContext.methodArguments();
+        Object chatMemoryId = invocationContext.chatMemoryId();
+        Instant eventTimestamp = invocationContext.timestamp();
+
+        log.info("【AI服务完成】调用唯一标识符: {}", invocationId);
+        log.info("【AI服务完成】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【AI服务完成】调用的方法名: {}", aiServiceMethodName);
+        log.info("【AI服务完成】AI服务方法参数: {}", aiServiceMethodArgs);
+        log.info("【AI服务完成】聊天记忆ID: {}", chatMemoryId);
+        log.info("【AI服务完成】调用发生的时间: {}", eventTimestamp);
+        log.info("【AI服务完成】调用结果: {}", result);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyAiServiceErrorListener.java

@@ -0,0 +1,33 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.observability.api.event.AiServiceErrorEvent;
+import dev.langchain4j.observability.api.listener.AiServiceErrorListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.UUID;
+
+/**
+ * 自定义的 AiServiceErrorEvent 的监听器。
+ * 它表示在 AI 服务调用失败时发生的事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyAiServiceErrorListener implements AiServiceErrorListener {
+
+    @Override
+    public void onEvent(AiServiceErrorEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        Throwable error = event.error();
+
+        log.error("【AI服务错误】调用唯一标识符: {}", invocationId);
+        log.error("【AI服务错误】AI服务接口名: {}", aiServiceInterfaceName);
+        log.error("【AI服务错误】调用的方法名: {}", aiServiceMethodName);
+        log.error("【AI服务错误】错误类型: {}", error.getClass().getName());
+        log.error("【AI服务错误】错误信息: {}", error.getMessage(), error);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyAiServiceRequestIssuedListener.java

@@ -0,0 +1,33 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.model.chat.request.ChatRequest;
+import dev.langchain4j.observability.api.event.AiServiceRequestIssuedEvent;
+import dev.langchain4j.observability.api.listener.AiServiceRequestIssuedListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.UUID;
+
+/**
+ * 自定义的 AiServiceRequestIssuedEvent 的监听器。
+ * 它表示在向 LLM 发送请求之前发生的事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyAiServiceRequestIssuedListener implements AiServiceRequestIssuedListener {
+
+    @Override
+    public void onEvent(AiServiceRequestIssuedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        ChatRequest request = event.request();
+
+        log.info("【请求已发出】调用唯一标识符: {}", invocationId);
+        log.info("【请求已发出】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【请求已发出】调用的方法名: {}", aiServiceMethodName);
+        log.info("【请求已发出】发送给LLM的请求: {}", request);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyAiServiceResponseReceivedListener.java

@@ -0,0 +1,37 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.model.chat.request.ChatRequest;
+import dev.langchain4j.model.chat.response.ChatResponse;
+import dev.langchain4j.observability.api.event.AiServiceResponseReceivedEvent;
+import dev.langchain4j.observability.api.listener.AiServiceResponseReceivedListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.UUID;
+
+/**
+ * 自定义的 AiServiceResponseReceivedEvent 的监听器。
+ * 它表示在从 LLM 接收到响应时发生的事件。
+ * 在涉及工具或 guardrail 的单个 AI 服务调用期间，可能会被调用多次。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyAiServiceResponseReceivedListener implements AiServiceResponseReceivedListener {
+
+    @Override
+    public void onEvent(AiServiceResponseReceivedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        ChatRequest request = event.request();
+        ChatResponse response = event.response();
+
+        log.info("【响应已接收】调用唯一标识符: {}", invocationId);
+        log.info("【响应已接收】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【响应已接收】调用的方法名: {}", aiServiceMethodName);
+        log.info("【响应已接收】发送给LLM的请求: {}", request);
+        log.info("【响应已接收】从LLM收到的响应: {}", response);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyAiServiceStartedListener.java

@@ -0,0 +1,38 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.data.message.ChatMessage;
+import dev.langchain4j.data.message.SystemMessage;
+import dev.langchain4j.data.message.UserMessage;
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.observability.api.event.AiServiceStartedEvent;
+import dev.langchain4j.observability.api.listener.AiServiceStartedListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.Optional;
+import java.util.UUID;
+
+/**
+ * 自定义的 AiServiceStartedEvent 的监听器。
+ * 它表示在 AI 服务调用开始时发生的事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyAiServiceStartedListener implements AiServiceStartedListener {
+
+    @Override
+    public void onEvent(AiServiceStartedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        Optional<SystemMessage> systemMessage = event.systemMessage();
+        UserMessage userMessage = event.userMessage();
+
+        log.info("【AI服务启动】调用唯一标识符: {}", invocationId);
+        log.info("【AI服务启动】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【AI服务启动】调用的方法名: {}", aiServiceMethodName);
+        log.info("【AI服务启动】系统消息: {}", systemMessage.orElse(null));
+        log.info("【AI服务启动】用户消息: {}", userMessage);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyChatModelListener.java

@@ -0,0 +1,43 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.model.chat.listener.ChatModelErrorContext;
+import dev.langchain4j.model.chat.listener.ChatModelListener;
+import dev.langchain4j.model.chat.listener.ChatModelRequestContext;
+import dev.langchain4j.model.chat.listener.ChatModelResponseContext;
+import dev.langchain4j.model.chat.request.ChatRequest;
+import dev.langchain4j.model.chat.response.ChatResponse;
+import lombok.extern.slf4j.Slf4j;
+
+/**
+ * 自定义的 ChatModelListener 的监听器。
+ * 它监听 ChatModel 的请求、响应和错误事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyChatModelListener implements ChatModelListener {
+
+    @Override
+    public void onRequest(ChatModelRequestContext requestContext) {
+        ChatRequest request = requestContext.chatRequest();
+        log.info("【ChatModel请求】发送给模型的请求: {}", request);
+        log.info("【ChatModel请求】模型提供商: {}", requestContext.modelProvider());
+    }
+
+    @Override
+    public void onResponse(ChatModelResponseContext responseContext) {
+        ChatRequest request = responseContext.chatRequest();
+        ChatResponse response = responseContext.chatResponse();
+        log.info("【ChatModel响应】原始请求: {}", request);
+        log.info("【ChatModel响应】收到的响应: {}", response);
+        log.info("【ChatModel响应】模型提供商: {}", responseContext.modelProvider());
+    }
+
+    @Override
+    public void onError(ChatModelErrorContext errorContext) {
+        log.error("【ChatModel错误】错误类型: {}", errorContext.error().getClass().getName());
+        log.error("【ChatModel错误】错误信息: {}", errorContext.error().getMessage());
+        log.error("【ChatModel错误】原始请求: {}", errorContext.chatRequest());
+        log.error("【ChatModel错误】模型提供商: {}", errorContext.modelProvider());
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyEmbeddingModelListener.java

@@ -0,0 +1,47 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.Experimental;
+import dev.langchain4j.data.embedding.Embedding;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelErrorContext;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelListener;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelRequestContext;
+import dev.langchain4j.model.embedding.listener.EmbeddingModelResponseContext;
+import dev.langchain4j.model.output.Response;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.List;
+
+/**
+ * 自定义的 EmbeddingModelListener 的监听器。
+ * 它监听 EmbeddingModel 的请求、响应和错误事件。
+ *
+ * @author evo
+ */
+@Slf4j
+@Experimental
+public class MyEmbeddingModelListener implements EmbeddingModelListener {
+
+    @Override
+    public void onRequest(EmbeddingModelRequestContext requestContext) {
+        log.info("【EmbeddingModel请求】输入文本段落数量: {}", requestContext.textSegments().size());
+        log.info("【EmbeddingModel请求】嵌入模型: {}", requestContext.embeddingModel());
+    }
+
+    @Override
+    public void onResponse(EmbeddingModelResponseContext responseContext) {
+        Response<List<Embedding>> response = responseContext.response();
+        List<Embedding> embeddings = response.content();
+        log.info("【EmbeddingModel响应】嵌入向量数量: {}", embeddings.size());
+        log.info("【EmbeddingModel响应】嵌入维度: {}", embeddings.isEmpty() ? 0 : embeddings.get(0).dimension());
+        log.info("【EmbeddingModel响应】嵌入模型: {}", responseContext.embeddingModel());
+        log.info("【EmbeddingModel响应】输入文本段落: {}", responseContext.textSegments());
+    }
+
+    @Override
+    public void onError(EmbeddingModelErrorContext errorContext) {
+        log.error("【EmbeddingModel错误】错误类型: {}", errorContext.error().getClass().getName());
+        log.error("【EmbeddingModel错误】错误信息: {}", errorContext.error().getMessage());
+        log.error("【EmbeddingModel错误】输入文本段落数量: {}", errorContext.textSegments().size());
+        log.error("【EmbeddingModel错误】嵌入模型: {}", errorContext.embeddingModel());
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyInputGuardrailExecutedListener.java

