[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f1b300lqgtfmoj":3,"$f250wz8h39yydc":8,"$f1v844swroohou":29,"toc--1713645937-5-5":39},{"description":4,"domain":5,"id":6,"keywords":4,"title":7},"","api.kongjs.com",1,"默认",[9,20,24],{"children":10,"icon":4,"id":11,"linkType":12,"linkUrl":13,"parentId":14,"sort":15,"status":16,"target":17,"title":18,"visible":19},null,"447369447469600768","URL","\u002F","0",0,"1","_self","首页",true,{"children":10,"icon":4,"id":21,"linkType":12,"linkUrl":22,"parentId":14,"sort":15,"status":16,"target":17,"title":23,"visible":19},"447369693859794944","\u002Fnotes","笔记",{"children":10,"icon":4,"id":25,"linkType":26,"linkUrl":27,"parentId":14,"sort":15,"status":16,"target":17,"title":28,"visible":19},"447369977960976384","PAGE","\u002Fabout","关于我",{"allowComment":19,"category":30,"categoryId":31,"collectCount":15,"commentCount":15,"content":34,"contentType":35,"cover":10,"description":4,"fileId":10,"id":36,"likeCount":15,"publishAt":10,"rating":15,"source":10,"sourceUrl":10,"tags":37,"title":38,"viewCount":15,"wordCount":15},{"articleCount":10,"children":10,"cover":10,"description":4,"fileId":10,"icon":4,"id":31,"parentId":14,"slug":32,"sort":15,"title":33},"453164804731838464","ai","人工智能","## 1\\. 文档概述\n\n本文档基于 LangChain4j 框架下的 `AgenticServices` 核心工具类源码，完成全方位代码分析。该类是**智能体体系的统一工厂入口**，核心职责是提供静态工厂方法，用于创建、配置各类智能体服务、工作流智能体、远程通信智能体及声明式智能体系统，是 LangChain4j Agentic 模块的顶层入口类。\n\n**核心定位**：统一封装所有智能体构建能力，整合工作流、编排、远程通信、人工介入、规划调度等全场景智能体创建逻辑，支撑声明式智能体体系的快速构建。\n\n## 2\\. 类基础信息\n\n|项目|详情|\n|---|---|\n|包路径|`dev.langchain4j.agentic`|\n|类修饰符|public final 工具类（私有构造，禁止实例化）|\n|核心职责|提供全场景智能体静态工厂方法、声明式智能体系统构建、智能体配置与扩展、工作流加载与自定义|\n|设计模式|静态工厂模式、建造者模式、服务提供者模式\\(SPI\\)、枚举单例模式|\n\n## 3\\. 核心结构拆解\n\n### 3\\.1 私有构造方法\n\n类构造方法私有化，强制作为工具类使用，禁止外部实例化，符合静态工具类设计规范。\n\n### 3\\.2 内部枚举：WorkflowBuilderProvider\n\n基于枚举实现**线程安全单例**，用于加载、缓存、自定义全局工作流构建器 `WorkflowAgentsBuilder`，核心能力：\n\n- **SPI 自动加载**：通过 `ServiceLoader` 加载自定义工作流构建器，无自定义实现时使用默认 `WorkflowAgentsBuilderImpl`\n\n- **动态覆盖**：支持通过外部API手动设置工作流构建器，实现业务自定义扩展\n\n- **全局缓存**：单例缓存构建器实例，避免重复加载资源\n\n### 3\\.3 内部辅助实体类\n\n封装智能体构建上下文与配置参数，简化复杂构建逻辑传参：\n\n1. **DeclarativeAgentCreationContext**：声明式智能体创建上下文，存储智能体服务类、智能体构建器实例\n\n2. **DefaultDeclarativeAgentCreationContext**：上述接口的默认只读实现（Record 