Langgraph - Multi-agent supervisor

本文翻译整理自：Multi-agent supervisor
https://langchain-ai.github.io/langgraph/tutorials/multi_agent/agent_supervisor/

---

一、前言

根据初始研究代理的输出，前面示例 自动路由消息。

我们还可以选择使用LLM来编排不同的代理。

下面，我们将创建一个代理组，其中有一个代理主管来帮助委派任务。

---

为了简化每个代理节点中的代码，我们将使用LangGraph的预构建create_react_agent。这个和其他“高级代理”笔记本旨在展示如何在LangGraph中实现某些设计模式。如果该模式适合您的需求，我们建议将其与文档中其他地方描述的一些其他基本模式相结合，以获得最佳性能。

---

二、设置

首先，让我们安装所需的包并设置我们的API密钥

%%capture --no-stderr
%pip install -U langgraph langchain langchain_openai langchain_experimental langsmith pandas
import getpass
import os

def _set_if_undefined(var: str):
    if not os.environ.get(var):
        os.environ[var] = getpass.getpass(f"Please provide your {var}")

_set_if_undefined("OPENAI_API_KEY")
_set_if_undefined("TAVILY_API_KEY")

设置LangSmith用于LangGraph开发

注册 LangSmith 以快速发现问题并提高 LangGraph 项目的性能。

LangSmith允许您使用跟踪数据来调试、测试和监控使用LangGraph构建的LLM应用程序 - 在此处阅读有关如何开始的更多信息。

---

三、创建工具

对于此示例，您将使一个代理 使用搜索引擎 进行网络研究，并使一个 Agent 创建 Graph。在下面定义他们将使用的工具：

from typing import Annotated

from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_experimental.tools import PythonREPLTool

tavily_tool = TavilySearchResults(max_results=5) 

# This executes code locally, which can be unsafe
python_repl_tool = PythonREPLTool()

API Reference: TavilySearchResults | PythonREPLTool

---

四、Helper Utilities

定义一个我们将用来创建 图中节点的辅助函数——它负责将代理响应转换为人工消息。这很重要，因为这就是我们将如何将它添加到图的全局状态

from langchain_core.messages import HumanMessage

def agent_node(state, agent, name):
    result = agent.invoke(state)
    return {
        "messages": [HumanMessage(content=result["messages"][-1].content, name=name)]
    }

API Reference: HumanMessage

---

创建 Agent Supervisor

它将使用函数调用来选择下一个工作节点或完成处理。

from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_openai import ChatOpenAI
from pydantic import BaseModel
from typing import Literal

members = ["Researcher", "Coder"]
system_prompt = (
    "You are a supervisor tasked with managing a conversation between the"
    " following workers:  {members}. Given the following user request,"
    " respond with the worker to act next. Each worker will perform a"
    " task and respond with their results and status. When finished,"
    " respond with FINISH."
)
# Our team supervisor is an LLM node. It just picks the next agent to process
# and decides when the work is completed
options = ["FINISH"] + members

class routeResponse(BaseModel):
    next: Literal[*options]

prompt = ChatPromptTemplate.from_messages(
    [
        ("system", system_prompt),
        MessagesPlaceholder(variable_name="messages"),
        (
            "system",
            "Given the conversation above, who should act next?"
            " Or should we FINISH? Select one of: {options}",
        ),
    ]
).partial(options=str(options), members=", ".join(members))

llm = ChatOpenAI(model="gpt-4o")

def supervisor_agent(state):
    supervisor_chain = prompt | llm.with_structured_output(routeResponse)
    return supervisor_chain.invoke(state)

API Reference: ChatPromptTemplate | MessagesPlaceholder | ChatOpenAI

---

五、构造 Graph

我们准备开始构建 Graph。下面，使用我们刚刚定义的函数定义状态和工作节点。

import functools
import operator
from typing import Sequence
from typing_extensions import TypedDict

from langchain_core.messages import BaseMessage

from langgraph.graph import END, StateGraph, START
from langgraph.prebuilt import create_react_agent

# The agent state is the input to each node in the graph
class AgentState(TypedDict):
    # The annotation tells the graph that new messages will always be added to the current states
    messages: Annotated[Sequence[BaseMessage], operator.add]
    # The 'next' field indicates where to route to next
    next: str

research_agent = create_react_agent(llm, tools=[tavily_tool])
research_node = functools.partial(agent_node, agent=research_agent, name="Researcher")

# NOTE: THIS PERFORMS ARBITRARY CODE EXECUTION. PROCEED WITH CAUTION
code_agent = create_react_agent(llm, tools=[python_repl_tool])
code_node = functools.partial(agent_node, agent=code_agent, name="Coder")

workflow = StateGraph(AgentState)
workflow.add_node("Researcher", research_node)
workflow.add_node("Coder", code_node)
workflow.add_node("supervisor", supervisor_agent)

API Reference: BaseMessage | END | StateGraph | START | create_react_agent

---

现在连接图中的所有边。

for member in members:
    # We want our workers to ALWAYS "report back" to the supervisor when done
    workflow.add_edge(member, "supervisor")

    
# The supervisor populates the "next" field in the graph state
# which routes to a node or finishes
conditional_map = {k: k for k in members}
conditional_map["FINISH"] = END

workflow.add_conditional_edges("supervisor", lambda x: x["next"], conditional_map)
# Finally, add entrypoint
workflow.add_edge(START, "supervisor")

graph = workflow.compile()

---

六、调起 team

创建 graph 后，我们现在可以调起它，看看它的表现如何！

for s in graph.stream(
    {
        "messages": [
            HumanMessage(content="Code hello world and print it to the terminal")
        ]
    }
):
    if "__end__" not in s:
        print(s)
        print("----")
{'supervisor': {'next': 'Coder'}}

---

----
``````output
Python REPL can execute arbitrary code. Use with caution.
``````output
{'Coder': {'messages': [HumanMessage(content='The code to print "Hello, World!" in the terminal has been executed successfully. Here is the output:\n\n\`\`\`\nHello, World!\n\`\`\`', additional_kwargs={}, response_metadata={}, name='Coder')]}}
----
{'supervisor': {'next': 'FINISH'}}
----

---

for s in graph.stream(
    {"messages": [HumanMessage(content="Write a brief research report on pikas.")]},
    {"recursion_limit": 100},
):
    if "__end__" not in s:
        print(s)
        print("----")
{'supervisor': {'next': 'Researcher'}}
----
{'Researcher': {'messages': [HumanMessage(content='### Research ...s/pikas.pdf)\n7. NatureMapping Foundation - American Pika: [Link](http://naturemappingfoundation.org/natmap/facts/american_pika_712.html)\n8. USDA Forest Service - Conservation Status of Pikas: [Link](https://www.fs.usda.gov/psw/publications/millar/psw_2022_millar002.pdf)', additional_kwargs={}, response_metadata={}, name='Researcher')]}}
----
{'supervisor': {'next': 'Coder'}}
----
{'Coder': {'messages': [HumanMessage(content='### R...(http://naturemappingfoundation.org/natmap/facts/american_pika_712.html)\n8. USDA Forest Service - Conservation Status of Pikas: [Link](https://www.fs.usda.gov/psw/publications/millar/psw_2022_millar002.pdf)', additional_kwargs={}, response_metadata={}, name='Coder')]}}
----
{'supervisor': {'next': 'FINISH'}}
----

---

2024-10-18（五）