@@ -0,0 +1,45 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.data.message.UserMessage;
+import dev.langchain4j.guardrail.InputGuardrail;
+import dev.langchain4j.guardrail.InputGuardrailRequest;
+import dev.langchain4j.guardrail.InputGuardrailResult;
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.observability.api.event.InputGuardrailExecutedEvent;
+import dev.langchain4j.observability.api.listener.InputGuardrailExecutedListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.time.Duration;
+import java.util.UUID;
+
+/**
+ * 自定义的 InputGuardrailExecutedEvent 的监听器。
+ * 它表示在输入 guardrail 验证执行时发生的事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyInputGuardrailExecutedListener implements InputGuardrailExecutedListener {
+
+    @Override
+    public void onEvent(InputGuardrailExecutedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        InputGuardrailRequest request = event.request();
+        InputGuardrailResult result = event.result();
+        Class<InputGuardrail> guardrailClass = event.guardrailClass();
+        Duration duration = event.duration();
+        UserMessage rewrittenUserMessage = event.rewrittenUserMessage();
+
+        log.info("【输入Guardrail已执行】调用唯一标识符: {}", invocationId);
+        log.info("【输入Guardrail已执行】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【输入Guardrail已执行】调用的方法名: {}", aiServiceMethodName);
+        log.info("【输入Guardrail已执行】Guardrail类名: {}", guardrailClass.getName());
+        log.info("【输入Guardrail已执行】输入Guardrail请求: {}", request);
+        log.info("【输入Guardrail已执行】输入Guardrail结果: {}", result);
+        log.info("【输入Guardrail已执行】重写后的用户消息: {}", rewrittenUserMessage);
+        log.info("【输入Guardrail已执行】执行耗时: {}ms", duration.toMillis());
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyMcpClientListener.java

@@ -0,0 +1,136 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.mcp.client.McpCallContext;
+import dev.langchain4j.mcp.client.McpClientListener;
+import dev.langchain4j.mcp.client.McpGetPromptResult;
+import dev.langchain4j.mcp.client.McpReadResourceResult;
+import dev.langchain4j.mcp.protocol.McpClientMessage;
+import dev.langchain4j.service.tool.ToolExecutionResult;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.Map;
+
+/**
+ * 自定义的 McpClientListener 的监听器。
+ * 监听 MCP 客户端相关的所有可观测性事件，包括：
+ * <ul>
+ *   <li>MCP 工具执行的开始/成功/错误事件</li>
+ *   <li>MCP 资源读取的开始/成功/错误事件</li>
+ *   <li>MCP 提示词获取的开始/成功/错误事件</li>
+ * </ul>
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyMcpClientListener implements McpClientListener {
+
+    // ==================== 工具执行 ====================
+
+    @Override
+    public void beforeExecuteTool(McpCallContext context) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.info("【MCP工具执行前】调用唯一标识符: {}", invocationContext.invocationId());
+        log.info("【MCP工具执行前】MCP消息ID: {}", message.getId());
+        log.info("【MCP工具执行前】MCP方法: {}", message.method);
+    }
+
+    @Override
+    public void afterExecuteTool(McpCallContext context, ToolExecutionResult result, Map<String, Object> rawResult) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.info("【MCP工具执行后】调用唯一标识符: {}", invocationContext.invocationId());
+        log.info("【MCP工具执行后】MCP消息ID: {}", message.getId());
+        log.info("【MCP工具执行后】MCP方法: {}", message.method);
+        log.info("【MCP工具执行后】工具执行结果: {}", result);
+        log.info("【MCP工具执行后】原始结果: {}", rawResult);
+    }
+
+    @Override
+    public void onExecuteToolError(McpCallContext context, Throwable error) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.error("【MCP工具执行错误】调用唯一标识符: {}", invocationContext.invocationId());
+        log.error("【MCP工具执行错误】MCP消息ID: {}", message.getId());
+        log.error("【MCP工具执行错误】MCP方法: {}", message.method);
+        log.error("【MCP工具执行错误】错误类型: {}", error.getClass().getName());
+        log.error("【MCP工具执行错误】错误信息: {}", error.getMessage(), error);
+    }
+
+    // ==================== 资源读取 ====================
+
+    @Override
+    public void beforeResourceGet(McpCallContext context) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.info("【MCP资源读取前】调用唯一标识符: {}", invocationContext.invocationId());
+        log.info("【MCP资源读取前】MCP消息ID: {}", message.getId());
+        log.info("【MCP资源读取前】MCP方法: {}", message.method);
+    }
+
+    @Override
+    public void afterResourceGet(McpCallContext context, McpReadResourceResult result, Map<String, Object> rawResult) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.info("【MCP资源读取后】调用唯一标识符: {}", invocationContext.invocationId());
+        log.info("【MCP资源读取后】MCP消息ID: {}", message.getId());
+        log.info("【MCP资源读取后】MCP方法: {}", message.method);
+        log.info("【MCP资源读取后】资源内容数量: {}", result.contents() != null ? result.contents().size() : 0);
+        log.info("【MCP资源读取后】原始结果: {}", rawResult);
+    }
+
+    @Override
+    public void onResourceGetError(McpCallContext context, Throwable error) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.error("【MCP资源读取错误】调用唯一标识符: {}", invocationContext.invocationId());
+        log.error("【MCP资源读取错误】MCP消息ID: {}", message.getId());
+        log.error("【MCP资源读取错误】MCP方法: {}", message.method);
+        log.error("【MCP资源读取错误】错误类型: {}", error.getClass().getName());
+        log.error("【MCP资源读取错误】错误信息: {}", error.getMessage(), error);
+    }
+
+    // ==================== 提示词获取 ====================
+
+    @Override
+    public void beforePromptGet(McpCallContext context) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.info("【MCP提示词获取前】调用唯一标识符: {}", invocationContext.invocationId());
+        log.info("【MCP提示词获取前】MCP消息ID: {}", message.getId());
+        log.info("【MCP提示词获取前】MCP方法: {}", message.method);
+    }
+
+    @Override
+    public void afterPromptGet(McpCallContext context, McpGetPromptResult result, Map<String, Object> rawResult) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.info("【MCP提示词获取后】调用唯一标识符: {}", invocationContext.invocationId());
+        log.info("【MCP提示词获取后】MCP消息ID: {}", message.getId());
+        log.info("【MCP提示词获取后】MCP方法: {}", message.method);
+        log.info("【MCP提示词获取后】提示词描述: {}", result.description());
+        log.info("【MCP提示词获取后】提示词消息数量: {}", result.messages() != null ? result.messages().size() : 0);
+        log.info("【MCP提示词获取后】原始结果: {}", rawResult);
+    }
+
+    @Override
+    public void onPromptGetError(McpCallContext context, Throwable error) {
+        InvocationContext invocationContext = context.invocationContext();
+        McpClientMessage message = context.message();
+
+        log.error("【MCP提示词获取错误】调用唯一标识符: {}", invocationContext.invocationId());
+        log.error("【MCP提示词获取错误】MCP消息ID: {}", message.getId());
+        log.error("【MCP提示词获取错误】MCP方法: {}", message.method);
+        log.error("【MCP提示词获取错误】错误类型: {}", error.getClass().getName());
+        log.error("【MCP提示词获取错误】错误信息: {}", error.getMessage(), error);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyOutputGuardrailExecutedListener.java

@@ -0,0 +1,42 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.guardrail.OutputGuardrail;
+import dev.langchain4j.guardrail.OutputGuardrailRequest;
+import dev.langchain4j.guardrail.OutputGuardrailResult;
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.observability.api.event.OutputGuardrailExecutedEvent;
+import dev.langchain4j.observability.api.listener.OutputGuardrailExecutedListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.time.Duration;
+import java.util.UUID;
+
+/**
+ * 自定义的 OutputGuardrailExecutedEvent 的监听器。
+ * 它表示在输出 guardrail 验证执行时发生的事件。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyOutputGuardrailExecutedListener implements OutputGuardrailExecutedListener {
+
+    @Override
+    public void onEvent(OutputGuardrailExecutedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        OutputGuardrailRequest request = event.request();
+        OutputGuardrailResult result = event.result();
+        Class<OutputGuardrail> guardrailClass = event.guardrailClass();
+        Duration duration = event.duration();
+
+        log.info("【输出Guardrail已执行】调用唯一标识符: {}", invocationId);
+        log.info("【输出Guardrail已执行】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【输出Guardrail已执行】调用的方法名: {}", aiServiceMethodName);
+        log.info("【输出Guardrail已执行】Guardrail类名: {}", guardrailClass.getName());
+        log.info("【输出Guardrail已执行】输出Guardrail请求: {}", request);
+        log.info("【输出Guardrail已执行】输出Guardrail结果: {}", result);
+        log.info("【输出Guardrail已执行】执行耗时: {}ms", duration.toMillis());
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyToolExecutedEventListener.java

