autogen_agentchat.teams#
本模块提供了多种预定义多智能体团队的实现。 每个团队都继承自 BaseGroupChat 类。
- class BaseGroupChat(participants: List[ChatAgent], group_chat_manager_name: str, group_chat_manager_class: type[SequentialRoutedAgent], termination_condition: TerminationCondition | None = None, max_turns: int | None = None, runtime: AgentRuntime | None = None, custom_message_types: List[type[BaseAgentEvent | BaseChatMessage]] | None = None, emit_team_events: bool = False)[源代码]#
基类:
Team
,ABC
,ComponentBase
[BaseModel
]群组聊天团队的基础类。
要实现一个群组聊天团队,首先创建
BaseGroupChatManager
的子类,然后 创建使用该群组聊天管理器的BaseGroupChat
的子类。- component_type: ClassVar[ComponentType] = 'team'#
组件的逻辑类型。
- async load_state(state: Mapping[str, Any]) None [源代码]#
加载外部状态并覆盖当前群聊团队的状态。
通过调用每个参与者和群聊管理器的
agent_load_state()
方法(使用其内部代理ID)来加载状态。 关于状态的预期格式,请参阅save_state()
方法。
- async pause() None [源代码]#
在团队运行时通过直接RPC调用参与者的
on_pause()
方法来暂停它们。注意
这是v0.4.9引入的实验性功能,未来可能会变更或移除。
团队必须初始化后才能暂停。
与终止不同,暂停团队不会导致
run()
或run_stream()
方法返回。它会调用每个参与者的on_pause()
方法,如果参与者未实现该方法, 则不会有任何操作。备注
代理类需要负责处理暂停操作,并确保代理后续可以恢复。 请确保在你的代理类中实现
on_pause()
方法来自定义暂停行为。 默认情况下,代理被调用时不会执行任何操作。- 抛出:
RuntimeError -- 如果团队未初始化。来自参与者的异常当调用它们的
on_pause
实现时 会传播到该方法并抛出。
- async reset() None [源代码]#
将团队及其参与者重置到初始状态。
团队必须停止后才能进行重置。
- 抛出:
RuntimeError -- 如果团队未初始化或当前正在运行。
使用
RoundRobinGroupChat
团队的示例:import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import RoundRobinGroupChat from autogen_ext.models.openai import OpenAIChatCompletionClient async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent("Assistant1", model_client=model_client) agent2 = AssistantAgent("Assistant2", model_client=model_client) termination = MaxMessageTermination(3) team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) stream = team.run_stream(task="Count from 1 to 10, respond one at a time.") async for message in stream: print(message) # Reset the team. await team.reset() stream = team.run_stream(task="Count from 1 to 10, respond one at a time.") async for message in stream: print(message) asyncio.run(main())
- async resume() None [源代码]#
在团队运行且暂停时通过直接RPC调用参与者的
on_resume()
方法来恢复它们。注意
这是v0.4.9引入的实验性功能,未来可能会变更或移除。
团队必须初始化后才能恢复。
与终止和用新任务重启不同,恢复团队不会导致
run()
或run_stream()
方法返回。它会调用每个参与者的on_resume()
方法,如果参与者未实现该方法, 则不会有任何操作。备注
代理类需要负责处理恢复操作,并确保代理从暂停处继续执行。 请确保在你的代理类中实现
on_resume()
方法来自定义恢复行为。- 抛出:
RuntimeError -- 如果团队未初始化。来自参与者的异常当调用它们的
on_resume
方法实现时 会传播到该方法并抛出。
- async run(*, task: str | BaseChatMessage | Sequence[BaseChatMessage] | None = None, cancellation_token: CancellationToken | None = None) TaskResult [源代码]#
运行团队并返回结果。基础实现使用
run_stream()
运行团队,然后返回最终结果。 团队停止后,终止条件会被重置。- 参数:
task (str | BaseChatMessage | Sequence[BaseChatMessage] | None) -- 运行团队的任务。可以是字符串、单个
BaseChatMessage
或BaseChatMessage
列表。cancellation_token (CancellationToken | None) -- 用于立即终止任务的取消令牌。 设置取消令牌可能导致团队处于不一致状态, 并且可能不会重置终止条件。 要优雅地停止团队,请改用
ExternalTermination
。
- Returns:
result -- 任务结果,类型为
TaskResult
。结果包含团队产生的消息和停止原因。
使用
RoundRobinGroupChat
团队的示例:import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import RoundRobinGroupChat from autogen_ext.models.openai import OpenAIChatCompletionClient async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent("Assistant1", model_client=model_client) agent2 = AssistantAgent("Assistant2", model_client=model_client) termination = MaxMessageTermination(3) team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) result = await team.run(task="Count from 1 to 10, respond one at a time.") print(result) # 再次运行团队时不带任务以继续之前的任务。 result = await team.run() print(result) asyncio.run(main())
使用
CancellationToken
取消任务的示例:import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import RoundRobinGroupChat from autogen_core import CancellationToken from autogen_ext.models.openai import OpenAIChatCompletionClient async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent("Assistant1", model_client=model_client) agent2 = AssistantAgent("Assistant2", model_client=model_client) termination = MaxMessageTermination(3) team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) cancellation_token = CancellationToken() # 创建后台运行团队的任务。 run_task = asyncio.create_task( team.run( task="Count from 1 to 10, respond one at a time.", cancellation_token=cancellation_token, ) ) # 等待1秒后取消任务。 await asyncio.sleep(1) cancellation_token.cancel() # 这将引发取消错误。 await run_task asyncio.run(main())
