08. Hierarchical Multi-Agent Teams

In this tutorial Hierarchical agent team Let's look at how to configure.

Complex tasks that are difficult to respond with a single agent or a single level supervisor Hierarchical structure Split through, and implement the way each sub-level supervisor manages a worker agent specialized in that area.

This hierarchical approach helps to efficiently solve complex tasks that are difficult for a single worker to handle, if there are too many workers.

This example AutoGen thesis An example of implementing the idea through LangGraph, which organizes two sub-tasks of web research and documentation into different teams, and suggests how to manage the entire process through top and middle level supervisors.

Copy

 Start tracking LangSmith. 
[Project name] 
CH17-LangGraph-Use-Cases 

Copy

# Set up LangSmith tracking. https://smith.langchain.com
# !pip install -qU langchain-teddynote
from langchain_teddynote import logging

# Enter a project name.
logging.langsmith("CH17-LangGraph-Use-Cases")

Copy

 True 

Copy

# Configuration file for managing API keys as environment variables
from dotenv import load_dotenv

# Load API key information
load_dotenv()

Preferences

• AutoGen thesis: Enabling Next-Gen LLM Applications via Multi-Agent Conversation (Wu et al.)

• LangGraph - Multi-Agent concept

Reference

---

1. Tool creation : Define agent tools for web research and documentation

2. Agent team definition : Hierarchically define and organize research teams and documentation teams

3. Add layer : Tiered the entire task through high-level graphs and mid-level supervisors

4. Combination : Integrate all elements to build a final hierarchical agent team

What to cover in this tutorial

---

In this situation, the higher level supervisor is the lower level sub-supervisor Assign tasks to the people, each sub-supervisor You can construct a hierarchical structure where you are reassigned to a team of specialized workers.

• Increased work complexity : A single supervisor may require expertise from various sub-areas that cannot be handled at once.

• Increase in the number of workers : When managing a large number of workers, the administrative burden increases when a single supervisor gives orders directly to all workers.

In the previous Supervisor example, we looked at the process of one supervisor node assigning tasks to multiple worker nodes and gathering results. This method is efficient in simple cases. However, hierarchical structures may be necessary in the following situations:

Why a hierarchical agent team?

Copy

Copyfrom pathlib import Path
from typing import Dict, Optional, List
from typing_extensions import Annotated


# Create a temporary directory and set a working directory
WORKING_DIRECTORY = Path("./tmp")

# If the tmp folder does not exist, create it.
WORKING_DIRECTORY.mkdir(exist_ok=True)

# Create outline and save to file
@tool
def create_outline(
    points: Annotated[List[str], "List of main points or sections."],
    file_name: Annotated[str, "File path to save the outline."],
) -> Annotated[str, "Path of the saved outline file."]:
    """Create and save an outline."""
    # Save outline with given file name
    with (WORKING_DIRECTORY / file_name).open("w") as file:
        for i, point in enumerate(points):
            file.write(f"{i + 1}. {point}\n")
    return f"Outline saved to {file_name}"


# Read the document
@tool
def read_document(
    file_name: Annotated[str, "File path to read the document."],
    start: Annotated[Optional[int], "The start line. Default is 0"] = None,
    end: Annotated[Optional[int], "The end line. Default is None"] = None,
) -> str:
    """Read the specified document."""
    # Read document with given file name
    with (WORKING_DIRECTORY / file_name).open("r") as file:
        lines = file.readlines()
    # Set default if start line is not specified
    if start is not None:
        start = 0
    return "\n".join(lines[start:end])


# Write and save documents
@tool
def write_document(
    content: Annotated[str, "Text content to be written into the document."],
    file_name: Annotated[str, "File path to save the document."],
) -> Annotated[str, "Path of the saved document file."]:
    """Create and save a text document."""
    # Save document with given file name
    with (WORKING_DIRECTORY / file_name).open("w") as file:
        file.write(content)
    return f"Document saved to {file_name}"


# editing documents
@tool
def edit_document(
    file_name: Annotated[str, "File path of the document to be edited."],
    inserts: Annotated[
        Dict[int, str],
        "Dictionary where key is the line number (1-indexed) and value is the text to be inserted at that line.",
    ],
) -> Annotated[str, "File path of the edited document."]:
    """Edit a document by inserting text at specific line numbers."""

    # Read document with given file name
    with (WORKING_DIRECTORY / file_name).open("r") as file:
        lines = file.readlines()

    # Process by aligning the text to be inserted
    sorted_inserts = sorted(inserts.items())

    # Insert text at a specified line number
    for line_number, text in sorted_inserts:
        if 1 <= line_number <= len(lines) + 1:
            lines.insert(line_number - 1, text + "\n")
        else:
            return f"Error: Line number {line_number} is out of range."

    # Save the edited document to a file
    with (WORKING_DIRECTORY / file_name).open("w") as file:
        file.writelines(lines)

    return f"Document edited and saved to {file_name}"

This tool allows agents to access the file system, which may not be safe. Therefore, attention is required to use.

Next, define the tools (file access tools) that the documentation team will use.

Documenting team tools

Copy

 USER_AGENT environment variable not set, consider setting it to identify your requests.

Copy

from typing import List
from langchain_community.document_loaders import WebBaseLoader
from langchain_teddynote.tools.tavily import TavilySearch
from langchain_core.tools import tool

# Defining a search tool (TavilySearch)
tavily_tool = TavilySearch(max_results=5)


# Define a tool to scrape details from web pages.
@tool
def scrape_webpages(urls: List[str]) -> str:
    """Use requests and bs4 to scrape the provided web pages for detailed information."""
    # Load web pages using a given list of URLs
    loader = WebBaseLoader(
        web_path=urls,
        header_template={
            "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/102.0.0.0 Safari/537.36",
        },
    )
    docs = loader.load()

    # Generate a string containing the title and contents of the loaded document.
    return "\n\n".join(
        [
            f'<Document name="{doc.metadata.get("title", "")}">\n{doc.page_content}\n</Document>'
            for doc in docs
        ]
    )

ResearchTeam can use search engines and URL scrapers to find information on the web. To improve the performance of ResearchTeam, you can freely add additional features below.

ResearchTeam tools

Let's look at the research team first.

Each team consists of one or more agents, and each agent is equipped with one or more tools. Below we define all the tools to be used by various teams.

Tool creation

Copy

 gpt-4o

Copy

from langchain_teddynote.models import get_model_name, LLMs

# Get the latest model name
MODEL_NAME = get_model_name(LLMs.GPT4o)
print(MODEL_NAME)

Here are the code execution tools PythonREPLTool Defines.

Copy

from langchain_experimental.tools import PythonREPLTool

# PythonREPL tool
python_repl_tool = PythonREPLTool()

Define utility functions for multi-agent creation

Here's how to create some utility functions to concisely perform tasks.

At this time, covered in the previous tutorial functools.partial Create an agent node using functions.

1. worker agent create.

2. sub-graph of supervisor create.

Copy

from langgraph.graph import START, END
from langchain_core.messages import HumanMessage
from langchain_openai.chat_models import ChatOpenAI
from langgraph.prebuilt import create_react_agent


# Agent Factory Class
class AgentFactory:
    def __init__(self, model_name):
        self.llm = ChatOpenAI(model=model_name, temperature=0)

    def create_agent_node(self, agent, name: str):
        # Node creation function
        def agent_node(state):
            result = agent.invoke(state)
            return {
                "messages": [
                    HumanMessage(content=result["messages"][-1].content, name=name)
                ]
            }

        return agent_node


# LLM Initialization
llm = ChatOpenAI(model=MODEL_NAME, temperature=0)

# Create an Agent Factory instance
agent_factory = AgentFactory(MODEL_NAME)

Below AgentFactory An example of using to generate an agent node.