@@ -0,0 +1,38 @@
+package org.ruoyi.observability;
+
+import dev.langchain4j.agent.tool.ToolExecutionRequest;
+import dev.langchain4j.invocation.InvocationContext;
+import dev.langchain4j.observability.api.event.ToolExecutedEvent;
+import dev.langchain4j.observability.api.listener.ToolExecutedEventListener;
+import lombok.extern.slf4j.Slf4j;
+
+import java.util.UUID;
+
+/**
+ * 自定义的 ToolExecutedEvent 的监听器。
+ * 它表示在工具执行完成后发生的事件。
+ * 在单个 AI 服务调用期间，可能会被调用多次。
+ *
+ * @author evo
+ */
+@Slf4j
+public class MyToolExecutedEventListener implements ToolExecutedEventListener {
+
+    @Override
+    public void onEvent(ToolExecutedEvent event) {
+        InvocationContext invocationContext = event.invocationContext();
+        UUID invocationId = invocationContext.invocationId();
+        String aiServiceInterfaceName = invocationContext.interfaceName();
+        String aiServiceMethodName = invocationContext.methodName();
+        ToolExecutionRequest request = event.request();
+        String resultText = event.resultText();
+
+        log.info("【工具已执行】调用唯一标识符: {}", invocationId);
+        log.info("【工具已执行】AI服务接口名: {}", aiServiceInterfaceName);
+        log.info("【工具已执行】调用的方法名: {}", aiServiceMethodName);
+        log.info("【工具已执行】工具执行请求 ID: {}", request.id());
+        log.info("【工具已执行】工具名称: {}", request.name());
+        log.info("【工具已执行】工具参数: {}", request.arguments());
+        log.info("【工具已执行】工具执行结果: {}", resultText);
+    }
+}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/handler/AgentChatHandler.java

@@ -1,132 +0,0 @@
-//package org.ruoyi.service.chat.handler;
-//
-//import dev.langchain4j.agentic.AgenticServices;
-//import dev.langchain4j.community.model.dashscope.QwenChatModel;
-//import dev.langchain4j.service.tool.ToolProvider;
-//import lombok.RequiredArgsConstructor;
-//import lombok.extern.slf4j.Slf4j;
-//import org.ruoyi.agent.McpAgent;
-//import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
-//import org.ruoyi.common.chat.entity.chat.ChatContext;
-//import org.ruoyi.common.chat.service.chatMessage.IChatMessageService;
-//import org.ruoyi.common.sse.utils.SseMessageUtils;
-//import org.ruoyi.mcp.service.core.ToolProviderFactory;
-//import org.springframework.core.annotation.Order;
-//import org.springframework.stereotype.Component;
-//import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
-//
-//import java.util.ArrayList;
-//import java.util.List;
-//
-///**
-// * Agent 深度思考处理器
-// * <p>
-// * 处理 enableThinking=true 的场景，使用 Agent 进行深度思考和工具调用
-// *
-// * @author ageerle@163.com
-// * @date 2025/12/13
-// */
-//@Slf4j
-//@Component
-//@Order(3)
-//@RequiredArgsConstructor
-//public class AgentChatHandler implements ChatHandler {
-//
-//    private final ToolProviderFactory toolProviderFactory;
-//
-//    @Override
-//    public boolean supports(ChatContext context) {
-//        Boolean enableThinking = context.getChatRequest().getEnableThinking();
-//        return enableThinking != null && enableThinking;
-//    }
-//
-//    @Override
-//    public SseEmitter handle(ChatContext context) {
-//        log.info("处理 Agent 深度思考，用户: {}", context.getUserId());
-//
-//        Long userId = context.getUserId();
-//        String tokenValue = context.getTokenValue();
-//        ChatModelVo chatModelVo = context.getChatModelVo();
-//
-//        try {
-//            // 1. 保存用户消息
-//            String content = extractUserContent(context);
-////            saveChatMessage(context.getChatRequest(), userId, content,
-////                RoleType.USER.getName(), chatModelVo);
-//
-//            // 2. 执行 Agent 任务
-//            String result = doAgent(content, chatModelVo);
-//
-//            // 3. 发送结果并保存
-//            SseMessageUtils.sendMessage(userId, result);
-//            SseMessageUtils.completeConnection(userId, tokenValue);
-//
-////            saveChatMessage(context.getChatRequest(), userId, result,
-////                RoleType.ASSISTANT.getName(), chatModelVo);
-//            // todo 保存消息
-//        } catch (Exception e) {
-//            log.error("Agent 执行失败: {}", e.getMessage(), e);
-//            SseMessageUtils.sendMessage(userId, "Agent 执行失败：" + e.getMessage());
-//            SseMessageUtils.completeConnection(userId, tokenValue);
-//        }
-//
-//        return context.getEmitter();
-//    }
-//
-//    /**
-//     * 执行 Agent 任务
-//     */
-//    private String doAgent(String userMessage, ChatModelVo chatModelVo) {
-//        log.info("执行 Agent 任务，消息: {}", userMessage);
-//
-//        try {
-//            // 1. 加载 LLM 模型
-//            QwenChatModel qwenChatModel = QwenChatModel.builder()
-//                .apiKey(chatModelVo.getApiKey())
-//                .modelName(chatModelVo.getModelName())
-//                .build();
-//
-//            // 2. 获取内置工具
-//            List<Object> builtinTools = toolProviderFactory.getAllBuiltinToolObjects();
-//            List<Object> allTools = new ArrayList<>(builtinTools);
-//            log.debug("加载 {} 个内置工具", builtinTools.size());
-//
-//            // 3. 获取 MCP 工具提供者
-//            ToolProvider mcpToolProvider = toolProviderFactory.getAllEnabledMcpToolsProvider();
-//
-//            // 4. 创建 MCP Agent
-//            var agentBuilder = AgenticServices.agentBuilder(McpAgent.class)
-//                .chatModel(qwenChatModel);
-//
-//            if (!allTools.isEmpty()) {
-//                agentBuilder.tools(allTools.toArray(new Object[0]));
-//            }
-//            if (mcpToolProvider != null) {
-//                agentBuilder.toolProvider(mcpToolProvider);
-//            }
-//
-//            McpAgent mcpAgent = agentBuilder.build();
-//
-//            // 5. 调用 Agent
-//            String result = mcpAgent.callMcpTool(userMessage);
-//            log.info("Agent 执行完成，结果长度: {}", result.length());
-//            return result;
-//
-//        } catch (Exception e) {
-//            log.error("Agent 模式执行失败: {}", e.getMessage(), e);
-//            return "Agent 执行失败: " + e.getMessage();
-//        }
-//    }
-//
-//    /**
-//     * 提取用户消息内容
-//     */
-//    private String extractUserContent(ChatContext context) {
-//        var messages = context.getChatRequest().getMessages();
-//        if (messages != null && !messages.isEmpty()) {
-//            return messages.get(0).getContent();
-//        }
-//        return "";
-//    }
-//
-//}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/handler/ResumeChatHandler.java

@@ -1,53 +0,0 @@
-//package org.ruoyi.service.chat.handler;
-//
-//import lombok.RequiredArgsConstructor;
-//import lombok.extern.slf4j.Slf4j;
-//import org.ruoyi.common.chat.domain.dto.request.ReSumeRunner;
-//import org.ruoyi.common.chat.entity.chat.ChatContext;
-//import org.ruoyi.common.chat.service.workFlow.IWorkFlowStarterService;
-//import org.ruoyi.common.core.utils.ObjectUtils;
-//import org.springframework.core.annotation.Order;
-//import org.springframework.stereotype.Component;
-//import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
-//
-///**
-// * 人机交互恢复处理器
-// * <p>
-// * 处理 isResume=true 的场景，恢复工作流的人机交互
-// *
-// * @author ageerle@163.com
-// * @date 2025/12/13
-// */
-//@Slf4j
-//@Component
-//@Order(1)
-//@RequiredArgsConstructor
-//public class ResumeChatHandler implements ChatHandler {
-//
-//    private final IWorkFlowStarterService workFlowStarterService;
-//
-//    @Override
-//    public boolean supports(ChatContext context) {
-//        Boolean isResume = context.getChatRequest().getIsResume();
-//        return isResume != null && isResume;
-//    }
-//
-//    @Override
-//    public SseEmitter handle(ChatContext context) {
-//        log.info("处理人机交互恢复，用户: {}", context.getUserId());
-//
-//        ReSumeRunner reSumeRunner = context.getChatRequest().getReSumeRunner();
-//        if (ObjectUtils.isEmpty(reSumeRunner)) {
-//            log.warn("人机交互恢复参数为空");
-//            return context.getEmitter();
-//        }
-//
-//        workFlowStarterService.resumeFlow(
-//            reSumeRunner.getRuntimeUuid(),
-//            reSumeRunner.getFeedbackContent(),
-//            context.getEmitter()
-//        );
-//
-//        return context.getEmitter();
-//    }
-//}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/handler/WorkflowChatHandler.java