不可变结构体）\n\n3. **AgentConfigurator**：智能体全局配置器，封装四大扩展能力：智能体配置回调、子智能体解析器、智能体实例工厂、默认内存ID供给器，支持空默认实例\n\n### 3\\.4 内部工具封装类\n\n内置两个函数式内部类，用于快速封装自定义智能体行为：\n\n- **AgentAction**：封装无参可执行任务，适配简单智能体动作，捕获异常并包装为运行时异常\n\n- **AgenticScopeAction \u002F AgenticScopeFunction**：封装基于 `AgenticScope` 上下文的消费逻辑与函数逻辑，支持上下文参数传递\n\n## 4\\. 核心对外API能力分类\n\n所有对外能力均为 `public static` 方法，按场景可分为**基础智能体构建、工作流智能体构建、高阶智能体构建、远程通信智能体、声明式系统构建**五大类。\n\n### 4\\.1 基础智能体构建API\n\n用于创建通用基础智能体构建器，支持泛型强类型与无类型两种场景：\n\n- `agentBuilder()`：创建无类型智能体构建器 `UntypedAgentBuilder`\n\n- `agentBuilder(Class\u003CT>)`：创建强类型智能体构建器，绑定指定智能体服务接口\n\n- `humanInTheLoopBuilder()`：创建人工介入智能体构建器，用于人机协同工作流\n\n### 4\\.2 工作流智能体构建API\n\n基于全局 `WorkflowAgentsBuilder` 构建各类编排工作流，全部支持**无类型\u002F强类型**双模式：\n\n|工作流类型|API方法|能力说明|\n|---|---|---|\n|顺序工作流|`sequenceBuilder()` \u002F `sequenceBuilder(Class)`|串行执行多个子智能体|\n|并行工作流|`parallelBuilder()` \u002F `parallelBuilder(Class)`|并行执行多个子智能体|\n|并行映射工作流|`parallelMapperBuilder()` \u002F `parallelMapperBuilder(Class)`|根据集合元素批量创建子智能体并行执行|\n|循环工作流|`loopBuilder()` \u002F `loopBuilder(Class)`|循环执行子智能体，支持最大迭代次数限制|\n|条件工作流|`conditionalBuilder()` \u002F `conditionalBuilder(Class)`|根据上下文条件动态触发对应子智能体|\n\n### 4\\.3 高阶智能体构建API\n\n面向复杂智能体场景，提供调度、规划类高级智能体构建能力：\n\n- **监督者智能体**：`supervisorBuilder()` 支持无类型\u002F强类型，用于管理、调度、监控其他子智能体，适配复杂多智能体系统\n\n- **规划者智能体**：`plannerBuilder()` 支持无类型\u002F强类型，具备任务拆解、规划执行能力，适配复杂任务场景\n\n### 4\\.4 远程通信智能体API\n\n提供 A2A（Agent to Agent）远程通信客户端构建能力，支持跨服务智能体交互：\n\n- `a2aBuilder(String serverUrl)`：创建无类型A2A客户端，绑定远程服务地址\n\n- `a2aBuilder(String serverUrl, Class\u003CT>)`：创建强类型A2A客户端，绑定自定义智能体服务接口\n\n### 4\\.5 声明式智能体系统核心API\n\n本类最核心能力，支持通过**注解声明**快速构建完整智能体系统，提供多权重重载方法，适配不同自定义需求：\n\n1. `createAgenticSystem(Class\u003CT>)`：最简方式，自动读取类注解配置与默认模型\n\n2. `createAgenticSystem(Class\u003CT>, ChatModel)`：指定全局对话模型，覆盖默认配置\n\n3. `createAgenticSystem(Class\u003CT>, AgentConfigurator)`：指定智能体全局配置器，实现自定义扩展\n\n4. `createAgenticSystem(Class\u003CT>, ChatModel, AgentConfigurator)`：全参数自定义，支持模型\\+配置双重定制\n\n## 5\\. 