就在昨天，langgraph的ultra版本项目开源，命名为langgraph-swarm-py，可以通过 pip install langgraph-swarm 安装。所以swarm ? openai 一脸问号？
他的核心升级是，拓展了多智能体应用，可以让多个具有不同专长的智能体协同工作，并在它们之间传递上下文信息。 其次是内置了智能体之间通信的工具，用户可以根据需求进行定制。
下面的图，是一个系统在订票和订酒店2个场景自动切换的demo。
用readme脚本简单测试了下，Agent切换是通过工具来实现，每个agent会绑定一个handoff_tool到别的agent上。
from langgraph.checkpoint.memory import InMemorySaver
from langgraph.prebuilt import create_react_agent
from langgraph_swarm import create_handoff_tool, create_swarm
model = ZhipuAIChat(provider="zhipuai",
model_name="glm-4-plus",
temperature=0.7)
def add(a: int, b: int) -> int:
"""
add function, return a add b
"""
return a + b
alice = create_react_agent(
model,
[add,
create_handoff_tool(agent_name="Bob")],
prompt="You are Alice, an addition expert.",
name="Alice",
)
bob = create_react_agent(
model,
[create_handoff_tool(agent_name="Alice", description="Transfer to Alice, she can help with math")],
prompt="You are Bob, you speak like a pirate.",
name="Bob",
)
checkpointer = InMemorySaver()
workflow = create_swarm(
[alice, bob],
default_active_agent="Alice"
)
app = workflow.compile(checkpointer=checkpointer)
config = {"configurable": {"thread_id": "1"}}
turn_1 = app.invoke(
{"messages": [{"role": "user", "content": "i'd like to speak to Bob"}]},
config,
)
print(turn_1)
turn_2 = app.invoke(
{"messages": [{"role": "user", "content": "what's 5 + 7?"}]},
config,
)
print(turn_2)