- async run_stream(*, task: str | BaseChatMessage | Sequence[BaseChatMessage] | None = None, cancellation_token: CancellationToken | None = None) AsyncGenerator[BaseAgentEvent | BaseChatMessage | TaskResult, None] [源代码]#
运行团队并生成消息流和最终结果, 最终结果的类型为
TaskResult
,作为流的最后一项。 团队停止后,终止条件会被重置。备注
如果代理产生
ModelClientStreamingChunkEvent
, 该消息将在流中产生,但不会包含在messages
中。- 参数:
task (str | BaseChatMessage | Sequence[BaseChatMessage] | None) -- 运行团队的任务。可以是字符串、单个
BaseChatMessage
或BaseChatMessage
列表。cancellation_token (CancellationToken | None) -- 用于立即终止任务的取消令牌。 设置取消令牌可能导致团队处于不一致状态, 并且可能不会重置终止条件。 要优雅地停止团队,请改用
ExternalTermination
。
- Returns:
stream -- 一个
AsyncGenerator
,产生BaseAgentEvent
、BaseChatMessage
以及最终的TaskResult
结果作为流的最后一项。
使用
RoundRobinGroupChat
团队的示例:import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import RoundRobinGroupChat from autogen_ext.models.openai import OpenAIChatCompletionClient async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent("Assistant1", model_client=model_client) agent2 = AssistantAgent("Assistant2", model_client=model_client) termination = MaxMessageTermination(3) team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) stream = team.run_stream(task="Count from 1 to 10, respond one at a time.") async for message in stream: print(message) # 再次运行团队时不带任务以继续之前的任务。 stream = team.run_stream() async for message in stream: print(message) asyncio.run(main())
使用
CancellationToken
取消任务的示例:import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.ui import Console from autogen_agentchat.teams import RoundRobinGroupChat from autogen_core import CancellationToken from autogen_ext.models.openai import OpenAIChatCompletionClient async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent("Assistant1", model_client=model_client) agent2 = AssistantAgent("Assistant2", model_client=model_client) termination = MaxMessageTermination(3) team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) cancellation_token = CancellationToken() # 创建后台运行团队的任务。 run_task = asyncio.create_task( Console( team.run_stream( task="Count from 1 to 10, respond one at a time.", cancellation_token=cancellation_token, ) ) ) # 等待1秒后取消任务。 await asyncio.sleep(1) cancellation_token.cancel() # 这将引发取消错误。 await run_task asyncio.run(main())
- async save_state() Mapping[str, Any] [源代码]#
保存群聊团队的状态。
通过调用每个参与者和群聊管理器的
agent_save_state()
方法(使用其内部代理ID)来保存状态。 状态以嵌套字典形式返回:一个包含 agent_states 键的字典,该键对应的值是另一个字典,其中代理名称作为键,状态作为值。{ "agent_states": { "agent1": ..., "agent2": ..., "RoundRobinGroupChatManager": ... } }
备注
从 v0.4.9 版本开始,状态使用代理名称作为键而非代理ID,并且 team_id 字段已从状态中移除。 这使得状态可以在不同团队和运行时之间移植。以旧格式保存的状态未来可能与新格式不兼容。
小心
在团队运行时调用
save_state()
可能导致状态不一致并产生意外结果。 建议在团队未运行或停止后调用此方法。
- pydantic model DiGraph[源代码]#
基类:
BaseModel
定义了一个包含节点和边的有向图结构。
GraphFlow
使用此结构来确定执行顺序和条件。警告
这是一个实验性功能,API 将在未来版本中发生变化。
Show JSON schema
{ "title": "DiGraph", "description": "\u5b9a\u4e49\u4e86\u4e00\u4e2a\u5305\u542b\u8282\u70b9\u548c\u8fb9\u7684\u6709\u5411\u56fe\u7ed3\u6784\u3002\n:class:`GraphFlow` \u4f7f\u7528\u6b64\u7ed3\u6784\u6765\u786e\u5b9a\u6267\u884c\u987a\u5e8f\u548c\u6761\u4ef6\u3002\n\n.. warning::\n\n \u8fd9\u662f\u4e00\u4e2a\u5b9e\u9a8c\u6027\u529f\u80fd\uff0cAPI \u5c06\u5728\u672a\u6765\u7248\u672c\u4e2d\u53d1\u751f\u53d8\u5316\u3002", "type": "object", "properties": { "nodes": { "additionalProperties": { "$ref": "#/$defs/DiGraphNode" }, "title": "Nodes", "type": "object" }, "default_start_node": { "anyOf": [ { "type": "string" }, { "type": "null" } ], "default": null, "title": "Default Start Node" } }, "$defs": { "DiGraphEdge": { "description": "\u8868\u793a :class:`DiGraph` \u4e2d\u7684\u6709\u5411\u8fb9\uff0c\u5e26\u6709\u53ef\u9009\u7684\u6267\u884c\u6761\u4ef6\u3002\n\n.. warning::\n\n \u8fd9\u662f\u4e00\u4e2a\u5b9e\u9a8c\u6027\u529f\u80fd\uff0cAPI \u5c06\u5728\u672a\u6765\u7248\u672c\u4e2d\u53d8\u66f4\u3002", "properties": { "target": { "title": "Target", "type": "string" }, "condition": { "anyOf": [ { "type": "string" }, { "type": "null" } ], "default": null, "title": "Condition" } }, "required": [ "target" ], "title": "DiGraphEdge", "type": "object" }, "DiGraphNode": { "description": "\u8868\u793a :class:`DiGraph` \u4e2d\u7684\u8282\u70b9(\u4ee3\u7406)\uff0c\u5305\u542b\u5176\u51fa\u8fb9\u548c\u6fc0\u6d3b\u7c7b\u578b\u3002\n\n.. warning::\n\n \u8fd9\u662f\u4e00\u4e2a\u5b9e\u9a8c\u6027\u529f\u80fd\uff0cAPI \u5c06\u5728\u672a\u6765\u7248\u672c\u4e2d\u53d8\u66f4\u3002", "properties": { "name": { "title": "Name", "type": "string" }, "edges": { "default": [], "items": { "$ref": "#/$defs/DiGraphEdge" }, "title": "Edges", "type": "array" }, "activation": { "default": "all", "enum": [ "all", "any" ], "title": "Activation", "type": "string" } }, "required": [ "name" ], "title": "DiGraphNode", "type": "object" } }, "required": [ "nodes" ] }