@@ -1,54 +0,0 @@
-//package org.ruoyi.service.chat.handler;
-//
-//import lombok.RequiredArgsConstructor;
-//import lombok.extern.slf4j.Slf4j;
-//import org.ruoyi.common.chat.base.ThreadContext;
-//import org.ruoyi.common.chat.domain.dto.request.WorkFlowRunner;
-//import org.ruoyi.common.chat.entity.chat.ChatContext;
-//import org.ruoyi.common.chat.service.workFlow.IWorkFlowStarterService;
-//import org.ruoyi.common.core.utils.ObjectUtils;
-//import org.springframework.core.annotation.Order;
-//import org.springframework.stereotype.Component;
-//import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
-//
-///**
-// * 工作流对话处理器
-// * <p>
-// * 处理 enableWorkFlow=true 的场景，启动工作流对话
-// *
-// * @author ageerle@163.com
-// * @date 2025/12/13
-// */
-//@Slf4j
-//@Component
-//@Order(2)
-//@RequiredArgsConstructor
-//public class WorkflowChatHandler implements ChatHandler {
-//
-//    private final IWorkFlowStarterService workFlowStarterService;
-//
-//    @Override
-//    public boolean supports(ChatContext context) {
-//        Boolean enableWorkFlow = context.getChatRequest().getEnableWorkFlow();
-//        return enableWorkFlow != null && enableWorkFlow;
-//    }
-//
-//    @Override
-//    public SseEmitter handle(ChatContext context) {
-//        log.info("处理工作流对话，用户: {}, 会话: {}",
-//            context.getUserId(), context.getChatRequest().getSessionId());
-//
-//        WorkFlowRunner runner = context.getChatRequest().getWorkFlowRunner();
-//        if (ObjectUtils.isEmpty(runner)) {
-//            log.warn("工作流参数为空");
-//            return context.getEmitter();
-//        }
-//
-//        return workFlowStarterService.streaming(
-//            ThreadContext.getCurrentUser(),
-//            runner.getUuid(),
-//            runner.getInputs(),
-//            context.getChatRequest().getSessionId()
-//        );
-//    }
-//}

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/ChatServiceFacade.java

@@ -1,25 +1,53 @@
 package org.ruoyi.service.chat.impl;
 
 import cn.dev33.satoken.stp.StpUtil;
-import dev.langchain4j.data.message.*;
+import cn.hutool.core.util.StrUtil;
+import dev.langchain4j.agentic.AgenticServices;
+import dev.langchain4j.agentic.supervisor.SupervisorAgent;
+import dev.langchain4j.agentic.supervisor.SupervisorResponseStrategy;
+import dev.langchain4j.data.message.AiMessage;
+import dev.langchain4j.data.message.ChatMessage;
+import dev.langchain4j.data.message.UserMessage;
+import dev.langchain4j.mcp.McpToolProvider;
+import dev.langchain4j.mcp.client.DefaultMcpClient;
+import dev.langchain4j.mcp.client.McpClient;
+import dev.langchain4j.mcp.client.transport.McpTransport;
+import dev.langchain4j.mcp.client.transport.stdio.StdioMcpTransport;
 import dev.langchain4j.memory.chat.MessageWindowChatMemory;
 import dev.langchain4j.model.chat.StreamingChatModel;
 import dev.langchain4j.model.chat.response.ChatResponse;
 import dev.langchain4j.model.chat.response.StreamingChatResponseHandler;
+import dev.langchain4j.model.openai.OpenAiChatModel;
+import dev.langchain4j.service.tool.ToolProvider;
 import lombok.RequiredArgsConstructor;
 import lombok.SneakyThrows;
 import lombok.extern.slf4j.Slf4j;
+import org.ruoyi.agent.ChartGenerationAgent;
+import org.ruoyi.agent.SqlAgent;
+import org.ruoyi.agent.WebSearchAgent;
+import org.ruoyi.agent.tool.ExecuteSqlQueryTool;
+import org.ruoyi.agent.tool.QueryAllTablesTool;
+import org.ruoyi.agent.tool.QueryTableSchemaTool;
+import org.ruoyi.common.chat.base.ThreadContext;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
+import org.ruoyi.common.chat.domain.dto.request.ReSumeRunner;
+import org.ruoyi.common.chat.domain.dto.request.WorkFlowRunner;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
 import org.ruoyi.common.chat.enums.RoleType;
 import org.ruoyi.common.chat.service.chat.IChatModelService;
 import org.ruoyi.common.chat.service.chat.IChatService;
+import org.ruoyi.common.chat.service.workFlow.IWorkFlowStarterService;
+import org.ruoyi.common.core.utils.ObjectUtils;
 import org.ruoyi.common.satoken.utils.LoginHelper;
 import org.ruoyi.common.sse.core.SseEmitterManager;
 import org.ruoyi.common.sse.utils.SseMessageUtils;
 import org.ruoyi.domain.bo.vector.QueryVectorBo;
 import org.ruoyi.domain.vo.knowledge.KnowledgeInfoVo;
 import org.ruoyi.factory.ChatServiceFactory;
+import org.ruoyi.mcp.service.core.ToolProviderFactory;
+import org.ruoyi.observability.MyAgentListener;
+import org.ruoyi.observability.MyChatModelListener;
+import org.ruoyi.observability.MyMcpClientListener;
 import org.ruoyi.service.chat.AbstractChatService;
 import org.ruoyi.service.chat.IChatMessageService;
 import org.ruoyi.service.chat.impl.memory.PersistentChatMemoryStore;
@@ -62,12 +90,18 @@ public class ChatServiceFacade implements IChatService {
 
     private final IChatMessageService chatMessageService;
 
+    private final IWorkFlowStarterService workFlowStarterService;
+
+    private final ToolProviderFactory toolProviderFactory;
+
     /**
      * 内存实例缓存，避免同一会话重复创建
      * Key: sessionId, Value: MessageWindowChatMemory实例
      */
     private static final Map<Object, MessageWindowChatMemory> memoryCache = new ConcurrentHashMap<>();
 
+
+
     /**
      * 统一聊天入口 - SSE流式响应
      *
@@ -76,6 +110,11 @@ public class ChatServiceFacade implements IChatService {
      */
     public SseEmitter sseChat(ChatRequest chatRequest) {
 
+        // 4. 具体的服务实现
+        Long userId = LoginHelper.getUserId();
+        String tokenValue = StpUtil.getTokenValue();
+        SseEmitter emitter = sseEmitterManager.connect(userId, tokenValue);
+
         // 1. 根据模型名称查询完整配置
         ChatModelVo chatModelVo = chatModelService.selectModelByName(chatRequest.getModel());
         if (chatModelVo == null) {
@@ -85,14 +124,17 @@ public class ChatServiceFacade implements IChatService {
         // 2. 构建上下文消息列表
         List<ChatMessage> contextMessages = buildContextMessages(chatRequest);
 
-        // 3. 路由服务提供商
+        // 3. 处理特殊聊天模式（工作流、人机交互恢复、思考模式）
+        SseEmitter specialResult = handleSpecialChatModes(chatRequest, contextMessages, chatModelVo, emitter, userId, tokenValue);
+        if (specialResult != null) {
+            return specialResult;
+        }
+
+        // 4. 路由服务提供商
         String providerCode = chatModelVo.getProviderCode();
         log.info("路由到服务提供商: {}, 模型: {}", providerCode, chatRequest.getModel());
         AbstractChatService chatService = chatServiceFactory.getOriginalService(providerCode);
-        // 4. 具体的服务实现
+
-        Long userId = LoginHelper.getUserId();
-        String tokenValue = StpUtil.getTokenValue();
-        SseEmitter emitter = sseEmitterManager.connect(userId, tokenValue);
 
         StreamingChatResponseHandler handler = createResponseHandler(userId, tokenValue,chatRequest);
 
@@ -105,6 +147,144 @@ public class ChatServiceFacade implements IChatService {
         return emitter;
     }
 