In the example, we will look at how to create a search agent.

Copy

# Agent Definition
search_agent = create_react_agent(llm, tools=[tavily_tool])
# Create an agent node
search_node = agent_factory.create_agent_node(search_agent, name="Searcher")

Here is a function that creates a Team Supervisor.

Copy

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


def create_team_supervisor(model_name, system_prompt, members) -> str:
    # Define the following worker selection options list
    options_for_next = ["FINISH"] + members

    # Define a worker selection response model: Select the next worker or indicate task completion.
    class RouteResponse(BaseModel):
        next: Literal[*options_for_next]

    # ChatPromptTemplate generation
    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_for_next))

    # LLM Initialization
    llm = ChatOpenAI(model=model_name, temperature=0)

    # Combine prompts and LLM to build a chain
    supervisor_chain = prompt | llm.with_structured_output(RouteResponse)

    return supervisor_chain

Agent team definition

Define the Research Team and the Doc Writing Team.

Research Team

The research team search agent Wow web scraping In charge research_agent Ra has two worker nodes. Let's create these and set up a team supervisor.

Copy

import operator
from typing import List, TypedDict
from typing_extensions import Annotated

from langchain_core.messages import BaseMessage, HumanMessage
from langchain_openai.chat_models import ChatOpenAI
from langgraph.prebuilt import create_react_agent


# state definition
class ResearchState(TypedDict):
    messages: Annotated[List[BaseMessage], operator.add]  # message
    team_members: List[str]  # member agent list
    next: str  # Instruct the Supervisor agent to select the next worker.


# LLM Initialization
llm = ChatOpenAI(model=MODEL_NAME, temperature=0)

# Create a search node
search_agent = create_react_agent(llm, tools=[tavily_tool])
search_node = agent_factory.create_agent_node(search_agent, name="Searcher")

# Creating a web scraping node
web_scraping_agent = create_react_agent(llm, tools=[scrape_webpages])
web_scraping_node = agent_factory.create_agent_node(
    web_scraping_agent, name="WebScraper"
)

# Create a Supervisor Agent
supervisor_agent = create_team_supervisor(
    MODEL_NAME,
    "You are a supervisor tasked with managing a conversation between the"
    " following workers: Search, WebScraper. 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.",
    ["Searcher", "WebScraper"],
)

Next, define the function that selects the node to be routed to.

Copy

def get_next_node(x):
    return x["next"]

Research Team Graph Creation

Copy

from langchain_teddynote.graphs import visualize_graph
from langgraph.graph import StateGraph
from langgraph.checkpoint.memory import MemorySaver

# create a graph
web_research_graph = StateGraph(ResearchState)

# add node
web_research_graph.add_node("Searcher", search_node)
web_research_graph.add_node("WebScraper", web_scraping_node)
web_research_graph.add_node("Supervisor", supervisor_agent)

# add edge
web_research_graph.add_edge("Searcher", "Supervisor")
web_research_graph.add_edge("WebScraper", "Supervisor")

# Conditional edge definition: moving to the next node based on the decision of the supervisor node.
web_research_graph.add_conditional_edges(
    "Supervisor",
    get_next_node,
    {"Searcher": "Searcher", "WebScraper": "WebScraper", "FINISH": END},
)

# setting the start node
web_research_graph.set_entry_point("Supervisor")

# compile the graph
web_research_app = web_research_graph.compile(checkpointer=MemorySaver())

# graph visualization
visualize_graph(web_research_app, xray=True)

Copy

# output the final result
print(output["messages"][-1].content)

Copy

 ================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
WebScraper 
================================================== 

================================================== 
🔄 Node: agent in [WebScraper] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  scrap_webpages (call_Yvsu5vbo7UiKBJ6wEoTsGLDc) 
 Call ID: call_Yvsu5vbo7UiKBJ6wEoTsGLDc 
  Args: 
    urls: ['https://finance.naver.com/news'] 
================================================== 

================================================== 
🔄 Node: tools in [WebScraper] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: scrape_webpages 


Naverpei Securities 

(...sway...) 

Breaking news 

Global interest rate cuts... 6000 billion dollars inflows into the bond fund this year'Important Max' 
[Economic greetings] Shinhan Bank 
Padu cheats on sales and inflated conspiracy... Goldsmith, sent to prosecution 
“Debt Sheer ”... The recession seems to be the best of all time this year. 
[Weekly Economic Indicators] December 20, 2024 
[IPO undefeated old saying ③]'Step vs.' shift-up, content is'horizontal' stock price'flat' why? 
“Can I call this ETFra ”... These days, well-going ETFs, I'm thrilled with individual events. 
Ant'taken times'... "minus if you just buy" 
Meet Trump and return home... "I told Trump aides that Korea is low." 
“There is no rice cake, only the rice cake ”... This year, 71 of the 108 listings of Kodak's listings fell. 

Ant'taken times'... "minus if you just buy" 
Bitcoin and Ethereum investment at the same time...SEC approves first complex ETF 
Thanks to export stock, test exchange rate... This year's rise rate is high 
Bitcoin, downtrend in three days ‘stop ’... 90,000 dollars deal 

96,000 dollars regression... 美 Fed Shock Day Day [Bitcoin Now] 
“Samsung · Hyundai makes a profit. 
"Still, you're buying a Samsung company?"... ants'difficult' [Event +] 
Hundingsu Han Dong-tae, Haejoo, 65,000 won → 16000 won [Rae-ju now] 

View more 
Popular color 

Samsung 
53,000 

SK Hynix 
168,500 

Eco Probiem 
114,100 

Imat 
65,800 

Orient 
6,100 

================================================== 

================================================== 
🔄 Node: agent in [WebScraper] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

Below are the main news and sources on the Naver Financial News page: 

One. ** Paddles of sales and bloated conspiracy... Golden Senate sends to prosecution**  
   Source: Chosun Ilbo 

2. **“ Debt Sheer ”... As a recession, it seems to be the best of all time this year**  
   Source: Chosun Ilbo 

3. **'Actioned Counter' Shift-up, content is'horizontal' stock price'flat' Why?[IPO undefeated old saying ③]**  
   Source: Duff 

4. **“ Can I call this ETFra ”... These days, ETFs that go well, and they plump into individual events**  
   Source: Daily Economy 

5. ** Ant'the age of trouble'... "minus when fraudulent"**  
   Source: Korean Economic TV 

6. ** Meet Trump and return home... "I told Trump aides that Korea is low-power"**  
   Source: MBN 

These news can be found in [Naver Finance News] (https://finance.naver.com/news)에서). 
================================================== 

================================================== 
🔄 Node: WebScraper 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: WebScraper 

Below are the main news and sources on the Naver Financial News page: 

One. ** Paddles of sales and bloated conspiracy... Golden Senate sends to prosecution**  
   Source: Chosun Ilbo 

2. **“ Debt Sheer ”... As a recession, it seems to be the best of all time this year**  
   Source: Chosun Ilbo 

3. **'Actioned Counter' Shift-up, content is'horizontal' stock price'flat' Why?[IPO undefeated old saying ③]**  
   Source: Duff 

4. **“ Can I call this ETFra ”... These days, ETFs that go well, and they plump into individual events**  
   Source: Daily Economy 

5. ** Ant'the age of trouble'... "minus when fraudulent"**  
   Source: Korean Economic TV 

6. ** Meet Trump and return home... "I told Trump aides that Korea is low-power"**  
   Source: MBN 

These news can be found in [Naver Finance News] (https://finance.naver.com/news)에서). 
================================================== 

================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
FINISH 
================================================== 