- Fields:
default_start_node (str | None)
nodes (Dict[str, autogen_agentchat.teams._group_chat._graph._digraph_group_chat.DiGraphNode])
- field nodes: Dict[str, DiGraphNode] [Required]#
- class DiGraphBuilder[源代码]#
基类:
object
用于构建
GraphFlow
中使用的DiGraph
执行图的流式构建器。警告
这是实验性功能,API 将在未来版本中变更。
该工具提供了一种便捷的编程方式来构建智能体交互图,支持包括以下复杂执行流程:
顺序链式执行
并行扇出
条件分支
带安全出口的循环
图中每个节点代表一个智能体。边定义了智能体间的执行路径,并可选择性地基于消息内容进行条件控制。
该构建器兼容 Graph 运行器,同时支持标准智能体和过滤型智能体。
- 方法说明:
add_node(agent, activation): 向图中添加智能体节点
add_edge(source, target, condition): 连接两个节点,可选添加条件
add_conditional_edges(source, condition_to_target): 从源节点添加多个条件边
set_entry_point(agent): 设置默认起始节点(可选)
build(): 生成验证过的 DiGraph
get_participants(): 返回已添加的智能体列表
- 示例 — 顺序流 A → B → C:
>>> builder = GraphBuilder() >>> builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) >>> builder.add_edge(agent_a, agent_b).add_edge(agent_b, agent_c) >>> team = Graph( ... participants=builder.get_participants(), ... graph=builder.build(), ... termination_condition=MaxMessageTermination(5), ... )
- 示例 — 并行扇出 A → (B, C):
>>> builder = GraphBuilder() >>> builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) >>> builder.add_edge(agent_a, agent_b).add_edge(agent_a, agent_c)
- 示例 — 条件分支 A → B ("yes"), A → C ("no"):
>>> builder = GraphBuilder() >>> builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) >>> builder.add_conditional_edges(agent_a, {"yes": agent_b, "no": agent_c})
- 示例 — 循环: A → B → A ("loop"), B → C ("exit"):
>>> builder = GraphBuilder() >>> builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) >>> builder.add_edge(agent_a, agent_b) >>> builder.add_conditional_edges(agent_b, {"loop": agent_a, "exit": agent_c})
- add_conditional_edges(source: str | ChatAgent, condition_to_target: Dict[str, str | ChatAgent]) DiGraphBuilder [源代码]#
根据条件字符串从源节点添加多条条件边。
- add_edge(source: str | ChatAgent, target: str | ChatAgent, condition: str | None = None) DiGraphBuilder [源代码]#
添加从源节点到目标节点的有向边,可选择性地添加条件。
- add_node(agent: ChatAgent, activation: Literal['all', 'any'] = 'all') DiGraphBuilder [源代码]#
向图中添加节点并注册其智能体。
- set_entry_point(name: str | ChatAgent) DiGraphBuilder [源代码]#
设置图的默认起始节点。
- pydantic model DiGraphEdge[源代码]#
基类:
BaseModel
表示
DiGraph
中的有向边,带有可选的执行条件。警告
这是一个实验性功能,API 将在未来版本中变更。
Show JSON schema
{ "title": "DiGraphEdge", "description": "\u8868\u793a :class:`DiGraph` \u4e2d\u7684\u6709\u5411\u8fb9\uff0c\u5e26\u6709\u53ef\u9009\u7684\u6267\u884c\u6761\u4ef6\u3002\n\n.. warning::\n\n \u8fd9\u662f\u4e00\u4e2a\u5b9e\u9a8c\u6027\u529f\u80fd\uff0cAPI \u5c06\u5728\u672a\u6765\u7248\u672c\u4e2d\u53d8\u66f4\u3002", "type": "object", "properties": { "target": { "title": "Target", "type": "string" }, "condition": { "anyOf": [ { "type": "string" }, { "type": "null" } ], "default": null, "title": "Condition" } }, "required": [ "target" ] }
- Fields:
condition (str | None)
target (str)
- pydantic model DiGraphNode[源代码]#
基类:
BaseModel
表示
DiGraph
中的节点(代理),包含其出边和激活类型。警告
这是一个实验性功能,API 将在未来版本中变更。
Show JSON schema
{ "title": "DiGraphNode", "description": "\u8868\u793a :class:`DiGraph` \u4e2d\u7684\u8282\u70b9(\u4ee3\u7406)\uff0c\u5305\u542b\u5176\u51fa\u8fb9\u548c\u6fc0\u6d3b\u7c7b\u578b\u3002\n\n.. warning::\n\n \u8fd9\u662f\u4e00\u4e2a\u5b9e\u9a8c\u6027\u529f\u80fd\uff0cAPI \u5c06\u5728\u672a\u6765\u7248\u672c\u4e2d\u53d8\u66f4\u3002", "type": "object", "properties": { "name": { "title": "Name", "type": "string" }, "edges": { "default": [], "items": { "$ref": "#/$defs/DiGraphEdge" }, "title": "Edges", "type": "array" }, "activation": { "default": "all", "enum": [ "all", "any" ], "title": "Activation", "type": "string" } }, "$defs": { "DiGraphEdge": { "description": "\u8868\u793a :class:`DiGraph` \u4e2d\u7684\u6709\u5411\u8fb9\uff0c\u5e26\u6709\u53ef\u9009\u7684\u6267\u884c\u6761\u4ef6\u3002\n\n.. warning::\n\n \u8fd9\u662f\u4e00\u4e2a\u5b9e\u9a8c\u6027\u529f\u80fd\uff0cAPI \u5c06\u5728\u672a\u6765\u7248\u672c\u4e2d\u53d8\u66f4\u3002", "properties": { "target": { "title": "Target", "type": "string" }, "condition": { "anyOf": [ { "type": "string" }, { "type": "null" } ], "default": null, "title": "Condition" } }, "required": [ "target" ], "title": "DiGraphEdge", "type": "object" } }, "required": [ "name" ] }