+    /**
+     * 处理特殊聊天模式（工作流、人机交互恢复、思考模式）
+     *
+     * @param chatRequest      聊天请求
+     * @param contextMessages  上下文消息列表（可能被修改）
+     * @param chatModelVo      聊天模型配置
+     * @param emitter          SSE发射器
+     * @param userId           用户ID
+     * @param tokenValue       会话令牌
+     * @return 如果需要提前返回则返回SseEmitter，否则返回null
+     */
+    private SseEmitter handleSpecialChatModes(ChatRequest chatRequest, List<ChatMessage> contextMessages,
+                                              ChatModelVo chatModelVo, SseEmitter emitter,
+                                              Long userId, String tokenValue) {
+        // 处理工作流对话
+        if (chatRequest.getEnableWorkFlow()) {
+            log.info("处理工作流对话,会话: {}", chatRequest.getSessionId());
+
+            WorkFlowRunner runner = chatRequest.getWorkFlowRunner();
+            if (ObjectUtils.isEmpty(runner)) {
+                log.warn("工作流参数为空");
+            }
+
+            return workFlowStarterService.streaming(
+                ThreadContext.getCurrentUser(),
+                runner.getUuid(),
+                runner.getInputs(),
+                chatRequest.getSessionId()
+            );
+        }
+
+        // 处理人机交互恢复
+        if (chatRequest.getIsResume()) {
+            log.info("处理人机交互恢复");
+
+            ReSumeRunner reSumeRunner = chatRequest.getReSumeRunner();
+            if (ObjectUtils.isEmpty(reSumeRunner)) {
+                log.warn("人机交互恢复参数为空");
+                return emitter;
+            }
+
+            workFlowStarterService.resumeFlow(
+                reSumeRunner.getRuntimeUuid(),
+                reSumeRunner.getFeedbackContent(),
+                emitter
+            );
+
+            return emitter;
+        }
+
+        // 处理思考模式
+        if (chatRequest.getEnableThinking()) {
+            handleThinkingMode(chatRequest, contextMessages, chatModelVo, userId);
+        }
+
+        return null;
+    }
+
+    /**
+     * 处理思考模式
+     *
+     * @param chatRequest     聊天请求
+     * @param contextMessages 上下文消息列表
+     * @param chatModelVo     聊天模型配置
+     * @param userId          用户ID
+     */
+    private void handleThinkingMode(ChatRequest chatRequest, List<ChatMessage> contextMessages,
+                                    ChatModelVo chatModelVo, Long userId) {
+        // 步骤1: 配置MCP传输层 - 连接到bing-cn-mcp服务器
+        McpTransport transport = new StdioMcpTransport.Builder()
+            .command(List.of("C:\\Program Files\\nodejs\\npx.cmd", "-y", "bing-cn-mcp"))
+            .logEvents(true)
+            .build();
+
+        McpClient mcpClient = new DefaultMcpClient.Builder()
+            .transport(transport)
+            .listener(new MyMcpClientListener())
+            .build();
+
+        ToolProvider toolProvider = McpToolProvider.builder()
+            .mcpClients(List.of(mcpClient))
+            .build();
+
+        // 配置echarts MCP
+        McpTransport transport1 = new StdioMcpTransport.Builder()
+            .command(List.of("C:\\Program Files\\nodejs\\npx.cmd", "-y", "mcp-echarts"))
+            .logEvents(true)
+            .build();
+
+        McpClient mcpClient1 = new DefaultMcpClient.Builder()
+            .transport(transport1)
+            .listener(new MyMcpClientListener())
+            .build();
+
+        ToolProvider toolProvider1 = McpToolProvider.builder()
+            .mcpClients(List.of(mcpClient1))
+            .build();
+
+        // 配置模型
+        OpenAiChatModel plannerModel = OpenAiChatModel.builder()
+            .baseUrl(chatModelVo.getApiHost())
+            .apiKey(chatModelVo.getApiKey())
+            .listeners(List.of(new MyChatModelListener()))
+            .modelName(chatModelVo.getModelName())
+            .build();
+
+        // 构建各Agent
+        SqlAgent sqlAgent = AgenticServices.agentBuilder(SqlAgent.class)
+            .chatModel(plannerModel)
+            .listener(new MyAgentListener())
+            .tools(new QueryAllTablesTool(), new QueryTableSchemaTool(), new ExecuteSqlQueryTool())
+            .build();
+
+        WebSearchAgent searchAgent = AgenticServices.agentBuilder(WebSearchAgent.class)
+            .chatModel(plannerModel)
+            .listener(new MyAgentListener())
+            .toolProvider(toolProvider)
+            .build();
+
+        ChartGenerationAgent chartGenerationAgent = AgenticServices.agentBuilder(ChartGenerationAgent.class)
+            .chatModel(plannerModel)
+            .listener(new MyAgentListener())
+            .toolProvider(toolProvider1)
+            .build();
+
+        // 构建监督者Agent
+        SupervisorAgent supervisor = AgenticServices.supervisorBuilder()
+            .chatModel(plannerModel)
+            .listener(new MyAgentListener())
+            .subAgents(sqlAgent, searchAgent, chartGenerationAgent)
+            .responseStrategy(SupervisorResponseStrategy.LAST)
+            .build();
+
+        // 调用 supervisor
+        String invoke = supervisor.invoke(chatRequest.getContent());
+        log.info("supervisor.invoke() 返回: {}", invoke);
+    }
+
     /**
      * 支持外部 handler 的对话接口（跨模块调用）
      * 同时发送到 SSE 和外部 handler
@@ -179,44 +359,16 @@ public class ChatServiceFacade implements IChatService {
 
     /**
      * 构建上下文消息列表
+
+     * 消息顺序：历史消息 → 当前用户消息（确保 AI 正确理解对话上下文）
      *
      * @param chatRequest 聊天请求
      * @return 上下文消息列表
      */
     private List<ChatMessage> buildContextMessages(ChatRequest chatRequest) {
         List<ChatMessage> messages  = new ArrayList<>();
-        // 构建用户消息
-        UserMessage userMessage = UserMessage.userMessage(chatRequest.getContent());
-        messages.add(userMessage);
 
-        // 从向量库查询相关历史消息
+        // 从数据库查询历史对话消息（放在前面）
-        if (chatRequest.getKnowledgeId() != null) {
-            // 查询知识库信息
-            KnowledgeInfoVo knowledgeInfoVo = knowledgeInfoService.queryById(Long.valueOf(chatRequest.getKnowledgeId()));
-            if (knowledgeInfoVo == null) {
-                log.warn("知识库信息不存在，kid: {}", chatRequest.getKnowledgeId());
-                return messages;
-            }
-
-            // 查询向量模型配置信息
-            ChatModelVo chatModel = chatModelService.selectModelByName(knowledgeInfoVo.getEmbeddingModel());
-            if (chatModel == null) {
-                log.warn("向量模型配置不存在，模型名称: {}", knowledgeInfoVo.getEmbeddingModel());
-                return messages;
-            }
-
-            // 构建向量查询参数
-            QueryVectorBo queryVectorBo = buildQueryVectorBo(chatRequest, knowledgeInfoVo, chatModel);
-
-            // 获取向量查询结果
-            List<String> nearestList = vectorStoreService.getQueryVector(queryVectorBo);
-            for (String prompt : nearestList) {
-                // 知识库内容作为系统上下文添加
-                messages.add( new AiMessage(prompt));
-            }
-        }
-
-        // 从数据库查询历史对话消息
         if (chatRequest.getSessionId() != null) {
             MessageWindowChatMemory memory = createChatMemory(chatRequest.getSessionId());
             if (memory != null) {
@@ -228,6 +380,40 @@ public class ChatServiceFacade implements IChatService {
             }
         }
 
+        // 从向量库查询相关历史消息（知识库内容作为上下文）
+        if (chatRequest.getKnowledgeId() != null) {
+            // 查询知识库信息
+            KnowledgeInfoVo knowledgeInfoVo = knowledgeInfoService.queryById(Long.valueOf(chatRequest.getKnowledgeId()));
+            if (knowledgeInfoVo == null) {
+                log.warn("知识库信息不存在，kid: {}", chatRequest.getKnowledgeId());
+                // 继续添加当前用户消息
+                messages.add(UserMessage.userMessage(chatRequest.getContent()));
+                return messages;
+            }
+
+            // 查询向量模型配置信息
+            ChatModelVo chatModel = chatModelService.selectModelByName(knowledgeInfoVo.getEmbeddingModel());
+            if (chatModel == null) {
+                log.warn("向量模型配置不存在，模型名称: {}", knowledgeInfoVo.getEmbeddingModel());
+                messages.add(UserMessage.userMessage(chatRequest.getContent()));
+                return messages;
+            }
+
+            // 构建向量查询参数
+            QueryVectorBo queryVectorBo = buildQueryVectorBo(chatRequest, knowledgeInfoVo, chatModel);
+
+            // 获取向量查询结果（知识库内容作为系统上下文，放在历史消息之后）
+            List<String> nearestList = vectorStoreService.getQueryVector(queryVectorBo);
+            for (String prompt : nearestList) {
+                // 知识库内容作为系统上下文添加
+                messages.add(new AiMessage(prompt));
+            }
+        }
+
+        // 构建当前用户消息（放在最后）
+        UserMessage userMessage = UserMessage.userMessage(chatRequest.getContent());
+        messages.add(userMessage);
+
         return messages;
     }
 