Copy

output = run_graph(
    web_research_app,
    "https://finance.naver.com/news Please summarize and print out the main news. Also print out the source (URL).",
)

Copy

from langchain_core.runnables import RunnableConfig
from langchain_teddynote.messages import random_uuid, invoke_graph


def run_graph(app, message: str, recursive_limit: int = 50):
    # config settings (max recursion count, thread_id)
    config = RunnableConfig(
        recursion_limit=recursive_limit, configurable={"thread_id": random_uuid()}
    )

    # enter your question
    inputs = {
        "messages": [HumanMessage(content=message)],
    }

    # running the graph
    invoke_graph(app, inputs, config)

    return app.get_state(config).values

Copy

Below are the main news and sources on the Naver Financial News page: 

One. ** Paddles of sales and bloated conspiracy... Golden Senate sends to prosecution**  
   Source: Chosun Ilbo 

2. **“ Debt Sheer ”... As a recession, it seems to be the best of all time this year**  
   Source: Chosun Ilbo 

3. **'Actioned Counter' Shift-up, content is'horizontal' stock price'flat' Why?[IPO undefeated old saying ③]**  
   Source: Duff 

4. **“ Can I call this ETFra ”... These days, ETFs that go well, and they plump into individual events**  
   Source: Daily Economy 

5. ** Ant'the age of trouble'... "minus when fraudulent"**  
   Source: Korean Economic TV 

6. ** Meet Trump and return home... "I told Trump aides that Korea is low-power"**  
   Source: MBN 

These news can be found in [Naver Finance News] (https://finance.naver.com/news)에서). 

Doc Writing Team

This time create a documenting team. At this time, each agent is given access to different file-writing tools.

Copy

import operator
from typing import List, TypedDict, Annotated
from pathlib import Path


# Create a temporary directory and set a working directory
WORKING_DIRECTORY = Path("./tmp")
WORKING_DIRECTORY.mkdir(exist_ok=True)  # tmp 폴더가 없으면 생성


# state definition
class DocWritingState(TypedDict):
    messages: Annotated[List[BaseMessage], operator.add]
    team_members: str
    next: str
    current_files: str  # file currently being worked on


# State Preprocessing Node: Allows each agent to be more aware of the state of its current working directory.
def preprocess(state):
    # Initialize the list of written files
    written_files = []

    try:
        # Find all files in the working directory and convert them to relative paths.
        written_files = [
            f.relative_to(WORKING_DIRECTORY) for f in WORKING_DIRECTORY.rglob("*")
        ]
    except Exception:
        pass
    # If there is no file written, the status is "No files written." 추가
    if not written_files:
        return {**state, "current_files": "No files written."}

    # Add a list of created files to the status
    return {
        **state,
        "current_files": "\nBelow are files your team has written to the directory:\n"
        + "\n".join([f" - {f}" for f in written_files]),
    }


# LLM Initialization
llm = ChatOpenAI(model=MODEL_NAME)

# Create a document authoring agent
doc_writer_agent = create_react_agent(
    llm,
    tools=[write_document, edit_document, read_document],
    state_modifier="You are a arxiv researcher. Your mission is to write arxiv style paper on given topic/resources.",
)
context_aware_doc_writer_agent = preprocess | doc_writer_agent
doc_writing_node = agent_factory.create_agent_node(
    context_aware_doc_writer_agent, name="DocWriter"
)

# note writing nodewwww
note_taking_agent = create_react_agent(
    llm,
    tools=[create_outline, read_document],
    state_modifier="You are an expert in creating outlines for research papers. Your mission is to create an outline for a given topic/resources or documents.",
)
context_aware_note_taking_agent = preprocess | note_taking_agent
nwwote_taking_node = agent_factory.create_agent_node(
    context_aware_note_taking_agent, name="NoteTaker"
)


# create a chart generation agent
chart_generating_agent = create_react_agent(
    llm, tools=[read_document, python_repl_tool]
)
context_aware_chart_generating_agent = preprocess | chart_generating_agent
chart_generating_node = agent_factory.create_agent_node(
    context_aware_chart_generating_agent, name="ChartGenerator"
)

# Create a Documentation Team Supervisor
doc_writing_supervisor = create_team_supervisor(
    MODEL_NAME,
    "You are a supervisor tasked with managing a conversation between the"
    " following workers:  ['DocWriter', 'NoteTaker', 'ChartGenerator']. 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.",
    ["DocWriter", "NoteTaker", "ChartGenerator"],
)

Create Doc Writing Team graph

Copy

# create a graph
authoring_graph = StateGraph(DocWritingState)

# node definition
authoring_graph.add_node("DocWriter", doc_writing_node)
authoring_graph.add_node("NoteTaker", note_taking_node)
authoring_graph.add_node("ChartGenerator", chart_generating_node)
authoring_graph.add_node("Supervisor", doc_writing_supervisor)

# edge definition
authoring_graph.add_edge("DocWriter", "Supervisor")
authoring_graph.add_edge("NoteTaker", "Supervisor")
authoring_graph.add_edge("ChartGenerator", "Supervisor")

# Conditional edge definition: moving to the next node based on the decision of the supervisor node
authoring_graph.add_conditional_edges(
    "Supervisor",
    get_next_node,
    {
        "DocWriter": "DocWriter",
        "NoteTaker": "NoteTaker",
        "ChartGenerator": "ChartGenerator",
        "FINISH": END,
    },
)

# swetting the start node
authoring_graph.set_entry_point("Supervisor")

# compile the graph
authoring_app = authoring_graph.compile(checkpointer=MemorySaver())

Visualize the graph.

Copy

Copyfrom langchain_teddynote.graphs import visualize_graph

# graph visualization
visualize_graph(authoring_app, xray=True)

Copy

 Python REPL can execute arbitrary code. Use with caution.

Copy

 ================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
DocWriter 
================================================== 

================================================== 
🔄 Node: agent in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  write_document (call_SMJQbkGc5x7eMe4hNb5BmSU2) 
 Call ID: call_SMJQbkGc5x7eMe4hNb5BmSU2 
  Args: 
    content: # table of contents 

One. Introduction 
2. Transformer Background 
3. Transformer Structure 
   3.1. encoder 
   3.2. decoder 
   3.3. Attention mechanism 
4. Benefits of the transformer model 
5. Transformers Applications 
6. Conclusion and future prospects 
    file_name: Transformer_Structure_Analysis.txt 
================================================== 

================================================== 
🔄 Node: tools in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: write_document 

Document saved to Transformer_Structure_Analysis.txt 
================================================== 

================================================== 
🔄 Node: agent in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  edit_document (call_ZjKp9lvhY9h8zJWP0OASptjw) 
 Call ID: call_ZjKp9lvhY9h8zJWP0OASptjw 
  Args: 
    file_name: Transformer_Structure_Analysis.txt 
================================================== 

================================================== 
🔄 Node: tools in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: edit_document 

Error: 1 validation error for edit_document 
inserts 
  Field required [type=missing, input_value={'file_name':'Transforme...Structure_Analysis.txt'}, input_type=dict] 
    For further information visit https://errors.pydantic.dev/2.10/v/missing 
 Please fix your mistakes. 
================================================== 