- Fields:
activation (Literal['all', 'any'])
edges (List[autogen_agentchat.teams._group_chat._graph._digraph_group_chat.DiGraphEdge])
name (str)
- field edges: List[DiGraphEdge] = []#
- class GraphFlow(participants: List[ChatAgent], graph: DiGraph, termination_condition: TerminationCondition | None = None, max_turns: int | None = None, runtime: AgentRuntime | None = None, custom_message_types: List[type[BaseAgentEvent | BaseChatMessage]] | None = None)[源代码]#
基类:
BaseGroupChat
,Component
[GraphFlowConfig
]一个按照有向图执行模式运行的群聊团队。
警告
这是一个实验性功能,API 将在未来版本中变更。
该群聊基于有向图 (
DiGraph
) 结构执行代理,支持复杂工作流,包括顺序执行、并行分发、 条件分支、合并模式以及带有显式退出条件的循环。执行顺序由 DiGraph 中定义的边决定。图中每个节点对应一个代理,边定义代理间的消息流向。 节点可配置为在以下情况下激活:
通过边条件支持条件分支,根据聊天历史内容选择下一个代理。只要存在最终退出循环的条件, 就允许循环结构。
备注
使用
DiGraphBuilder
类可轻松创建DiGraph
。它提供了流畅的 API 用于添加节点和边、设置入口点以及验证图结构。 详见DiGraphBuilder
文档。GraphFlow
类设计用于与DiGraphBuilder
配合创建复杂工作流。- 参数:
participants (List[ChatAgent]) -- 群聊中的参与者列表。
termination_condition (TerminationCondition, optional) -- 聊天终止条件。
max_turns (int, optional) -- 强制终止前的最大轮次。
graph (DiGraph) -- 定义节点流向和条件的有向执行图。
- 抛出:
ValueError -- 如果参与者名称不唯一,或图验证失败(如存在无退出条件的循环)。
示例
顺序流: A → B → C
import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import DiGraphBuilder, GraphFlow from autogen_ext.models.openai import OpenAIChatCompletionClient async def main(): # Initialize agents with OpenAI model clients. model_client = OpenAIChatCompletionClient(model="gpt-4.1-nano") agent_a = AssistantAgent("A", model_client=model_client, system_message="You are a helpful assistant.") agent_b = AssistantAgent("B", model_client=model_client, system_message="Translate input to Chinese.") agent_c = AssistantAgent("C", model_client=model_client, system_message="Translate input to English.") # Create a directed graph with sequential flow A -> B -> C. builder = DiGraphBuilder() builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) builder.add_edge(agent_a, agent_b).add_edge(agent_b, agent_c) graph = builder.build() # Create a GraphFlow team with the directed graph. team = GraphFlow( participants=[agent_a, agent_b, agent_c], graph=graph, termination_condition=MaxMessageTermination(5), ) # Run the team and print the events. async for event in team.run_stream(task="Write a short story about a cat."): print(event) asyncio.run(main())
并行分发: A → (B, C)
import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import DiGraphBuilder, GraphFlow from autogen_ext.models.openai import OpenAIChatCompletionClient async def main(): # Initialize agents with OpenAI model clients. model_client = OpenAIChatCompletionClient(model="gpt-4.1-nano") agent_a = AssistantAgent("A", model_client=model_client, system_message="You are a helpful assistant.") agent_b = AssistantAgent("B", model_client=model_client, system_message="Translate input to Chinese.") agent_c = AssistantAgent("C", model_client=model_client, system_message="Translate input to Japanese.") # Create a directed graph with fan-out flow A -> (B, C). builder = DiGraphBuilder() builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) builder.add_edge(agent_a, agent_b).add_edge(agent_a, agent_c) graph = builder.build() # Create a GraphFlow team with the directed graph. team = GraphFlow( participants=[agent_a, agent_b, agent_c], graph=graph, termination_condition=MaxMessageTermination(5), ) # Run the team and print the events. async for event in team.run_stream(task="Write a short story about a cat."): print(event) asyncio.run(main())
条件分支: A → B (若'yes') 或 C (若'no')
import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import DiGraphBuilder, GraphFlow from autogen_ext.models.openai import OpenAIChatCompletionClient async def main(): # Initialize agents with OpenAI model clients. model_client = OpenAIChatCompletionClient(model="gpt-4.1-nano") agent_a = AssistantAgent( "A", model_client=model_client, system_message="Detect if the input is in Chinese. If it is, say 'yes', else say 'no', and nothing else.", ) agent_b = AssistantAgent("B", model_client=model_client, system_message="Translate input to English.") agent_c = AssistantAgent("C", model_client=model_client, system_message="Translate input to Chinese.") # Create a directed graph with conditional branching flow A -> B ("yes"), A -> C ("no"). builder = DiGraphBuilder() builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) builder.add_edge(agent_a, agent_b, condition="yes") builder.add_edge(agent_a, agent_c, condition="no") graph = builder.build() # Create a GraphFlow team with the directed graph. team = GraphFlow( participants=[agent_a, agent_b, agent_c], graph=graph, termination_condition=MaxMessageTermination(5), ) # Run the team and print the events. async for event in team.run_stream(task="AutoGen is a framework for building AI agents."): print(event) asyncio.run(main())