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/provider/DeepseekServiceImpl.java

@@ -1,22 +1,30 @@
 package org.ruoyi.service.chat.impl.provider;
 
+
 import dev.langchain4j.model.chat.StreamingChatModel;
 import dev.langchain4j.model.openai.OpenAiStreamingChatModel;
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
 import org.ruoyi.enums.ChatModeType;
+import org.ruoyi.observability.ChatModelListenerProvider;
+import org.ruoyi.observability.MyChatModelListener;
 import org.ruoyi.service.chat.AbstractChatService;
 import org.springframework.stereotype.Service;
 
+import java.util.List;
+
 
 /**
- * @Author: xiaoen
+ * Deepseek服务调用
- * @Description: deepseek 服务调用
+ *
- * @Date: Created in 19:12 2026/3/17
+ * @author xiaoen
+ * @date 2026/3/17
  */
 @Service
 @Slf4j
+@RequiredArgsConstructor
 public class DeepseekServiceImpl implements AbstractChatService {
 
     @Override
@@ -25,6 +33,7 @@ public class DeepseekServiceImpl implements AbstractChatService {
             .baseUrl(chatModelVo.getApiHost())
             .apiKey(chatModelVo.getApiKey())
             .modelName(chatModelVo.getModelName())
+            .listeners(List.of(new MyChatModelListener()))
             .returnThinking(chatRequest.getEnableThinking())
             .build();
     }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/provider/OllamaServiceImpl.java

@@ -1,13 +1,19 @@
 package org.ruoyi.service.chat.impl.provider;
 
+
 import dev.langchain4j.model.chat.StreamingChatModel;
 import dev.langchain4j.model.ollama.OllamaStreamingChatModel;
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.enums.ChatModeType;
-import org.ruoyi.service.chat.AbstractChatService;
-import org.springframework.stereotype.Service;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
+import org.ruoyi.enums.ChatModeType;
+import org.ruoyi.observability.ChatModelListenerProvider;
+import org.ruoyi.observability.MyChatModelListener;
+import org.ruoyi.service.chat.AbstractChatService;
+import org.springframework.stereotype.Service;
+
+import java.util.List;
 
 /**
  * OllamaAI服务调用
@@ -17,13 +23,17 @@ import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
  */
 @Service
 @Slf4j
+@RequiredArgsConstructor
 public class OllamaServiceImpl implements AbstractChatService {
 
+    private final ChatModelListenerProvider listenerProvider;
+
     @Override
     public StreamingChatModel buildStreamingChatModel(ChatModelVo chatModelVo, ChatRequest chatRequest) {
         return OllamaStreamingChatModel.builder()
                 .baseUrl(chatModelVo.getApiHost())
                 .modelName(chatModelVo.getModelName())
+                .listeners(List.of(new MyChatModelListener()))
                 .build();
     }
 

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/provider/OpenAIServiceImpl.java

@@ -3,13 +3,18 @@ package org.ruoyi.service.chat.impl.provider;
 
 import dev.langchain4j.model.chat.StreamingChatModel;
 import dev.langchain4j.model.openai.OpenAiStreamingChatModel;
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
 import org.ruoyi.enums.ChatModeType;
+import org.ruoyi.observability.ChatModelListenerProvider;
+import org.ruoyi.observability.MyChatModelListener;
 import org.ruoyi.service.chat.AbstractChatService;
 import org.springframework.stereotype.Service;
 
+import java.util.List;
+
 
 /**
  * OPENAI服务调用
@@ -19,14 +24,16 @@ import org.springframework.stereotype.Service;
  */
 @Service
 @Slf4j
+@RequiredArgsConstructor
 public class OpenAIServiceImpl implements AbstractChatService {
 
     @Override
-    public StreamingChatModel buildStreamingChatModel(ChatModelVo chatModelVo, ChatRequest chatRequest) {
+    public StreamingChatModel buildStreamingChatModel(ChatModelVo chatModelVo,ChatRequest chatRequest) {
         return OpenAiStreamingChatModel.builder()
                 .baseUrl(chatModelVo.getApiHost())
                 .apiKey(chatModelVo.getApiKey())
                 .modelName(chatModelVo.getModelName())
+                .listeners(List.of(new MyChatModelListener()))
                 .returnThinking(chatRequest.getEnableThinking())
                 .build();
     }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/provider/PPIOServiceImpl.java

@@ -1,22 +1,28 @@
 package org.ruoyi.service.chat.impl.provider;
 
+
 import dev.langchain4j.model.chat.StreamingChatModel;
 import dev.langchain4j.model.openai.OpenAiStreamingChatModel;
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
 import org.ruoyi.enums.ChatModeType;
+import org.ruoyi.observability.MyChatModelListener;
 import org.ruoyi.service.chat.AbstractChatService;
 import org.springframework.stereotype.Service;
 
+import java.util.List;
+
 /**
- * OPENAI服务调用
+ * PPIO服务调用
  *
  * @author ageerle@163.com
  * @date 2025/12/13
  */
 @Service
 @Slf4j
+@RequiredArgsConstructor
 public class PPIOServiceImpl implements AbstractChatService {
 
     @Override
@@ -25,6 +31,7 @@ public class PPIOServiceImpl implements AbstractChatService {
                 .baseUrl(chatModelVo.getApiHost())
                 .apiKey(chatModelVo.getApiKey())
                 .modelName(chatModelVo.getModelName())
+                .listeners(List.of(new MyChatModelListener()))
                 .returnThinking(chatRequest.getEnableThinking())
                 .build();
     }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/provider/QianWenChatServiceImpl.java

@@ -1,14 +1,20 @@
 package org.ruoyi.service.chat.impl.provider;
 
+
 import dev.langchain4j.community.model.dashscope.QwenStreamingChatModel;
 import dev.langchain4j.model.chat.StreamingChatModel;
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
 import org.ruoyi.enums.ChatModeType;
+import org.ruoyi.observability.ChatModelListenerProvider;
+import org.ruoyi.observability.MyChatModelListener;
 import org.ruoyi.service.chat.AbstractChatService;
 import org.springframework.stereotype.Service;
 
+import java.util.List;
+
 
 /**
  * qianWenAI服务调用
@@ -18,14 +24,17 @@ import org.springframework.stereotype.Service;
  */
 @Service
 @Slf4j
+@RequiredArgsConstructor
 public class QianWenChatServiceImpl implements AbstractChatService {
 
-    // 添加文档解析的前缀字段
+    private final ChatModelListenerProvider listenerProvider;
+
     @Override
     public StreamingChatModel buildStreamingChatModel(ChatModelVo chatModelVo,ChatRequest chatRequest) {
         return QwenStreamingChatModel.builder()
                 .apiKey(chatModelVo.getApiKey())
                 .modelName(chatModelVo.getModelName())
+                .listeners(List.of(new MyChatModelListener()))
                 .build();
     }
 

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/provider/ZhiPuChatServiceImpl.java

@@ -1,14 +1,19 @@
 package org.ruoyi.service.chat.impl.provider;
 
+
 import dev.langchain4j.community.model.zhipu.ZhipuAiStreamingChatModel;
 import dev.langchain4j.model.chat.StreamingChatModel;
+import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.chat.domain.dto.request.ChatRequest;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
 import org.ruoyi.enums.ChatModeType;
+import org.ruoyi.observability.MyChatModelListener;
 import org.ruoyi.service.chat.AbstractChatService;
 import org.springframework.stereotype.Service;
 
+import java.util.List;
+
 
 /**
  * 智谱AI服务调用
@@ -18,6 +23,7 @@ import org.springframework.stereotype.Service;
  */
 @Service
 @Slf4j
+@RequiredArgsConstructor
 public class ZhiPuChatServiceImpl implements AbstractChatService {
 
     @Override
@@ -25,6 +31,7 @@ public class ZhiPuChatServiceImpl implements AbstractChatService {
         return ZhipuAiStreamingChatModel.builder()
             .apiKey(chatModelVo.getApiKey())
             .model(chatModelVo.getModelName())
+            .listeners(List.of(new MyChatModelListener()))
             .build();
     }
 

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/embed/impl/AliBaiLianBaseEmbedProvider.java