================================================== 
🔄 Node: agent in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  edit_document (call_4qNiVcNaBPHWRgplOytLnXh1) 
 Call ID: call_4qNiVcNaBPHWRgplOytLnXh1 
  Args: 
    file_name: Transformer_Structure_Analysis.txt 
    inserts: [{'line': 1,'text':'\n\n# 1. Introduction\n\ntransformers are innovative models in the field of natural language processing, designed to effectively handle complex dependencies. This paper deeply analyzes the structure of the transformer and its key elements, and explains why this model is important. Transformers offer the advantage of parallel processing in particular, which is very suitable for modern applications where large amounts of data have to be processed. In addition, this model has been successfully applied in various fields, which demonstrates the flexibility and strength of transformers. This paper will develop a discussion around the structural features of these transformers.\n\n 
    (...sway...) 
    # 6. Conclusions and Future Prospects\n\n Transformers are getting a lot of attention thanks to their innovative structure and outstanding performance. This paper looked at the structural features of transformers and their benefits. In the future, transformers will be utilized in more fields, which will provide an important direction for AI research. Moreover, the development of transformers is expected to lead to the development of new models and technologies. In conclusion, transformers are playing a key role in advancing AI technology, and the possibilities are endless.'}] 
================================================== 

================================================== 
🔄 Node: tools in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: edit_document 

Error: 1 validation error for edit_document 
inserts 
  Input should be be a valid dictionary [type=dict_type, input_value=[{'line': 1,'text':'\n\...I'm in trouble.'}], input_type=list] 
    For further information visit https://errors.pydantic.dev/2.10/v/dict_type 
 Please fix your mistakes. 
================================================== 

================================================== 
🔄 Node: agent in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  edit_document (call_YziJh8x5GWBv0fehTA0l2i4F) 
 Call ID: call_YziJh8x5GWBv0fehTA0l2i4F 
  Args: 
    file_name: Transformer_Structure_Analysis.txt 
    inserts: {'1':'\n\n# 1. Introduction\n\ntransformers are innovative models in the field of natural language processing, designed to effectively handle complex dependencies. This paper deeply analyzes the structure of the transformer and its key elements, and explains why this model is important. Transformers offer the advantage of parallel processing in particular, which is very suitable for modern applications where large amounts of data have to be processed. In addition, this model has been successfully applied in various fields, which demonstrates the flexibility and strength of transformers. This paper will develop a discussion around the structural features of these transformers.\n\n 
    (...sway...) 
    In this section, we will look at the various applications with transformers and their achievements.\n\n# 6. Conclusions and Future Prospects\n\n Transformers are getting a lot of attention thanks to their innovative structure and outstanding performance. This paper looked at the structural features of transformers and their benefits. In the future, transformers will be utilized in more fields, which will provide an important direction for AI research. Moreover, the development of transformers is expected to lead to the development of new models and technologies. In conclusion, transformers are playing a key role in advancing AI technology, and the possibilities are endless.'} 
================================================== 

================================================== 
🔄 Node: tools in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: edit_document 

Document edited and saved to Transformer_Structure_Analysis.txt 
================================================== 

================================================== 
🔄 Node: agent in [DocWriter] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

I wrote a paper on the structure of the transformer, and I saved the document in "Transformer_Structure_Analysis.txt". This document contains details of the transformer's introduction, background, structure, benefits, applications, conclusions and future prospects. The document in-depth describes the transformer's encoder, decoder, and attention mechanism. 
================================================== 

================================================== 
🔄 Node: DocWriter 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: DocWriter 

I wrote a paper on the structure of the transformer, and I saved the document in "Transformer_Structure_Analysis.txt". This document contains details of the transformer's introduction, background, structure, benefits, applications, conclusions and future prospects. The document in-depth describes the transformer's encoder, decoder, and attention mechanism. 
================================================== 

================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
NoteTaker 
================================================== 

================================================== 
🔄 Node: agent in [NoteTaker] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  read_document (call_AHWWnWBgshpk1hGdwFbTq6Ei) 
 Call ID: call_AHWWnWBgshpk1hGdwFbTq6Ei 
  Args: 
    file_name: Transformer_Structure_Analysis.txt 
================================================== 

================================================== 
🔄 Node: tools in [NoteTaker] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: read_document 

# One. Introduction 

Transformers are innovative models in the field of natural language processing, designed to effectively handle complex dependencies. This paper deeply analyzes the structure of the transformer and its key elements, and explains why this model is important. Transformers offer the advantage of parallel processing in particular, which is very suitable for modern applications where large amounts of data have to be processed. In addition, this model has been successfully applied in various fields, which demonstrates the flexibility and strength of transformers. This paper will develop discussions around the structural features of these transformers. 

(...sway...) 

# 6. Conclusion and future prospects 

Transformers are getting a lot of attention thanks to their innovative structure and outstanding performance. This paper looked at the structural features of transformers and their benefits. In the future, transformers will be utilized in more fields, which will provide an important direction for AI research. Moreover, the development of transformers is expected to lead to the development of new models and technologies. In conclusion, transformers are playing a key role in advancing AI technology, and the possibilities are endless. 

# Table of Contents 

One. Introduction 

2. Transformer Background 

3. Transformer Structure 

   3.1. encoder 

   3.2. decoder 

   3.3. Attention mechanism 

4. Benefits of the transformer model 

5. Transformers Applications 

6. Conclusion and future prospects 
================================================== 

================================================== 
🔄 Node: agent in [NoteTaker] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  create_outline (call_PIqz0USGzgK2716P7OVcBq25) 
 Call ID: call_PIqz0USGzgK2716P7OVcBq25 
  Args: 
    points: ['1. Introduction', '2. Transformer Background', '3. Transformer Structure', '3.1. Encoder', '3.2. Decoder', '3.3. Attention mechanism', '4. Benefits of the transformer model', '5. Transformers' Applications', '6. Conclusion and future prospects'] 
    file_name: Transformer_Structure_Outline.txt 
================================================== 

================================================== 
🔄 Node: tools in [NoteTaker] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: create_outline 

Outline saved to Transformer_Structure_Outline.txt 
================================================== 

================================================== 
🔄 Node: agent in [NoteTaker] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

The table of contents of the paper on the structure of the transformer was stored in the "Transformer_Structure_Outline.txt" file. I will write a detailed description of each table of contents. 

### 1. Introduction 
-Transformers are innovative models in the field of natural language processing, designed to effectively deal with complex dependencies.  
-This paper deeply analyzes the structure of the transformer and its key elements, and explains why this model is important.  
-Transformers offer the advantage of parallel processing in particular, which is very suitable for modern applications where large amounts of data have to be processed.  
-In addition, this model has been successfully applied in various fields, which demonstrates the flexibility and strength of transformers.  
-This paper will develop discussions around the structural features of these transformers. 

(...sway...) 

### 5. Transformers Applications 
-Transformers are proving their usefulness in various fields.  
-In the field of natural language processing (NLP), it is utilized in a variety of tasks, including translation, summary, and emotional analysis.  
-In addition, in the field of image processing, it has been successfully applied in several tasks, including image generation.  
-Recently, we are expanding our application to various fields such as bioinformatics and music generation.  
-In this section, we will look at the various applications with transformers and their performance. 

### 6. Conclusion and future prospects 
-Transformers are getting a lot of attention thanks to their innovative structure and outstanding performance.  
-This paper looked at the structural features of the transformer and its benefits.  
-In the future, transformers will be utilized in more fields, which will provide an important direction for AI research.  
- Moreover, the development of transformers is expected to lead to the development of new models and technologies.  
-In conclusion, transformers are playing a key role in advancing AI technology, and the possibilities are endless. 

This description is more than 5 sentences for each table of contents required to write a research paper. 
================================================== 

================================================== 
🔄 Node: NoteTaker 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: NoteTaker 

The table of contents of the paper on the structure of the transformer was stored in the "Transformer_Structure_Outline.txt" file. I will write a detailed description of each table of contents. 

### 1. Introduction 
-Transformers are innovative models in the field of natural language processing, designed to effectively deal with complex dependencies.  
-This paper deeply analyzes the structure of the transformer and its key elements, and explains why this model is important.  
-Transformers offer the advantage of parallel processing in particular, which is very suitable for modern applications where large amounts of data have to be processed.  
-In addition, this model has been successfully applied in various fields, which demonstrates the flexibility and strength of transformers.  
-This paper will develop discussions around the structural features of these transformers. 