带退出条件的循环: A → B → C (若'APPROVE') 或 A (若'REJECT')
import asyncio from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import MaxMessageTermination from autogen_agentchat.teams import DiGraphBuilder, GraphFlow from autogen_ext.models.openai import OpenAIChatCompletionClient async def main(): # Initialize agents with OpenAI model clients. model_client = OpenAIChatCompletionClient(model="gpt-4.1") agent_a = AssistantAgent( "A", model_client=model_client, system_message="You are a helpful assistant.", ) agent_b = AssistantAgent( "B", model_client=model_client, system_message="Provide feedback on the input, if your feedback has been addressed, " "say 'APPROVE', else say 'REJECT' and provide a reason.", ) agent_c = AssistantAgent( "C", model_client=model_client, system_message="Translate the final product to Korean." ) # Create a loop graph with conditional exit: A -> B -> C ("APPROVE"), B -> A ("REJECT"). builder = DiGraphBuilder() builder.add_node(agent_a).add_node(agent_b).add_node(agent_c) builder.add_edge(agent_a, agent_b) builder.add_conditional_edges(agent_b, {"APPROVE": agent_c, "REJECT": agent_a}) builder.set_entry_point(agent_a) graph = builder.build() # Create a GraphFlow team with the directed graph. team = GraphFlow( participants=[agent_a, agent_b, agent_c], graph=graph, termination_condition=MaxMessageTermination(20), # Max 20 messages to avoid infinite loop. ) # Run the team and print the events. async for event in team.run_stream(task="Write a short poem about AI Agents."): print(event) asyncio.run(main())
- component_config_schema#
GraphFlowConfig
的别名
- class MagenticOneGroupChat(participants: List[ChatAgent], model_client: ChatCompletionClient, *, termination_condition: TerminationCondition | None = None, max_turns: int | None = 20, runtime: AgentRuntime | None = None, max_stalls: int = 3, final_answer_prompt: str = ORCHESTRATOR_FINAL_ANSWER_PROMPT, custom_message_types: List[type[BaseAgentEvent | BaseChatMessage]] | None = None, emit_team_events: bool = False)[源代码]#
基类:
BaseGroupChat
,Component
[MagenticOneGroupChatConfig
]一个由 MagenticOneOrchestrator 管理参与者并运行群聊的团队。
该协调器负责处理对话流程,通过管理参与者间的交互确保任务高效完成。
协调器基于 Magentic-One 架构,这是一个用于解决复杂任务的通用多智能体系统(参见下方参考文献)。
- 参数:
participants (List[ChatAgent]) -- 群聊中的参与者列表。
model_client (ChatCompletionClient) -- 用于生成响应的模型客户端。
termination_condition (TerminationCondition, optional) -- 群聊的终止条件。默认为 None。 若无终止条件,群聊将根据协调器逻辑运行或直到达到最大轮数。
max_turns (int, optional) -- 群聊停止前的最大对话轮数。默认为 20。
max_stalls (int, optional) -- 重新规划前允许的最大停滞次数。默认为 3。
final_answer_prompt (str, optional) -- 用于从团队对话记录生成最终答案的 LLM 提示语。已提供默认值(适用于 GPT-4o 级别模型)。
custom_message_types (List[type[BaseAgentEvent | BaseChatMessage]], optional) -- 将在群聊中使用的自定义消息类型列表。 若使用自定义消息类型或您的智能体会产生自定义消息类型,需在此指定。 请确保自定义消息类型是
BaseAgentEvent
或BaseChatMessage
的子类。emit_team_events (bool, optional) -- 是否通过
BaseGroupChat.run_stream()
发出团队事件。默认为 False。
- 抛出:
ValueError -- 当协调逻辑中进度记录缺失必要字段或下一发言者无效时抛出。
Examples:
含单个助手智能体的 MagenticOneGroupChat:
import asyncio from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import MagenticOneGroupChat from autogen_agentchat.ui import Console async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") assistant = AssistantAgent( "Assistant", model_client=model_client, ) team = MagenticOneGroupChat([assistant], model_client=model_client) await Console(team.run_stream(task="Provide a different proof to Fermat last theorem")) asyncio.run(main())
引用
若在研究中使用了 MagenticOneGroupChat,请引用以下论文:
@article{fourney2024magentic, title={Magentic-one: A generalist multi-agent system for solving complex tasks}, author={Fourney, Adam and Bansal, Gagan and Mozannar, Hussein and Tan, Cheng and Salinas, Eduardo and Niedtner, Friederike and Proebsting, Grace and Bassman, Griffin and Gerrits, Jack and Alber, Jacob and others}, journal={arXiv preprint arXiv:2411.04468}, year={2024} }