@@ -4,10 +4,11 @@ package org.ruoyi.service.embed.impl;
 import dev.langchain4j.community.model.dashscope.QwenEmbeddingModel;
 import dev.langchain4j.data.embedding.Embedding;
 import dev.langchain4j.data.segment.TextSegment;
+import dev.langchain4j.model.embedding.EmbeddingModel;
 import dev.langchain4j.model.output.Response;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
-import org.springframework.stereotype.Component;
 import org.ruoyi.enums.ModalityType;
+import org.springframework.stereotype.Component;
 
 import java.util.List;
 import java.util.Set;
@@ -20,7 +21,6 @@ import java.util.Set;
 @Component("alibailian")
 public class AliBaiLianBaseEmbedProvider extends OpenAiEmbeddingProvider {
 
-
     private ChatModelVo chatModelVo;
 
     @Override
@@ -35,12 +35,13 @@ public class AliBaiLianBaseEmbedProvider extends OpenAiEmbeddingProvider {
 
     @Override
     public Response<List<Embedding>> embedAll(List<TextSegment> textSegments) {
-        return QwenEmbeddingModel.builder()
+        EmbeddingModel model = QwenEmbeddingModel.builder()
                 .apiKey(chatModelVo.getApiKey())
                 .modelName(chatModelVo.getModelName())
                 .dimension(chatModelVo.getModelDimension())
-                .build()
+                .build();
-                .embedAll(textSegments);
+
+        return model.embedAll(textSegments);
     }
 
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/embed/impl/OllamaEmbeddingProvider.java

@@ -2,6 +2,7 @@ package org.ruoyi.service.embed.impl;
 
 import dev.langchain4j.data.embedding.Embedding;
 import dev.langchain4j.data.segment.TextSegment;
+import dev.langchain4j.model.embedding.EmbeddingModel;
 import dev.langchain4j.model.ollama.OllamaEmbeddingModel;
 import dev.langchain4j.model.output.Response;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
@@ -34,10 +35,11 @@ public class OllamaEmbeddingProvider implements BaseEmbedModelService {
     // ollama不能设置embedding维度，使用milvus时请注意！！创建向量表时需要先设定维度大小
     @Override
     public Response<List<Embedding>> embedAll(List<TextSegment> textSegments) {
-        return OllamaEmbeddingModel.builder()
+        EmbeddingModel model = OllamaEmbeddingModel.builder()
                 .baseUrl(chatModelVo.getApiHost())
                 .modelName(chatModelVo.getModelName())
-                .build()
+                .build();
-                .embedAll(textSegments);
+
+        return model.embedAll(textSegments);
     }
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/embed/impl/OpenAiEmbeddingProvider.java

@@ -2,6 +2,7 @@ package org.ruoyi.service.embed.impl;
 
 import dev.langchain4j.data.embedding.Embedding;
 import dev.langchain4j.data.segment.TextSegment;
+import dev.langchain4j.model.embedding.EmbeddingModel;
 import dev.langchain4j.model.openai.OpenAiEmbeddingModel;
 import dev.langchain4j.model.output.Response;
 import org.ruoyi.common.chat.domain.vo.chat.ChatModelVo;
@@ -33,12 +34,13 @@ public class OpenAiEmbeddingProvider implements BaseEmbedModelService {
 
     @Override
     public Response<List<Embedding>> embedAll(List<TextSegment> textSegments) {
-        return OpenAiEmbeddingModel.builder()
+        EmbeddingModel model = OpenAiEmbeddingModel.builder()
                 .baseUrl(chatModelVo.getApiHost())
                 .apiKey(chatModelVo.getApiKey())
                 .modelName(chatModelVo.getModelName())
                 .dimensions(chatModelVo.getModelDimension())
-                .build()
+                .build();
-                .embedAll(textSegments);
+
+        return model.embedAll(textSegments);
     }
 }

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/knowledge/impl/split/CharacterTextSplitter.java

@@ -1,7 +1,10 @@
 package org.ruoyi.service.knowledge.impl.split;
 
+import lombok.AllArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.core.utils.StringUtils;
+import org.ruoyi.domain.vo.knowledge.KnowledgeInfoVo;
+import org.ruoyi.service.knowledge.IKnowledgeInfoService;
 import org.ruoyi.service.knowledge.TextSplitter;
 import org.springframework.context.annotation.Primary;
 import org.springframework.stereotype.Component;
@@ -13,14 +16,37 @@ import java.util.List;
 @Component
 @Slf4j
 @Primary
+@AllArgsConstructor
 public class CharacterTextSplitter implements TextSplitter {
 
+    private final IKnowledgeInfoService knowledgeInfoService;
+
     @Override
     public List<String> split(String content, String kid) {
-        // 使用默认配置
+        // 默认配置值
         String knowledgeSeparator = "#";
-        int textBlockSize = 10000;
+        int textBlockSize = 1000;
-        int overlapChar = 500;
+        int overlapChar = 50;
+
+        // 根据知识库ID查询配置，覆盖默认值
+        if (StringUtils.isNotBlank(kid)) {
+            try {
+                KnowledgeInfoVo info = knowledgeInfoService.queryById(Long.parseLong(kid));
+                if (info != null) {
+                    if (StringUtils.isNotBlank(info.getSeparator())) {
+                        knowledgeSeparator = info.getSeparator();
+                    }
+                    if (info.getTextBlockSize() != null && info.getTextBlockSize() > 0) {
+                        textBlockSize = info.getTextBlockSize().intValue();
+                    }
+                    if (info.getOverlapChar() != null && info.getOverlapChar() > 0) {
+                        overlapChar = info.getOverlapChar().intValue();
+                    }
+                }
+            } catch (Exception e) {
+                log.warn("查询知识库配置失败，使用默认配置, kid={}", kid, e);
+            }
+        }
 
         List<String> chunkList = new ArrayList<>();
         if (content.contains(knowledgeSeparator) && StringUtils.isNotBlank(knowledgeSeparator)) {

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/knowledge/impl/split/ExcelTextSplitter.java

@@ -3,6 +3,8 @@ package org.ruoyi.service.knowledge.impl.split;
 import lombok.AllArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
 import org.ruoyi.common.core.utils.StringUtils;
+import org.ruoyi.domain.vo.knowledge.KnowledgeInfoVo;
+import org.ruoyi.service.knowledge.IKnowledgeInfoService;
 import org.ruoyi.service.knowledge.TextSplitter;
 import org.springframework.stereotype.Component;
 
@@ -15,13 +17,34 @@ import java.util.List;
 @Slf4j
 public class ExcelTextSplitter implements TextSplitter {
 
+    private final IKnowledgeInfoService knowledgeInfoService;
 
     @Override
     public List<String> split(String content, String kid) {
-        // 使用默认配置
+        // 默认配置
         String knowledgeSeparator = "#";
-        int textBlockSize = 10000;
+        int textBlockSize = 1000;
-        int overlapChar = 500;
+        int overlapChar = 50;
+
+        // 根据知识库ID查询配置，覆盖默认值
+        if (StringUtils.isNotBlank(kid)) {
+            try {
+                KnowledgeInfoVo info = knowledgeInfoService.queryById(Long.parseLong(kid));
+                if (info != null) {
+                    if (StringUtils.isNotBlank(info.getSeparator())) {
+                        knowledgeSeparator = info.getSeparator();
+                    }
+                    if (info.getTextBlockSize() != null && info.getTextBlockSize() > 0) {
+                        textBlockSize = info.getTextBlockSize().intValue();
+                    }
+                    if (info.getOverlapChar() != null && info.getOverlapChar() > 0) {
+                        overlapChar = info.getOverlapChar().intValue();
+                    }
+                }
+            } catch (Exception e) {
+                log.warn("查询知识库配置失败，使用默认配置, kid={}", kid, e);
+            }
+        }
         List<String> chunkList = new ArrayList<>();
         if (content.contains(knowledgeSeparator) && StringUtils.isNotBlank(knowledgeSeparator)) {
             // 按自定义分隔符切分

ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/vector/impl/QdrantVectorStoreStrategy.java