核心内部实现逻辑\n\n### 5\\.1 声明式智能体构建核心流程\n\n`createAgenticSystem` 整体执行链路：\n\n1. **读取默认模型**：通过 `declarativeChatModel` 扫描目标类 `@ChatModelSupplier` 注解，获取自定义对话模型\n\n2. **构建组合智能体**：通过`createComposedAgent` 扫描类方法注解，匹配工作流\u002F高阶智能体类型（Sequence\u002FLoop\u002FConditional\u002FSupervisor等）\n\n3. **分支处理**：\n        \n\n    - 匹配到注解：调用对应 `buildXXXAgent` 方法构建对应类型智能体\n\n    - 未匹配到组合智能体：判断是否为A2A远程客户端，构建远程智能体\n\n    - 普通智能体：通过基础构建器配置并构建\n\n4. **异常兜底**：构建结果为空时，抛出非法参数异常，提示非有效智能体类\n\n### 5\\.2 各类智能体构建逻辑\n\n针对不同注解标记的智能体，提供专属构建逻辑，统一流程规范：\n\n- **通用构建步骤**：读取注解配置（名称、描述、输出Key、错误补偿、子智能体列表）→ 配置实例工厂与内存ID → 组装子智能体 → 构建实例\n\n- **条件智能体特殊逻辑**：扫描 `@ActivationCondition` 注解，绑定条件断言与对应子智能体，实现条件触发机制\n\n- **并行映射特殊逻辑**：绑定集合数据供给器，实现动态批量创建子智能体\n\n- **监督者智能体特殊逻辑**：配置最大调用次数、上下文生成策略、响应策略等专属调度参数\n\n### 5\\.3 子智能体创建逻辑\n\n`createSubagent \u002F createSubagents` 负责批量\u002F单个创建子智能体，支持扩展优先级：\n\n1. 优先使用 `AgentConfigurator` 自定义子智能体解析器\n\n2. 解析器为空时，创建内置智能体执行器\n\n3. 无匹配内置类型时，通过基础构建器构建普通智能体\n\n4. 统一封装为 `AgentExecutor` 执行器，统一调度入口\n\n### 5\\.4 内置智能体执行器创建\n\n`createBuiltInAgentExecutor` 扫描类所有支持的注解类型，覆盖全场景内置智能体：工作流智能体、A2A远程智能体、MCP客户端智能体、人工介入智能体、注册中心智能体、普通静态方法智能体。\n\n### 5\\.5 远程与MCP智能体特殊逻辑\n\n- **A2A客户端**：支持注解硬编码地址 \\+ 动态方法供给地址两种模式，二者不可同时配置；支持自定义客户端配置器、监听器\n\n- **MCP客户端**：强制要求 `@McpClientSupplier` 注解提供客户端实例，绑定工具名称、输入输出参数、异步策略\n\n## 6\\. 关键扩展点\n\n该类预留丰富的扩展接口，支持业务自定义改造：\n\n1. **全局工作流构建器自定义**：通过 `setWorkflowAgentsBuilder` 替换默认工作流实现\n\n2. **智能体全局配置**：通过 `AgentConfigurator` 自定义子智能体解析、实例创建、内存ID生成\n\n3. **动态模型供给**：通过 `@ChatModelSupplier` 注解动态提供对话模型\n\n4. **远程地址动态供给**：通过 `@A2AServerUrlSupplier` 动态提供远程服务地址\n\n5. **自定义监听器**：通过 `@AgentListenerSupplier` 为智能体绑定自定义监听\n\n## 7\\. 异常约束与校验规则\n\n- 禁止同时配置A2A地址注解与动态地址供给方法，否则抛出参数异常\n\n- MCP客户端必须配置客户端供给方法，否则抛出异常\n\n- 同一个智能体类中仅允许存在一个 `@Agent` 静态方法，多方法冲突抛异常\n\n- 非有效智能体类调用构建方法，抛出类型不匹配异常\n\n## 8\\. 核心设计亮点\n\n1. **统一入口**：所有智能体构建能力收敛到单一工具类，降低使用与维护成本\n\n2. **声明式编程**：基于注解实现零配置快速构建复杂智能体工作流，简化开发\n\n3. **全场景覆盖**：支持基础智能体、工作流、调度规划、远程通信、人机协同全场景\n\n4. **高可扩展性**：SPI机制\\+自定义配置器\\+动态供给方法，多层扩展能力\n\n5. **强类型兼容**：所有能力支持无类型快速使用 \\+ 强类型工程化使用，适配不同场景\n\n6. **统一异常处理**：封装底层异常，提供友好的参数校验与异常提示\n\n## 9\\. 总结\n\n`AgenticServices` 是 LangChain4j Agentic 模块的**核心门面类**，整合了智能体创建、工作流编排、远程通信、声明式构建、自定义扩展的全部核心能力。通过静态工厂模式简化调用，通过注解声明降低开发成本，通过SPI与自定义配置预留扩展空间，是构建标准化、可扩展、复杂化智能体系统的基础核心组件。","markdown","453165254168289280",[],"LangChain4j AgenticServices 源码分析文档",{"frontmatter":40,"meta":41,"nodes":119},{},{"toc":42},{"title":4,"searchDepth":43,"depth":43,"links":44},5,[45,49,52,69,88,107,110,113,116],{"id":46,"depth":47,"text":48},"_1",2,"1. 