- classmethod _from_config(config: MagenticOneGroupChatConfig) Self [源代码]#
从配置对象创建组件的新实例。
- 参数:
config (T) -- 配置对象。
- Returns:
Self -- 组件的新实例。
- _to_config() MagenticOneGroupChatConfig [源代码]#
导出当前组件实例的配置,该配置可用于创建具有相同配置的新组件实例。
- Returns:
T -- 组件的配置。
- component_config_schema#
MagenticOneGroupChatConfig
的别名
- class RoundRobinGroupChat(participants: List[ChatAgent], termination_condition: TerminationCondition | None = None, max_turns: int | None = None, runtime: AgentRuntime | None = None, custom_message_types: List[type[BaseAgentEvent | BaseChatMessage]] | None = None, emit_team_events: bool = False)[源代码]#
基类:
BaseGroupChat
,Component
[RoundRobinGroupChatConfig
]一个运行群聊的团队,参与者以轮询方式轮流向所有人发布消息。
如果团队中只有一个参与者,该参与者将是唯一的发言者。
- 参数:
participants (List[BaseChatAgent]) -- 群聊中的参与者列表。
termination_condition (TerminationCondition, optional) -- 群聊的终止条件。默认为 None。 如果没有终止条件,群聊将无限期运行。
max_turns (int, optional) -- 群聊停止前的最大轮次。默认为 None,表示无限制。
custom_message_types (List[type[BaseAgentEvent | BaseChatMessage]], optional) -- 群聊中使用的自定义消息类型列表。 如果使用自定义消息类型或您的代理生成了自定义消息类型,需要在此指定。 确保您的自定义消息类型是
BaseAgentEvent
或BaseChatMessage
的子类。emit_team_events (bool, optional) -- 是否通过
BaseGroupChat.run_stream()
方法发出团队事件。默认为 False。
- 抛出:
ValueError -- 如果未提供参与者或参与者名称不唯一。
Examples:
包含一个带工具参与者的团队:
import asyncio from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import RoundRobinGroupChat from autogen_agentchat.conditions import TextMentionTermination from autogen_agentchat.ui import Console async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") async def get_weather(location: str) -> str: return f"The weather in {location} is sunny." assistant = AssistantAgent( "Assistant", model_client=model_client, tools=[get_weather], ) termination = TextMentionTermination("TERMINATE") team = RoundRobinGroupChat([assistant], termination_condition=termination) await Console(team.run_stream(task="What's the weather in New York?")) asyncio.run(main())
包含多个参与者的团队:
import asyncio from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import RoundRobinGroupChat from autogen_agentchat.conditions import TextMentionTermination from autogen_agentchat.ui import Console async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent("Assistant1", model_client=model_client) agent2 = AssistantAgent("Assistant2", model_client=model_client) termination = TextMentionTermination("TERMINATE") team = RoundRobinGroupChat([agent1, agent2], termination_condition=termination) await Console(team.run_stream(task="Tell me some jokes.")) asyncio.run(main())
- classmethod _from_config(config: RoundRobinGroupChatConfig) Self [源代码]#
从配置对象创建组件的新实例。
- 参数:
config (T) -- 配置对象。
- Returns:
Self -- 组件的新实例。
- _to_config() RoundRobinGroupChatConfig [源代码]#
导出当前组件实例的配置,该配置可用于创建具有相同配置的新组件实例。
- Returns:
T -- 组件的配置。
- component_config_schema#
RoundRobinGroupChatConfig
的别名
- class SelectorGroupChat(participants: List[ChatAgent], model_client: ChatCompletionClient, *, termination_condition: TerminationCondition | None = None, max_turns: int | None = None, runtime: AgentRuntime | None = None, selector_prompt: str = 'You are in a role play game. The following roles are available:\n{roles}.\nRead the following conversation. Then select the next role from {participants} to play. Only return the role.\n\n{history}\n\nRead the above conversation. Then select the next role from {participants} to play. Only return the role.\n', allow_repeated_speaker: bool = False, max_selector_attempts: int = 3, selector_func: Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], str | None] | Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], Awaitable[str | None]] | None = None, candidate_func: Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], List[str]] | Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], Awaitable[List[str]]] | None = None, custom_message_types: List[type[BaseAgentEvent | BaseChatMessage]] | None = None, emit_team_events: bool = False, model_client_streaming: bool = False, model_context: ChatCompletionContext | None = None)[源代码]#
基类:
BaseGroupChat
,Component
[SelectorGroupChatConfig
]一个拥有参与者的群聊团队,参与者轮流向所有人发布消息, 使用 ChatCompletion 模型在每条消息后选择下一位发言者。
- 参数:
participants (List[ChatAgent]) -- 群聊中的参与者, 必须具有唯一名称且至少两名参与者。
model_client (ChatCompletionClient) -- 用于选择下一位发言者的 ChatCompletion 模型客户端。