(...sway...) 

### 6. Conclusion and future prospects 
-Transformers are getting a lot of attention thanks to their innovative structure and outstanding performance.  
-This paper looked at the structural features of the transformer and its benefits.  
-In the future, transformers will be utilized in more fields, which will provide an important direction for AI research.  
- Moreover, the development of transformers is expected to lead to the development of new models and technologies.  
-In conclusion, transformers are playing a key role in advancing AI technology, and the possibilities are endless. 

This description is more than 5 sentences for each table of contents required to write a research paper. 
================================================== 

================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
ChartGenerator 
================================================== 

================================================== 
🔄 Node: agent in [ChartGenerator] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 
Tool Calls: 
  read_document (call_LoYx78kH5tKzFEqENxTQirPZ) 
 Call ID: call_LoYx78kH5tKzFEqENxTQirPZ 
  Args: 
    file_name: Transformer_Structure_Outline.txt 
    start: 0 
    end: 50 
================================================== 

================================================== 
🔄 Node: tools in [ChartGenerator] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: read_document 

One. One. Introduction 

2. 2. Transformer Background 

3. 3. Transformer Structure 

4. 3.1. encoder 

5. 3.2. decoder 

6. 3.3. Attention mechanism 

7. 4. Benefits of the transformer model 

8. 5. Transformers Applications 

9. 6. Conclusion and future prospects 

================================================== 

Copy

output = run_graph(
    authoring_app,
    "Please go into depth about the structure of the transformer and write the table of contents of the paper in Korean. "
    "Then, write at least 5 sentences for each table of contents. "
    "If a chart is needed when writing detailed information, please create one. "
    "Save the final result. ",
)

Super-Graph generation

for this Super Graph We will define the previous two graphs, and add connection elements that define how these higher-level states are shared between different graphs. First, create a total supervisor node.

In this design Bottom-up planning policy Is applied. I have already created two graphs, but I have to decide how to route the work between them.

Copy

================================================== 
🔄 Node: agent in [ChartGenerator] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

The table of contents for the paper is as follows: 

One. Introduction 
2. Transformer Background 
3. Transformer Structure 
   3.1. encoder 
   3.2. decoder 
   3.3. Attention mechanism 
4. Benefits of the transformer model 
5. Transformers Applications 
6. Conclusion and future prospects 

The detailed description of each table of contents is the same as previously provided. I will save the table of contents and detailed description of the paper as a file "Transformer_Structure_Analysis.txt". 
Tool Calls: 
  Python_REPL (call_PdvTGUNNVbQ88aYTExoUFaMg) 
 Call ID: call_PdvTGUNNVbQ88aYTExoUFaMg 
  Args: 
    query: outline = "" 
One. Introduction 
-Transformers are innovative models in the field of natural language processing, designed to effectively deal with complex dependencies.  
-This paper deeply analyzes the structure of the transformer and its key elements, and explains why this model is important.  
-Transformers offer the advantage of parallel processing in particular, which is very suitable for modern applications where large amounts of data have to be processed.  
-In addition, this model has been successfully applied in various fields, which demonstrates the flexibility and strength of transformers.  
-This paper will develop discussions around the structural features of these transformers. 

(...sway...) 

5. Transformers Applications 
-Transformers are proving their usefulness in various fields.  
-In the field of natural language processing (NLP), it is utilized in a variety of tasks, including translation, summary, and emotional analysis.  
-In addition, in the field of image processing, it has been successfully applied in several tasks, including image generation.  
-Recently, we are expanding our application to various fields such as bioinformatics and music generation.  
-In this section, we will look at the various applications with transformers and their performance. 

6. Conclusion and future prospects 
-Transformers are getting a lot of attention thanks to their innovative structure and outstanding performance.  
-This paper looked at the structural features of the transformer and its benefits.  
-In the future, transformers will be utilized in more fields, which will provide an important direction for AI research.  
- Moreover, the development of transformers is expected to lead to the development of new models and technologies.  
-In conclusion, transformers are playing a key role in advancing AI technology, and the possibilities are endless. 
""" 

with open('Transformer_Structure_Analysis.txt','w', encoding='utf-8') as file: 
    file.write(outline) 
================================================== 

================================================== 
🔄 Node: tools in [ChartGenerator] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================= Tool Message ================================= 
Name: Python_REPL 


================================================== 

================================================== 
🔄 Node: agent in [ChartGenerator] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

A detailed description of the paper on the structure of the transformer was stored in the "Transformer_Structure_Analysis.txt" file. This file contains the structure of the transformer and its advantages, and a deep description of the application. 
================================================== 

================================================== 
🔄 Node: ChartGenerator 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: ChartGenerator 

A detailed description of the paper on the structure of the transformer was stored in the "Transformer_Structure_Analysis.txt" file. This file contains the structure of the transformer and its advantages, and a deep description of the application. 
================================================== 

================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
FINISH 
================================================== 

Copy

from langchain_teddynote.graphs import visualize_graph

# 그래프 시각화
visualize_graph(super_graph, xray=True)

Visualize the graph.

Copy

# graph definition
super_graph = StateGraph(State)

# node definition
super_graph.add_node("ResearchTeam", get_last_message | web_research_app | join_graph)
super_graph.add_node("PaperWritingTeam", get_last_message | authoring_app | join_graph)
super_graph.add_node("Supervisor", supervisor_node)

# edge definition
super_graph.add_edge("ResearchTeam", "Supervisor")
super_graph.add_edge("PaperWritingTeam", "Supervisor")

# Add conditional edge: Move to next node based on Supervisor's decision
super_graph.add_conditional_edges(
    "Supervisor",
    get_next_node,
    {
        "PaperWritingTeam": "PaperWritingTeam",
        "ResearchTeam": "ResearchTeam",
        "FINISH": END,
    },
)

# Set the Supervisor node as the starting node
super_graph.set_entry_point("Supervisor")

# compile the graph
super_graph = super_graph.compile(checkpointer=MemorySaver())

Now define a Super-Graph that connects 2 teams.

Super-Graph definition

Copy

from typing import TypedDict, List, Annotated
import operator


# swtate definition
class State(TypedDict):
    messages: Annotated[List[BaseMessage], operator.add]
    # routing decisions
    next: str


# Last message return node
def get_last_message(state: State) -> str:
    last_message = state["messages"][-1]
    if isinstance(last_message, str):
        return {"messages": [HumanMessage(content=last_message)]}
    else:
        return {"messages": [last_message.content]}


# Response Comprehensive Node
def join_graph(response: dict):
    # Extract the last message and return it as a message list
    return {"messages": [response["messages"][-1]]}

Super-Graph simply plays the role of routing Task.

The following defines the status and nodes of the Super-Graph.

Copy

from langchain_core.messages import BaseMessage
from langchain_openai.chat_models import ChatOpenAI

# Create a ChatOpenAI instance with basic LLM
llm = ChatOpenAI(model=MODEL_NAME)

# Create a Team Supervisor Node
supervisor_node = create_team_supervisor(
    MODEL_NAME,
    "You are a supervisor tasked with managing a conversation between the"
    " following teams: ['ResearchTeam', 'PaperWritingTeam']. 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.",
    ["ResearchTeam", "PaperWritingTeam"],
)

Copy

# How to perform complex tasks using a multi-agent architecture

## index
1. introduction
2. Overview of the Multi-gent System
3. Advantages of Multi-Agent Systems
4. Components of a Multi-Agent System
5. Applications of Multi-Agent Systems
6. Implementation Case of Multi-Agent System
7. conclusion
8. references

## 1. introduction
Multi-Agent 시스템은 여러 개의 에이전트가 협력하여 복잡한 문제를 해결하는 분산 인공지능 시스템을 의미합니다. 이러한 시스템은 각 에이전트가 독립적으로 작동하면서도 상호작용을 통해 전체 시스템의 목표를 달성합니다. 최근 들어, 복잡한 작업을 효율적으로 수행하기 위해 Multi-Agent 시스템의 중요성이 부각되고 있습니다. 본 리포트에서는 Multi-Agent 시스템의 구조와 이를 활용한 복잡한 작업 수행 방법에 대해 논의합니다. 또한, 다양한 응용 분야와 구현 사례를 통해 Multi-Agent 시스템의 실용성을 살펴봅니다.