文档概述",{"id":50,"depth":47,"text":51},"_2","2. 类基础信息",{"id":53,"depth":47,"text":54,"children":55},"_3","3. 核心结构拆解",[56,60,63,66],{"id":57,"depth":58,"text":59},"_3-_31",3,"3.1 私有构造方法",{"id":61,"depth":58,"text":62},"_3-_32-workflowbuilderprovider","3.2 内部枚举：WorkflowBuilderProvider",{"id":64,"depth":58,"text":65},"_3-_33","3.3 内部辅助实体类",{"id":67,"depth":58,"text":68},"_3-_34","3.4 内部工具封装类",{"id":70,"depth":47,"text":71,"children":72},"_4-api","4. 核心对外API能力分类",[73,76,79,82,85],{"id":74,"depth":58,"text":75},"_4-api-_41-api","4.1 基础智能体构建API",{"id":77,"depth":58,"text":78},"_4-api-_42-api","4.2 工作流智能体构建API",{"id":80,"depth":58,"text":81},"_4-api-_43-api","4.3 高阶智能体构建API",{"id":83,"depth":58,"text":84},"_4-api-_44-api","4.4 远程通信智能体API",{"id":86,"depth":58,"text":87},"_4-api-_45-api","4.5 声明式智能体系统核心API",{"id":89,"depth":47,"text":90,"children":91},"_5","5. 核心内部实现逻辑",[92,95,98,101,104],{"id":93,"depth":58,"text":94},"_5-_51","5.1 声明式智能体构建核心流程",{"id":96,"depth":58,"text":97},"_5-_52","5.2 各类智能体构建逻辑",{"id":99,"depth":58,"text":100},"_5-_53","5.3 子智能体创建逻辑",{"id":102,"depth":58,"text":103},"_5-_54","5.4 内置智能体执行器创建",{"id":105,"depth":58,"text":106},"_5-_55-mcp","5.5 远程与MCP智能体特殊逻辑",{"id":108,"depth":47,"text":109},"_6","6. 关键扩展点",{"id":111,"depth":47,"text":112},"_7","7. 异常约束与校验规则",{"id":114,"depth":47,"text":115},"_8","8. 核心设计亮点",{"id":117,"depth":47,"text":118},"_9","9. 总结",[120,123,137,143,145,199,201,204,207,209,220,249,251,254,275,277,280,298,300,311,313,316,339,341,351,449,451,454,475,477,480,494,496,503,529,531,533,539,592,594,597,627,629,635,656,658,664,666,684,686,689,734,736,754,756,794,796],[121,122,48],"h2",{"id":46},[124,125,126,127,131,132,136],"p",{},"本文档基于 LangChain4j 框架下的 ",[128,129,130],"code",{},"AgenticServices"," 核心工具类源码，完成全方位代码分析。该类是",[133,134,135],"strong",{},"智能体体系的统一工厂入口","，核心职责是提供静态工厂方法，用于创建、配置各类智能体服务、工作流智能体、远程通信智能体及声明式智能体系统，是 LangChain4j Agentic 