termination_condition (TerminationCondition, optional) -- 群聊的终止条件。默认为 None。 没有终止条件时,群聊将无限期运行。
max_turns (int, optional) -- 群聊停止前的最大轮数。默认为 None,表示无限制。
selector_prompt (str, optional) -- 用于选择下一位发言者的提示模板。 可用字段:'{roles}'、'{participants}' 和 '{history}'。 {participants} 是候选发言者名称列表,格式为 ["<name1>", "<name2>", ...]。 {roles} 是候选代理名称和描述的新行分隔列表,每行格式为:"<name> : <description>"。 {history} 是格式化为双新行分隔的名称和消息内容的对话历史,每条消息格式为:"<name> : <message content>"。
allow_repeated_speaker (bool, optional) -- 是否将上一位发言者包含在下轮候选列表中。 默认为 False。模型仍可能选择上一位发言者——若发生此情况将记录警告。
max_selector_attempts (int, optional) -- 使用模型选择发言者的最大尝试次数。默认为 3。 若达到最大尝试次数后模型仍未能选择发言者,将使用上一位发言者(若可用), 否则使用第一位参与者。
selector_func (Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], str | None], Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], Awaitable[str | None]], optional) -- 自定义选择函数, 接收对话历史并返回下一位发言者名称。 若提供,此函数将覆盖模型选择逻辑。 若函数返回 None,则使用模型选择下一位发言者。
candidate_func (Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], List[str]], Callable[[Sequence[BaseAgentEvent | BaseChatMessage]], Awaitable[List[str]]], optional) -- 自定义函数,接收对话历史并返回模型选择下位发言者的过滤候选列表。 若返回空列表或 None,SelectorGroupChat 将抛出 ValueError。 仅当未设置 selector_func 时使用此函数。若设置此函数,allow_repeated_speaker 将被忽略。
custom_message_types (List[type[BaseAgentEvent | BaseChatMessage]], optional) -- 群聊中使用的自定义消息类型列表。 若使用自定义消息类型或代理产生自定义消息类型,需在此指定。 确保自定义消息类型是
BaseAgentEvent
或BaseChatMessage
的子类。emit_team_events (bool, optional) -- 是否通过
BaseGroupChat.run_stream()
发出团队事件。默认为 False。model_client_streaming (bool, optional) -- 是否为模型客户端启用流式传输。(适用于 QwQ 等推理模型)。默认为 False。
model_context (ChatCompletionContext | None, optional) -- 存储和检索
LLMMessage
的模型上下文。 可预加载初始消息。模型上下文中的消息将用于发言者选择。团队重置时将清除初始消息。
- 抛出:
ValueError -- 若参与者少于两名或选择提示无效时抛出。
Examples:
含多个参与者的团队:
import asyncio from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import SelectorGroupChat from autogen_agentchat.conditions import TextMentionTermination from autogen_agentchat.ui import Console async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") async def lookup_hotel(location: str) -> str: return f"Here are some hotels in {location}: hotel1, hotel2, hotel3." async def lookup_flight(origin: str, destination: str) -> str: return f"Here are some flights from {origin} to {destination}: flight1, flight2, flight3." async def book_trip() -> str: return "Your trip is booked!" travel_advisor = AssistantAgent( "Travel_Advisor", model_client, tools=[book_trip], description="Helps with travel planning.", ) hotel_agent = AssistantAgent( "Hotel_Agent", model_client, tools=[lookup_hotel], description="Helps with hotel booking.", ) flight_agent = AssistantAgent( "Flight_Agent", model_client, tools=[lookup_flight], description="Helps with flight booking.", ) termination = TextMentionTermination("TERMINATE") team = SelectorGroupChat( [travel_advisor, hotel_agent, flight_agent], model_client=model_client, termination_condition=termination, ) await Console(team.run_stream(task="Book a 3-day trip to new york.")) asyncio.run(main())
含自定义选择函数的团队:
import asyncio from typing import Sequence from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import SelectorGroupChat from autogen_agentchat.conditions import TextMentionTermination from autogen_agentchat.ui import Console from autogen_agentchat.messages import BaseAgentEvent, BaseChatMessage async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") def check_calculation(x: int, y: int, answer: int) -> str: if x + y == answer: return "Correct!" else: return "Incorrect!" agent1 = AssistantAgent( "Agent1", model_client, description="For calculation", system_message="Calculate the sum of two numbers", ) agent2 = AssistantAgent( "Agent2", model_client, tools=[check_calculation], description="For checking calculation", system_message="Check the answer and respond with 'Correct!' or 'Incorrect!'", ) def selector_func(messages: Sequence[BaseAgentEvent | BaseChatMessage]) -> str | None: if len(messages) == 1 or messages[-1].to_text() == "Incorrect!": return "Agent1" if messages[-1].source == "Agent1": return "Agent2" return None termination = TextMentionTermination("Correct!") team = SelectorGroupChat( [agent1, agent2], model_client=model_client, selector_func=selector_func, termination_condition=termination, ) await Console(team.run_stream(task="What is 1 + 1?")) asyncio.run(main())
含自定义模型上下文的团队:
import asyncio from autogen_core.model_context import BufferedChatCompletionContext from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.conditions import TextMentionTermination from autogen_agentchat.teams import SelectorGroupChat from autogen_agentchat.ui import Console async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") model_context = BufferedChatCompletionContext(buffer_size=5) async def lookup_hotel(location: str) -> str: return f"Here are some hotels in {location}: hotel1, hotel2, hotel3." async def lookup_flight(origin: str, destination: str) -> str: return f"Here are some flights from {origin} to {destination}: flight1, flight2, flight3." async def book_trip() -> str: return "Your trip is booked!" travel_advisor = AssistantAgent( "Travel_Advisor", model_client, tools=[book_trip], description="Helps with travel planning.", ) hotel_agent = AssistantAgent( "Hotel_Agent", model_client, tools=[lookup_hotel], description="Helps with hotel booking.", ) flight_agent = AssistantAgent( "Flight_Agent", model_client, tools=[lookup_flight], description="Helps with flight booking.", ) termination = TextMentionTermination("TERMINATE") team = SelectorGroupChat( [travel_advisor, hotel_agent, flight_agent], model_client=model_client, termination_condition=termination, model_context=model_context, ) await Console(team.run_stream(task="Book a 3-day trip to new york.")) asyncio.run(main())
- classmethod _from_config(config: SelectorGroupChatConfig) Self [源代码]#
从配置对象创建组件的新实例。
- 参数:
config (T) -- 配置对象。
- Returns:
Self -- 组件的新实例。
- component_config_schema#
SelectorGroupChatConfig
的别名
- class Swarm(participants: List[ChatAgent], termination_condition: TerminationCondition | None = None, max_turns: int | None = None, runtime: AgentRuntime | None = None, custom_message_types: List[type[BaseAgentEvent | BaseChatMessage]] | None = None, emit_team_events: bool = False)[源代码]#
基类:
BaseGroupChat
,Component
[SwarmConfig
]一个仅基于交接消息选择下一位发言者的群聊团队。
参与者列表中的第一位成员是初始发言者。 下一位发言者根据当前发言者发送的
HandoffMessage
消息选择。 如果没有发送交接消息,则当前发言者继续发言。- 参数:
participants (List[ChatAgent]) -- 参与群聊的代理列表。列表中的第一个代理是初始发言者。
termination_condition (TerminationCondition, optional) -- 群聊的终止条件。默认为 None。 若无终止条件,群聊将无限运行。
max_turns (int, optional) -- 群聊停止前的最大轮次。默认为 None,表示无限制。
custom_message_types (List[type[BaseAgentEvent | BaseChatMessage]], optional) -- 群聊中使用的自定义消息类型列表。 如果使用自定义消息类型或代理会产生自定义消息类型,需在此指定。 确保自定义消息类型是
BaseAgentEvent
或BaseChatMessage
的子类。emit_team_events (bool, optional) -- 是否通过
BaseGroupChat.run_stream()
发出团队事件。默认为 False。
基础示例:
import asyncio from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import Swarm from autogen_agentchat.conditions import MaxMessageTermination async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent1 = AssistantAgent( "Alice", model_client=model_client, handoffs=["Bob"], system_message="You are Alice and you only answer questions about yourself.", ) agent2 = AssistantAgent( "Bob", model_client=model_client, system_message="You are Bob and your birthday is on 1st January." ) termination = MaxMessageTermination(3) team = Swarm([agent1, agent2], termination_condition=termination) stream = team.run_stream(task="What is bob's birthday?") async for message in stream: print(message) asyncio.run(main())
使用
HandoffTermination
实现人机协作交接:import asyncio from autogen_ext.models.openai import OpenAIChatCompletionClient from autogen_agentchat.agents import AssistantAgent from autogen_agentchat.teams import Swarm from autogen_agentchat.conditions import HandoffTermination, MaxMessageTermination from autogen_agentchat.ui import Console from autogen_agentchat.messages import HandoffMessage async def main() -> None: model_client = OpenAIChatCompletionClient(model="gpt-4o") agent = AssistantAgent( "Alice", model_client=model_client, handoffs=["user"], system_message="You are Alice and you only answer questions about yourself, ask the user for help if needed.", ) termination = HandoffTermination(target="user") | MaxMessageTermination(3) team = Swarm([agent], termination_condition=termination) # 开始对话。 await Console(team.run_stream(task="What is bob's birthday?")) # 根据用户反馈继续。 await Console( team.run_stream( task=HandoffMessage(source="user", target="Alice", content="Bob's birthday is on 1st January.") ) ) asyncio.run(main())
- classmethod _from_config(config: SwarmConfig) Swarm [源代码]#
从配置对象创建组件的新实例。
- 参数:
config (T) -- 配置对象。
- Returns:
Self -- 组件的新实例。
- component_config_schema#
SwarmConfig
的别名