## 2. Overview of the Multi-Agent System
Multi-Agent 시스템은 여러 개의 자율적인 에이전트로 구성되어 있으며, 각 에이전트는 독립적으로 의사결정을 내릴 수 있는 능력을 갖추고 있습니다. 이러한 시스템은 분산된 환경에서의 문제 해결에 적합하며, 각 에이전트는 특정 역할이나 기능을 수행합니다. 에이전트 간의 상호작용은 시스템의 효율성을 높이고, 복잡한 문제를 해결하는 데 필수적입니다. Multi-Agent 시스템은 주로 분산 컴퓨팅, 로봇 공학, 네트워크 관리 등 다양한 분야에서 활용됩니다. 이러한 시스템의 설계와 구현은 에이전트 간의 협력과 조정 메커니즘을 중심으로 이루어집니다.

## 3. Advantages of Multi-Agent Systems
Multi-Agent 시스템의 주요 장점 중 하나는 확장성입니다. 시스템에 새로운 에이전트를 추가함으로써 쉽게 확장할 수 있으며, 이는 시스템의 유연성을 높입니다. 또한, 에이전트 간의 병렬 처리를 통해 작업의 효율성을 극대화할 수 있습니다. 이러한 시스템은 고장 허용성이 높아, 일부 에이전트의 오류가 전체 시스템에 미치는 영향을 최소화합니다. Multi-Agent 시스템은 복잡한 문제를 작은 하위 문제로 나누어 해결할 수 있어, 문제 해결의 복잡성을 줄입니다. 마지막으로, 다양한 에이전트가 서로 다른 전문성을 가질 수 있어, 다양한 문제에 대한 포괄적인 접근이 가능합니다.

## 4. Components of a Multi-Agent System
Multi-Agent 시스템은 일반적으로 에이전트, 환경, 상호작용 메커니즘으로 구성됩니다. 에이전트는 자율적으로 행동하며, 환경은 에이전트가 상호작용하는 공간을 제공합니다. 상호작용 메커니즘은 에이전트 간의 통신과 협력을 가능하게 하며, 이를 통해 시스템의 목표를 달성합니다. 각 에이전트는 감지, 계획, 행동의 세 가지 주요 기능을 수행합니다. 이러한 구성 요소들은 시스템의 설계와 구현에 있어 중요한 역할을 하며, 각 요소의 효율적인 통합이 시스템의 성능을 좌우합니다.

## 5. Applications of Multi-Agent Systems
Multi-Agent 시스템은 다양한 분야에서 응용되고 있으며, 그 중 대표적인 분야로는 로봇 공학, 금융, 물류, 스마트 그리드 등이 있습니다. 로봇 공학에서는 다수의 로봇이 협력하여 작업을 수행하는 데 사용됩니다. 금융 분야에서는 에이전트 기반의 거래 시스템이 시장 분석과 자동 거래를 수행합니다. 물류 분야에서는 물류 최적화와 공급망 관리에 활용됩니다. 스마트 그리드에서는 에너지 관리와 최적화에 Multi-Agent 시스템이 사용됩니다. 이러한 응용 분야들은 Multi-Agent 시스템의 유연성과 효율성을 잘 보여줍니다.

## 6. Implementation Case of Multi-Agent System
Multi-Agent 시스템의 구현 사례로는 자율 주행 차량 시스템, 스마트 홈 자동화 시스템, 분산 네트워크 관리 시스템 등이 있습니다. 자율 주행 차량 시스템에서는 각 차량이 에이전트로 작동하여 교통 흐름을 최적화합니다. 스마트 홈 자동화 시스템에서는 가전 기기들이 에이전트로 작동하여 에너지 효율성을 높입니다. 분산 네트워크 관리 시스템에서는 네트워크 장비들이 에이전트로 작동하여 네트워크 성능을 최적화합니다. 이러한 구현 사례들은 Multi-Agent 시스템의 실용성과 효과를 입증합니다.

## 7. conclusion
Multi-Agent 시스템은 복잡한 문제를 효율적으로 해결할 수 있는 강력한 도구입니다. 이러한 시스템은 다양한 분야에서 응용 가능하며, 그 유연성과 확장성은 시스템 설계에 있어 큰 장점으로 작용합니다. Multi-Agent 시스템의 성공적인 구현을 위해서는 에이전트 간의 효과적인 상호작용과 협력 메커니즘이 필수적입니다. 앞으로도 Multi-Agent 시스템은 더욱 발전하여 다양한 복잡한 문제 해결에 기여할 것으로 기대됩니다.

## 8. references
- Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg.
- Weiss, G. (Ed.). (1999). Multiagent systems: a modern approach to distributed artificial intelligence. MIT press.
- Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons.

이 리포트는 Multi-Agent 시스템의 구조와 응용에 대한 전반적인 이해를 돕기 위해 작성되었습니다. 각 섹션은 Multi-Agent 시스템의 다양한 측면을 다루며, 복잡한 작업을 수행하는 데 있어 이러한 시스템의 중요성을 강조합니다.

Copy

from IPython.display import Markdown

if hasattr(output["messages"][-1], "content"):
    display(Markdown(output["messages"][-1].content))
else:
    display(Markdown(output["messages"][-1]))

Outputs the final result in a markdown format.

Copy

 ================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
ResearchTeam 
================================================== 

================================================== 
🔄 Node: Supervisor in [ResearchTeam] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
Searcher 
================================================== 

================================================== 
🔄 Node: agent in [Searcher] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

# How to perform complex tasks using Multi-Agent structures 

## Table of Contents 
One. Introduction 
2. Overview of the Multi-Agent system 
3. Advantages of the Multi-Agent system 
4. Components of the Multi-Agent system 
5. Applications of Multi-Agent systems 
6. Implementation examples of Multi-Agent systems 
7. conclusion 
8. references 

## One. Introduction 
The Multi-Agent system refers to a distributed artificial system in which multiple agents work together to solve complex problems. These systems achieve the goals of the entire system through mutual operation, while each agent works independently. Recently, the importance of Multi-Agent systems is emerging to efficiently perform complex tasks. This report discusses the structure of the Multi-Agent system and how to perform complex tasks that utilize it. In addition, we will look at the practicality of Multi-Agent systems through a variety of applications and implementation examples. 

## 2. Overview of the Multi-Agent system 
The Multi-Agent system consists of several autonomous agents, each of which has the ability to make decisions independently. These systems are suitable for solving problems in a decentralized environment, and each agent performs a specific role or function. Interaction between agents is essential to increasing the efficiency of the system and solving complex problems. Multi-Agent systems are mainly utilized in a variety of fields, including distributed computing, robotics, and network management. The design and implementation of these systems is centered around the cooperation and coordination mechanisms between agents. 