模块的顶层入口类。",[124,138,139,142],{},[133,140,141],{},"核心定位","：统一封装所有智能体构建能力，整合工作流、编排、远程通信、人工介入、规划调度等全场景智能体创建逻辑，支撑声明式智能体体系的快速构建。",[121,144,51],{"id":50},[146,147,148,161],"table",{},[149,150,151],"thead",{},[152,153,154,158],"tr",{},[155,156,157],"th",{},"项目",[155,159,160],{},"详情",[162,163,164,175,183,191],"tbody",{},[152,165,166,170],{},[167,168,169],"td",{},"包路径",[167,171,172],{},[128,173,174],{},"dev.langchain4j.agentic",[152,176,177,180],{},[167,178,179],{},"类修饰符",[167,181,182],{},"public final 工具类（私有构造，禁止实例化）",[152,184,185,188],{},[167,186,187],{},"核心职责",[167,189,190],{},"提供全场景智能体静态工厂方法、声明式智能体系统构建、智能体配置与扩展、工作流加载与自定义",[152,192,193,196],{},[167,194,195],{},"设计模式",[167,197,198],{},"静态工厂模式、建造者模式、服务提供者模式(SPI)、枚举单例模式",[121,200,54],{"id":53},[202,203,59],"h3",{"id":57},[124,205,206],{},"类构造方法私有化，强制作为工具类使用，禁止外部实例化，符合静态工具类设计规范。",[202,208,62],{"id":61},[124,210,211,212,215,216,219],{},"基于枚举实现",[133,213,214],{},"线程安全单例","，用于加载、缓存、自定义全局工作流构建器 ",[128,217,218],{},"WorkflowAgentsBuilder","，核心能力：",[221,222,223,237,243],"ul",{},[224,225,226,229,230,233,234],"li",{},[133,227,228],{},"SPI 自动加载","：通过 ",[128,231,232],{},"ServiceLoader"," 加载自定义工作流构建器，无自定义实现时使用默认 ",[128,235,236],{},"WorkflowAgentsBuilderImpl",[224,238,239,242],{},[133,240,241],{},"动态覆盖","：支持通过外部API手动设置工作流构建器，实现业务自定义扩展",[224,244,245,248],{},[133,246,247],{},"全局缓存","：单例缓存构建器实例，避免重复加载资源",[202,250,65],{"id":64},[124,252,253],{},"封装智能体构建上下文与配置参数，简化复杂构建逻辑传参：",[255,256,257,263,269],"ol",{},[224,258,259,262],{},[133,260,261],{},"DeclarativeAgentCreationContext","：声明式智能体创建上下文，存储智能体服务类、智能体构建器实例",[224,264,265,268],{},[133,266,267],{},"DefaultDeclarativeAgentCreationContext","：上述接口的默认只读实现（Record 不可变结构体）",[224,270,271,274],{},[133,272,273],{},"AgentConfigurator","：智能体全局配置器，封装四大扩展能力：智能体配置回调、子智能体解析器、智能体实例工厂、默认内存ID供给器，支持空默认实例",[202,276,68],{"id":67},[124,278,279],{},"内置两个函数式内部类，用于快速封装自定义智能体行为：",[221,281,282,288],{},[224,283,284,287],{},[133,285,286],{},"AgentAction","：封装无参可执行任务，适配简单智能体动作，捕获异常并包装为运行时异常",[224,289,290,293,294,297],{},[133,291,292],{},"AgenticScopeAction \u002F AgenticScopeFunction","：封装基于 ",[128,295,296],{},"AgenticScope"," 上下文的消费逻辑与函数逻辑，支持上下文参数传递",[121,299,71],{"id":70},[124,301,302,303,306,307,310],{},"所有对外能力均为 ",[128,304,305],{},"public static"," 方法，按场景可分为",[133,308,309],{},"基础智能体构建、工作流智能体构建、高阶智能体构建、远程通信智能体、声明式系统构建","五大类。",[202,312,75],{"id":74},[124,314,315],{},"用于创建通用基础智能体构建器，支持泛型强类型与无类型两种场景：",[221,317,318,327,333],{},[224,319,320,323,324],{},[128,321,322],{},"agentBuilder()","：创建无类型智能体构建器 ",[128,325,326],{},"UntypedAgentBuilder",[224,328,329,332],{},[128,330,331],{},"agentBuilder(Class\u003CT>)","：创建强类型智能体构建器，绑定指定智能体服务接口",[224,334,335,338],{},[128,336,337],{},"humanInTheLoopBuilder()","：创建人工介入智能体构建器，用于人机协同工作流",[202,340,78],{"id":77},[124,342,343,344,346,347,350],{},"基于全局 ",[128,345,218],{}," 构建各类编排工作流，全部支持",[133,348,349],{},"无类型\u002F强类型","双模式：",[146,352,353,366],{},[149,354,355],{},[152,356,357,360,363],{},[155,358,359],{},"工作流类型",[155,361,362],{},"API方法",[155,364,365],{},"能力说明",[162,367,368,385,401,417,433],{},[152,369,370,373,382],{},[167,371,372],{},"顺序工作流",[167,374,375,378,379],{},[128,376,377],{},"sequenceBuilder()"," \u002F ",[128,380,381],{},"sequenceBuilder(Class)",[167,383,384],{},"串行执行多个子智能体",[152,386,387,390,398],{},[167,388,389],{},"并行工作流",[167,391,392,378,395],{},[128,393,394],{},"parallelBuilder()",[128,396,397],{},"parallelBuilder(Class)",[167,399,400],{},"并行执行多个子智能体",[152,402,403,406,414],{},[167,404,405],{},"并行映射工作流",[167,407,408,378,411],{},[128,409,410],{},"parallelMapperBuilder()",[128,412,413],{},"parallelMapperBuilder(Class)",[167,415,416],{},"根据集合元素批量创建子智能体并行执行",[152,418,419,422,430],{},[167,420,421],{},"循环工作流",[167,423,424,378,427],{},[128,425,426],{},"loopBuilder()",[128,428,429],{},"loopBuilder(Class)",[167,431,432],{},"循环执行子智能体，支持最大迭代次数限制",[152,434,435,438,446],{},[167,436,437],{},"条件工作流",[167,439,440,378,443],{},[128,441,442],{},"conditionalBuilder()",[128,444,445],{},"conditionalBuilder(Class)",[167,447,448],{},"根据上下文条件动态触发对应子智能体",[202,450,81],{"id":80},[124,452,453],{},"面向复杂智能体场景，提供调度、规划类高级智能体构建能力：",[221,455,456,466],{},[224,457,458,461,462,465],{},[133,459,460],{},"监督者智能体","：",[128,463,464],{},"supervisorBuilder()"," 支持无类型\u002F强类型，用于管理、调度、监控其他子智能体，适配复杂多智能体系统",[224,467,468,461,471,474],{},[133,469,470],{},"规划者智能体",[128,472,473],{},"plannerBuilder()"," 支持无类型\u002F强类型，具备任务拆解、规划执行能力，适配复杂任务场景",[202,476,84],{"id":83},[124,478,479],{},"提供 A2A（Agent to Agent）远程通信客户端构建能力，支持跨服务智能体交互：",[221,481,482,488],{},[224,483,484,487],{},[128,485,486],{},"a2aBuilder(String serverUrl)","：创建无类型A2A客户端，绑定远程服务地址",[224,489,490,493],{},[128,491,492],{},"a2aBuilder(String