@@ -0,0 +1,204 @@
+package org.ruoyi.service.vector.impl;
+
+import dev.langchain4j.data.document.Metadata;
+import dev.langchain4j.data.embedding.Embedding;
+import dev.langchain4j.data.segment.TextSegment;
+import dev.langchain4j.model.embedding.EmbeddingModel;
+import dev.langchain4j.store.embedding.EmbeddingStore;
+import dev.langchain4j.store.embedding.filter.Filter;
+import dev.langchain4j.store.embedding.filter.MetadataFilterBuilder;
+import dev.langchain4j.store.embedding.qdrant.QdrantEmbeddingStore;
+import io.qdrant.client.QdrantClient;
+import io.qdrant.client.QdrantGrpcClient;
+import io.qdrant.client.grpc.Collections.Distance;
+import io.qdrant.client.grpc.Collections.VectorParams;
+import io.qdrant.client.grpc.JsonWithInt;
+import io.qdrant.client.grpc.Points.DenseVector;
+import io.qdrant.client.grpc.Points.Query;
+import io.qdrant.client.grpc.Points.QueryPoints;
+import io.qdrant.client.grpc.Points.ScoredPoint;
+import io.qdrant.client.grpc.Points.VectorInput;
+import lombok.extern.slf4j.Slf4j;
+import org.ruoyi.common.chat.service.chat.IChatModelService;
+import org.ruoyi.common.core.exception.ServiceException;
+import org.ruoyi.config.VectorStoreProperties;
+import org.ruoyi.domain.bo.vector.QueryVectorBo;
+import org.ruoyi.domain.bo.vector.StoreEmbeddingBo;
+import org.ruoyi.factory.EmbeddingModelFactory;
+import org.springframework.stereotype.Component;
+
+import static io.qdrant.client.VectorInputFactory.vectorInput;
+import static io.qdrant.client.WithPayloadSelectorFactory.enable;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.stream.IntStream;
+
+/**
+ * Qdrant向量库策略实现
+ */
+@Slf4j
+@Component
+public class QdrantVectorStoreStrategy extends AbstractVectorStoreStrategy {
+
+    private static final String VECTOR_STORE_TYPE   = "qdrant";
+    private static final String TEXT_SEGMENT_KEY    = "text_segment";
+    private static final String METADATA_FID_KEY    = "fid";
+    private static final String METADATA_KID_KEY    = "kid";
+    private static final String METADATA_DOC_ID_KEY = "doc_id";
+
+    public QdrantVectorStoreStrategy(VectorStoreProperties vectorStoreProperties,
+                                     IChatModelService chatModelService,
+                                     EmbeddingModelFactory embeddingModelFactory) {
+        super(vectorStoreProperties, embeddingModelFactory, chatModelService);
+    }
+
+    private EmbeddingStore<TextSegment> getQdrantStore(String collectionName) {
+        VectorStoreProperties.Qdrant cfg = vectorStoreProperties.getQdrant();
+        QdrantEmbeddingStore.Builder builder = QdrantEmbeddingStore.builder()
+                .host(cfg.getHost())
+                .port(cfg.getPort())
+                .collectionName(collectionName)
+                .useTls(cfg.isUseTls());
+        if (cfg.getApiKey() != null && !cfg.getApiKey().isEmpty()) {
+            builder.apiKey(cfg.getApiKey());
+        }
+        return builder.build();
+    }
+
+    private QdrantClient buildQdrantClient() {
+        VectorStoreProperties.Qdrant cfg = vectorStoreProperties.getQdrant();
+        QdrantGrpcClient.Builder grpcBuilder = QdrantGrpcClient.newBuilder(cfg.getHost(), cfg.getPort(), cfg.isUseTls());
+        if (cfg.getApiKey() != null && !cfg.getApiKey().isEmpty()) {
+            grpcBuilder.withApiKey(cfg.getApiKey());
+        }
+        return new QdrantClient(grpcBuilder.build());
+    }
+
+    private int getModelDimension(String modelName) {
+        return chatModelService.selectModelByName(modelName).getModelDimension();
+    }
+
+    @Override
+    public String getVectorStoreType() {
+        return VECTOR_STORE_TYPE;
+    }
+
+    @Override
+    public void createSchema(String kid, String modelName) {
+        String collectionName = vectorStoreProperties.getQdrant().getCollectionname() + kid;
+        int dimension = getModelDimension(modelName);
+        try (QdrantClient client = buildQdrantClient()) {
+            Boolean exists = client.collectionExistsAsync(collectionName).get();
+            if (!exists) {
+                VectorParams params = VectorParams.newBuilder()
+                        .setSize(dimension)
+                        .setDistance(Distance.Cosine)
+                        .build();
+                client.createCollectionAsync(collectionName, params).get();
+                log.info("Qdrant集合创建成功: {}, dimension: {}", collectionName, dimension);
+            } else {
+                log.info("Qdrant集合已存在: {}", collectionName);
+            }
+        } catch (Exception e) {
+            log.error("Qdrant集合创建失败: {}", collectionName, e);
+            throw new ServiceException("Qdrant集合创建失败: " + collectionName);
+        }
+    }
+
+    @Override
+    public void storeEmbeddings(StoreEmbeddingBo storeEmbeddingBo) {
+        EmbeddingModel embeddingModel = getEmbeddingModel(storeEmbeddingBo.getEmbeddingModelName());
+        List<String> chunkList = storeEmbeddingBo.getChunkList();
+        List<String> fidList = storeEmbeddingBo.getFids();
+        String kid = storeEmbeddingBo.getKid();
+        String docId = storeEmbeddingBo.getDocId();
+        String collectionName = vectorStoreProperties.getQdrant().getCollectionname() + kid;
+
+        EmbeddingStore<TextSegment> embeddingStore = getQdrantStore(collectionName);
+
+        log.info("Qdrant向量存储条数记录: {}", chunkList.size());
+        long startTime = System.currentTimeMillis();
+
+        IntStream.range(0, chunkList.size()).forEach(i -> {
+            String text = chunkList.get(i);
+            String fid = fidList.get(i);
+            Metadata metadata = new Metadata();
+            metadata.put(METADATA_FID_KEY, fid);
+            metadata.put(METADATA_KID_KEY, kid);
+            metadata.put(METADATA_DOC_ID_KEY, docId);
+            TextSegment textSegment = TextSegment.from(text, metadata);
+            Embedding embedding = embeddingModel.embed(text).content();
+            embeddingStore.add(embedding, textSegment);
+        });
+
+        long endTime = System.currentTimeMillis();
+        log.info("Qdrant向量存储完成消耗时间：{}秒", (endTime - startTime) / 1000);
+    }
+
+    @Override
+    public List<String> getQueryVector(QueryVectorBo queryVectorBo) {
+        EmbeddingModel embeddingModel = getEmbeddingModel(queryVectorBo.getEmbeddingModelName());
+        Embedding queryEmbedding = embeddingModel.embed(queryVectorBo.getQuery()).content();
+        String collectionName = vectorStoreProperties.getQdrant().getCollectionname() + queryVectorBo.getKid();
+
+        List<Float> vector = new ArrayList<>();
+        for (float f : queryEmbedding.vector()) {
+            vector.add(f);
+        }
+
+        try (QdrantClient client = buildQdrantClient()) {
+            QueryPoints request = QueryPoints.newBuilder()
+                    .setCollectionName(collectionName)
+                    .setQuery(Query.newBuilder()
+                            .setNearest(vectorInput(vector))
+                            .build())
+                    .setLimit(queryVectorBo.getMaxResults())
+                    .setWithPayload(enable(true))
+                    .build();
+
+            List<ScoredPoint> results = client.queryAsync(request).get();
+            List<String> resultList = new ArrayList<>();
+            for (ScoredPoint point : results) {
+                JsonWithInt.Value textValue = point.getPayloadMap().get(TEXT_SEGMENT_KEY);
+                if (textValue != null && textValue.hasStringValue()) {
+                    resultList.add(textValue.getStringValue());
+                }
+            }
+            return resultList;
+        } catch (Exception e) {
+            log.error("Qdrant查询失败: {}", collectionName, e);
+            throw new ServiceException("Qdrant向量查询失败");
+        }
+    }
+
+    @Override
+    public void removeById(String id, String modelName) {
+        String collectionName = vectorStoreProperties.getQdrant().getCollectionname() + id;
+        try (QdrantClient client = buildQdrantClient()) {
+            client.deleteCollectionAsync(collectionName).get();
+            log.info("Qdrant成功删除集合: {}", collectionName);
+        } catch (Exception e) {
+            log.error("Qdrant删除集合失败: {}", collectionName, e);
+            throw new ServiceException("失败删除向量数据!");
+        }
+    }
+
+    @Override
+    public void removeByDocId(String docId, String kid) {
+        String collectionName = vectorStoreProperties.getQdrant().getCollectionname() + kid;
+        EmbeddingStore<TextSegment> embeddingStore = getQdrantStore(collectionName);
+        Filter filter = MetadataFilterBuilder.metadataKey(METADATA_DOC_ID_KEY).isEqualTo(docId);
+        embeddingStore.removeAll(filter);
+        log.info("Qdrant成功删除 docId={} 的所有向量数据", docId);
+    }
+
+    @Override
+    public void removeByFid(String fid, String kid) {
+        String collectionName = vectorStoreProperties.getQdrant().getCollectionname() + kid;
+        EmbeddingStore<TextSegment> embeddingStore = getQdrantStore(collectionName);
+        Filter filter = MetadataFilterBuilder.metadataKey(METADATA_FID_KEY).isEqualTo(fid);
+        embeddingStore.removeAll(filter);
+        log.info("Qdrant成功删除 fid={} 的所有向量数据", fid);
+    }
+}