## 3. Advantages of the Multi-Agent system 
One of the main advantages of the Multi-Agent system is scalability. By adding a new agent to the system, it can be easily extended, which increases the flexibility of the system. In addition, parallel processing between agents allows you to maximize the efficiency of your work. These systems have high fault tolerance, minimizing the impact of some agent errors on the entire system. The Multi-Agent system can be solved by dividing complex problems into small sub-problems, reducing the complexity of problem solving. Finally, different agents can have different expertise, allowing comprehensive access to a variety of issues. 

## 4. Components of the Multi-Agent system 
Multi-Agent systems usually consist of agents, environments, and interaction mechanisms. Agents act autonomously, and the environment provides space for agents to interact with. The interaction mechanism enables communication and cooperation between agents, which achieves the goals of the system. Each agent performs three main functions: detection, planning, and action. These components play an important role in the design and implementation of the system, and the efficient integration of each element depends on the performance of the system. 

## 5. Applications of Multi-Agent systems 
Multi-Agent systems are being applied in a variety of fields, among which are representative areas: robotics, finance, logistics, smart grids, etc. In robotics, many robots are used to collaborate to perform tasks. In the financial sector, agent-based trading systems conduct market analysis and automatic trading. In the logistics sector, it is used for logistics optimization and supply chain management. In the smart grid, Multi-Agent systems are used for energy management and optimization. These applications demonstrate the flexibility and efficiency of the Multi-Agent system. 

## 6. Implementation examples of Multi-Agent systems 
Examples of implementation of Multi-Agent systems include autonomous vehicle systems, smart home automation systems, and distributed network management systems. In an autonomous vehicle system, each vehicle acts as an agent to optimize traffic flow. In smart home automation systems, consumer electronics act as agents to increase energy efficiency. In distributed network management systems, network equipment acts as agents to optimize network performance. These implementation cases demonstrate the practicality and effectiveness of the Multi-Agent system. 

## 7. conclusion 
The Multi-Agent system is a powerful tool for efficiently solving complex problems. These systems are applicable in a variety of fields, and their flexibility and scalability serve as a great advantage in system design. Effective interaction and cooperation mechanisms between agents are essential for the successful implementation of the Multi-Agent system. In the future, the Multi-Agent system is expected to further develop and contribute to solving various complex problems. 

## 8. references 
-Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg. 
-Weiss, G. (Ed.). (1999). Multiagent systems: a modern approx to distributed material intelligence. MIT press. 
-Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons. 

This report was written to help you understand the overall structure and application of the Multi-Agent system. Each section covers various aspects of the Multi-Agent system and highlights the importance of these systems in performing complex tasks. 
================================================== 

================================================== 
🔄 Node: Searcher in [ResearchTeam] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: Searcher 

# How to perform complex tasks using Multi-Agent structures 

## Table of Contents 
One. Introduction 
2. Overview of the Multi-Agent system 
3. Advantages of the Multi-Agent system 
4. Components of the Multi-Agent system 
5. Applications of Multi-Agent systems 
6. Implementation examples of Multi-Agent systems 
7. conclusion 
8. references 

## One. Introduction 
The Multi-Agent system refers to a distributed artificial system in which multiple agents work together to solve complex problems. These systems achieve the goals of the entire system through mutual operation, while each agent works independently. Recently, the importance of Multi-Agent systems is emerging to efficiently perform complex tasks. This report discusses the structure of the Multi-Agent system and how to perform complex tasks that utilize it. In addition, we will look at the practicality of Multi-Agent systems through a variety of applications and implementation examples. 

## 2. Overview of the Multi-Agent system 
The Multi-Agent system consists of several autonomous agents, each of which has the ability to make decisions independently. These systems are suitable for solving problems in a decentralized environment, and each agent performs a specific role or function. Interaction between agents is essential to increasing the efficiency of the system and solving complex problems. Multi-Agent systems are mainly utilized in a variety of fields, including distributed computing, robotics, and network management. The design and implementation of these systems is centered around the cooperation and coordination mechanisms between agents. 

## 3. Advantages of the Multi-Agent system 
One of the main advantages of the Multi-Agent system is scalability. By adding a new agent to the system, it can be easily extended, which increases the flexibility of the system. In addition, parallel processing between agents allows you to maximize the efficiency of your work. These systems have high fault tolerance, minimizing the impact of some agent errors on the entire system. The Multi-Agent system can be solved by dividing complex problems into small sub-problems, reducing the complexity of problem solving. Finally, different agents can have different expertise, allowing comprehensive access to a variety of issues. 

## 4. Components of the Multi-Agent system 
Multi-Agent systems usually consist of agents, environments, and interaction mechanisms. Agents act autonomously, and the environment provides space for agents to interact with. The interaction mechanism enables communication and cooperation between agents, which achieves the goals of the system. Each agent performs three main functions: detection, planning, and action. These components play an important role in the design and implementation of the system, and the efficient integration of each element depends on the performance of the system. 

## 5. Applications of Multi-Agent systems 
Multi-Agent systems are being applied in a variety of fields, among which are representative areas: robotics, finance, logistics, smart grids, etc. In robotics, many robots are used to collaborate to perform tasks. In the financial sector, agent-based trading systems conduct market analysis and automatic trading. In the logistics sector, it is used for logistics optimization and supply chain management. In the smart grid, Multi-Agent systems are used for energy management and optimization. These applications demonstrate the flexibility and efficiency of the Multi-Agent system. 

## 6. Implementation examples of Multi-Agent systems 
Examples of implementation of Multi-Agent systems include autonomous vehicle systems, smart home automation systems, and distributed network management systems. In an autonomous vehicle system, each vehicle acts as an agent to optimize traffic flow. In smart home automation systems, consumer electronics act as agents to increase energy efficiency. In distributed network management systems, network equipment acts as agents to optimize network performance. These implementation cases demonstrate the practicality and effectiveness of the Multi-Agent system. 

## 7. conclusion 
The Multi-Agent system is a powerful tool for efficiently solving complex problems. These systems are applicable in a variety of fields, and their flexibility and scalability serve as a great advantage in system design. Effective interaction and cooperation mechanisms between agents are essential for the successful implementation of the Multi-Agent system. In the future, the Multi-Agent system is expected to further develop and contribute to solving various complex problems. 

## 8. references 
-Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg. 
-Weiss, G. (Ed.). (1999). Multiagent systems: a modern approx to distributed material intelligence. MIT press. 
-Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons. 

This report was written to help you understand the overall structure and application of the Multi-Agent system. Each section covers various aspects of the Multi-Agent system and highlights the importance of these systems in performing complex tasks. 
================================================== 

================================================== 
🔄 Node: Supervisor in [ResearchTeam] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
WebScraper 
================================================== 

================================================== 
🔄 Node: agent in [WebScraper] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================== Ai Message ================================== 

# How to perform complex tasks using Multi-Agent structures 

## Table of Contents 
One. Introduction 
2. Overview of the Multi-Agent system 
3. Advantages of the Multi-Agent system 
4. Components of the Multi-Agent system 
5. Applications of Multi-Agent systems 
6. Implementation examples of Multi-Agent systems 
7. conclusion 
8. references 

## One. Introduction 
The Multi-Agent system refers to a distributed artificial system in which multiple agents work together to solve complex problems. These systems achieve the goals of the entire system through mutual operation, while each agent works independently. Recently, the importance of Multi-Agent systems is emerging to efficiently perform complex tasks. This report discusses the structure of the Multi-Agent system and how to perform complex tasks that utilize it. In addition, we will look at the practicality of Multi-Agent systems through a variety of applications and implementation examples. 

## 2. Overview of the Multi-Agent system 
The Multi-Agent system consists of several autonomous agents, each of which has the ability to make decisions independently. These systems are suitable for solving problems in a decentralized environment, and each agent performs a specific role or function. Interaction between agents is essential to increasing the efficiency of the system and solving complex problems. Multi-Agent systems are mainly utilized in a variety of fields, including distributed computing, robotics, and network management. The design and implementation of these systems is centered around the cooperation and coordination mechanisms between agents. 