serverUrl, Class\u003CT>)","：创建强类型A2A客户端，绑定自定义智能体服务接口",[202,495,87],{"id":86},[124,497,498,499,502],{},"本类最核心能力，支持通过",[133,500,501],{},"注解声明","快速构建完整智能体系统，提供多权重重载方法，适配不同自定义需求：",[255,504,505,511,517,523],{},[224,506,507,510],{},[128,508,509],{},"createAgenticSystem(Class\u003CT>)","：最简方式，自动读取类注解配置与默认模型",[224,512,513,516],{},[128,514,515],{},"createAgenticSystem(Class\u003CT>, ChatModel)","：指定全局对话模型，覆盖默认配置",[224,518,519,522],{},[128,520,521],{},"createAgenticSystem(Class\u003CT>, AgentConfigurator)","：指定智能体全局配置器，实现自定义扩展",[224,524,525,528],{},[128,526,527],{},"createAgenticSystem(Class\u003CT>, ChatModel, AgentConfigurator)","：全参数自定义，支持模型+配置双重定制",[121,530,90],{"id":89},[202,532,94],{"id":93},[124,534,535,538],{},[128,536,537],{},"createAgenticSystem"," 整体执行链路：",[255,540,541,554,564,586],{},[224,542,543,229,546,549,550,553],{},[133,544,545],{},"读取默认模型",[128,547,548],{},"declarativeChatModel"," 扫描目标类 ",[128,551,552],{},"@ChatModelSupplier"," 注解，获取自定义对话模型",[224,555,556,559,560,563],{},[133,557,558],{},"构建组合智能体","：通过",[128,561,562],{},"createComposedAgent"," 扫描类方法注解，匹配工作流\u002F高阶智能体类型（Sequence\u002FLoop\u002FConditional\u002FSupervisor等）",[224,565,566,571],{},[124,567,568,461],{},[133,569,570],{},"分支处理",[221,572,573,580,583],{},[224,574,575,576,579],{},"匹配到注解：调用对应 ",[128,577,578],{},"buildXXXAgent"," 方法构建对应类型智能体",[224,581,582],{},"未匹配到组合智能体：判断是否为A2A远程客户端，构建远程智能体",[224,584,585],{},"普通智能体：通过基础构建器配置并构建",[224,587,588,591],{},[133,589,590],{},"异常兜底","：构建结果为空时，抛出非法参数异常，提示非有效智能体类",[202,593,97],{"id":96},[124,595,596],{},"针对不同注解标记的智能体，提供专属构建逻辑，统一流程规范：",[221,598,599,605,615,621],{},[224,600,601,604],{},[133,602,603],{},"通用构建步骤","：读取注解配置（名称、描述、输出Key、错误补偿、子智能体列表）→ 配置实例工厂与内存ID → 组装子智能体 → 构建实例",[224,606,607,610,611,614],{},[133,608,609],{},"条件智能体特殊逻辑","：扫描 ",[128,612,613],{},"@ActivationCondition"," 注解，绑定条件断言与对应子智能体，实现条件触发机制",[224,616,617,620],{},[133,618,619],{},"并行映射特殊逻辑","：绑定集合数据供给器，实现动态批量创建子智能体",[224,622,623,626],{},[133,624,625],{},"监督者智能体特殊逻辑","：配置最大调用次数、上下文生成策略、响应策略等专属调度参数",[202,628,100],{"id":99},[124,630,631,634],{},[128,632,633],{},"createSubagent \u002F createSubagents"," 负责批量\u002F单个创建子智能体，支持扩展优先级：",[255,636,637,643,646,649],{},[224,638,639,640,642],{},"优先使用 ",[128,641,273],{}," 自定义子智能体解析器",[224,644,645],{},"解析器为空时，创建内置智能体执行器",[224,647,648],{},"无匹配内置类型时，通过基础构建器构建普通智能体",[224,650,651,652,655],{},"统一封装为 ",[128,653,654],{},"AgentExecutor"," 执行器，统一调度入口",[202,657,103],{"id":102},[124,659,660,663],{},[128,661,662],{},"createBuiltInAgentExecutor"," 扫描类所有支持的注解类型，覆盖全场景内置智能体：工作流智能体、A2A远程智能体、MCP客户端智能体、人工介入智能体、注册中心智能体、普通静态方法智能体。",[202,665,106],{"id":105},[221,667,668,674],{},[224,669,670,673],{},[133,671,672],{},"A2A客户端","：支持注解硬编码地址 + 动态方法供给地址两种模式，二者不可同时配置；支持自定义客户端配置器、监听器",[224,675,676,679,680,683],{},[133,677,678],{},"MCP客户端","：强制要求 ",[128,681,682],{},"@McpClientSupplier"," 注解提供客户端实例，绑定工具名称、输入输出参数、异步策略",[121,685,109],{"id":108},[124,687,688],{},"该类预留丰富的扩展接口，支持业务自定义改造：",[255,690,691,700,708,716,725],{},[224,692,693,229,696,699],{},[133,694,695],{},"全局工作流构建器自定义",[128,697,698],{},"setWorkflowAgentsBuilder"," 替换默认工作流实现",[224,701,702,229,705,707],{},[133,703,704],{},"智能体全局配置",[128,706,273],{}," 自定义子智能体解析、实例创建、内存ID生成",[224,709,710,229,713,715],{},[133,711,712],{},"动态模型供给",[128,714,552],{}," 注解动态提供对话模型",[224,717,718,229,721,724],{},[133,719,720],{},"远程地址动态供给",[128,722,723],{},"@A2AServerUrlSupplier"," 动态提供远程服务地址",[224,726,727,229,730,733],{},[133,728,729],{},"自定义监听器",[128,731,732],{},"@AgentListenerSupplier"," 为智能体绑定自定义监听",[121,735,112],{"id":111},[221,737,738,741,744,751],{},[224,739,740],{},"禁止同时配置A2A地址注解与动态地址供给方法，否则抛出参数异常",[224,742,743],{},"MCP客户端必须配置客户端供给方法，否则抛出异常",[224,745,746,747,750],{},"同一个智能体类中仅允许存在一个 ",[128,748,749],{},"@Agent"," 静态方法，多方法冲突抛异常",[224,752,753],{},"非有效智能体类调用构建方法，抛出类型不匹配异常",[121,755,115],{"id":114},[255,757,758,764,770,776,782,788],{},[224,759,760,763],{},[133,761,762],{},"统一入口","：所有智能体构建能力收敛到单一工具类，降低使用与维护成本",[224,765,766,769],{},[133,767,768],{},"声明式编程","：基于注解实现零配置快速构建复杂智能体工作流，简化开发",[224,771,772,775],{},[133,773,774],{},"全场景覆盖","：支持基础智能体、工作流、调度规划、远程通信、人机协同全场景",[224,777,778,781],{},[133,779,780],{},"高可扩展性","：SPI机制+自定义配置器+动态供给方法，多层扩展能力",[224,783,784,787],{},[133,785,786],{},"强类型兼容","：所有能力支持无类型快速使用 + 强类型工程化使用，适配不同场景",[224,789,790,793],{},[133,791,792],{},"统一异常处理","：封装底层异常，提供友好的参数校验与异常提示",[121,795,118],{"id":117},[124,797,798,800,801,804],{},[128,799,130],{}," 是 LangChain4j Agentic 模块的",[133,802,803],{},"核心门面类","，整合了智能体创建、工作流编排、远程通信、声明式构建、自定义扩展的全部核心能力。通过静态工厂模式简化调用，通过注解声明降低开发成本，通过SPI与自定义配置预留扩展空间，是构建标准化、可扩展、复杂化智能体系统的基础核心组件。"]