## 3. Advantages of the Multi-Agent system 
One of the main advantages of the Multi-Agent system is scalability. By adding a new agent to the system, it can be easily extended, which increases the flexibility of the system. In addition, parallel processing between agents allows you to maximize the efficiency of your work. These systems have high fault tolerance, minimizing the impact of some agent errors on the entire system. The Multi-Agent system can be solved by dividing complex problems into small sub-problems, reducing the complexity of problem solving. Finally, different agents can have different expertise, allowing comprehensive access to a variety of issues. 

## 4. Components of the Multi-Agent system 
Multi-Agent systems usually consist of agents, environments, and interaction mechanisms. Agents act autonomously, and the environment provides space for agents to interact with. The interaction mechanism enables communication and cooperation between agents, which achieves the goals of the system. Each agent performs three main functions: detection, planning, and action. These components play an important role in the design and implementation of the system, and the efficient integration of each element depends on the performance of the system. 

## 5. Applications of Multi-Agent systems 
Multi-Agent systems are being applied in a variety of fields, among which are representative areas: robotics, finance, logistics, smart grids, etc. In robotics, many robots are used to collaborate to perform tasks. In the financial sector, agent-based trading systems conduct market analysis and automatic trading. In the logistics sector, it is used for logistics optimization and supply chain management. In the smart grid, Multi-Agent systems are used for energy management and optimization. These applications demonstrate the flexibility and efficiency of the Multi-Agent system. 

## 6. Implementation examples of Multi-Agent systems 
Examples of implementation of Multi-Agent systems include autonomous vehicle systems, smart home automation systems, and distributed network management systems. In an autonomous vehicle system, each vehicle acts as an agent to optimize traffic flow. In smart home automation systems, consumer electronics act as agents to increase energy efficiency. In distributed network management systems, network equipment acts as agents to optimize network performance. These implementation cases demonstrate the practicality and effectiveness of the Multi-Agent system. 

## 7. conclusion 
The Multi-Agent system is a powerful tool for efficiently solving complex problems. These systems are applicable in a variety of fields, and their flexibility and scalability serve as a great advantage in system design. Effective interaction and cooperation mechanisms between agents are essential for the successful implementation of the Multi-Agent system. In the future, the Multi-Agent system is expected to further develop and contribute to solving various complex problems. 

## 8. references 
-Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg. 
-Weiss, G. (Ed.). (1999). Multiagent systems: a modern approx to distributed material intelligence. MIT press. 
-Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons. 

This report was written to help you understand the overall structure and application of the Multi-Agent system. Each section covers various aspects of the Multi-Agent system and highlights the importance of these systems in performing complex tasks. 
================================================== 

================================================== 
🔄 Node: WebScraper in [ResearchTeam] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: WebScraper 

# How to perform complex tasks using Multi-Agent structures 

## Table of Contents 
One. Introduction 
2. Overview of the Multi-Agent system 
3. Advantages of the Multi-Agent system 
4. Components of the Multi-Agent system 
5. Applications of Multi-Agent systems 
6. Implementation examples of Multi-Agent systems 
7. conclusion 
8. references 

## One. Introduction 
The Multi-Agent system refers to a distributed artificial system in which multiple agents work together to solve complex problems. These systems achieve the goals of the entire system through mutual operation, while each agent works independently. Recently, the importance of Multi-Agent systems is emerging to efficiently perform complex tasks. This report discusses the structure of the Multi-Agent system and how to perform complex tasks that utilize it. In addition, we will look at the practicality of Multi-Agent systems through a variety of applications and implementation examples. 

## 2. Overview of the Multi-Agent system 
The Multi-Agent system consists of several autonomous agents, each of which has the ability to make decisions independently. These systems are suitable for solving problems in a decentralized environment, and each agent performs a specific role or function. Interaction between agents is essential to increasing the efficiency of the system and solving complex problems. Multi-Agent systems are mainly utilized in a variety of fields, including distributed computing, robotics, and network management. The design and implementation of these systems is centered around the cooperation and coordination mechanisms between agents. 

## 3. Advantages of the Multi-Agent system 
One of the main advantages of the Multi-Agent system is scalability. By adding a new agent to the system, it can be easily extended, which increases the flexibility of the system. In addition, parallel processing between agents allows you to maximize the efficiency of your work. These systems have high fault tolerance, minimizing the impact of some agent errors on the entire system. The Multi-Agent system can be solved by dividing complex problems into small sub-problems, reducing the complexity of problem solving. Finally, different agents can have different expertise, allowing comprehensive access to a variety of issues. 

## 4. Components of the Multi-Agent system 
Multi-Agent systems usually consist of agents, environments, and interaction mechanisms. Agents act autonomously, and the environment provides space for agents to interact with. The interaction mechanism enables communication and cooperation between agents, which achieves the goals of the system. Each agent performs three main functions: detection, planning, and action. These components play an important role in the design and implementation of the system, and the efficient integration of each element depends on the performance of the system. 

## 5. Applications of Multi-Agent systems 
Multi-Agent systems are being applied in a variety of fields, among which are representative areas: robotics, finance, logistics, smart grids, etc. In robotics, many robots are used to collaborate to perform tasks. In the financial sector, agent-based trading systems conduct market analysis and automatic trading. In the logistics sector, it is used for logistics optimization and supply chain management. In the smart grid, Multi-Agent systems are used for energy management and optimization. These applications demonstrate the flexibility and efficiency of the Multi-Agent system. 

## 6. Implementation examples of Multi-Agent systems 
Examples of implementation of Multi-Agent systems include autonomous vehicle systems, smart home automation systems, and distributed network management systems. In an autonomous vehicle system, each vehicle acts as an agent to optimize traffic flow. In smart home automation systems, consumer electronics act as agents to increase energy efficiency. In distributed network management systems, network equipment acts as agents to optimize network performance. These implementation cases demonstrate the practicality and effectiveness of the Multi-Agent system. 

## 7. conclusion 
The Multi-Agent system is a powerful tool for efficiently solving complex problems. These systems are applicable in a variety of fields, and their flexibility and scalability serve as a great advantage in system design. Effective interaction and cooperation mechanisms between agents are essential for the successful implementation of the Multi-Agent system. In the future, the Multi-Agent system is expected to further develop and contribute to solving various complex problems. 

## 8. references 
-Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg. 
-Weiss, G. (Ed.). (1999). Multiagent systems: a modern approx to distributed material intelligence. MIT press. 
-Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons. 

This report was written to help you understand the overall structure and application of the Multi-Agent system. Each section covers various aspects of the Multi-Agent system and highlights the importance of these systems in performing complex tasks. 
================================================== 

================================================== 
🔄 Node: Supervisor in [ResearchTeam] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
FINISH 
================================================== 

================================================== 
🔄 Node: ResearchTeam 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
================================ Human Message ================================= 
Name: WebScraper 

# How to perform complex tasks using Multi-Agent structures 

## Table of Contents 
One. Introduction 
2. Overview of the Multi-Agent system 
3. Advantages of the Multi-Agent system 
4. Components of the Multi-Agent system 
5. Applications of Multi-Agent systems 
6. Implementation examples of Multi-Agent systems 
7. conclusion 
8. references 

## One. Introduction 
The Multi-Agent system refers to a distributed artificial system in which multiple agents work together to solve complex problems. These systems achieve the goals of the entire system through mutual operation, while each agent works independently. Recently, the importance of Multi-Agent systems is emerging to efficiently perform complex tasks. This report discusses the structure of the Multi-Agent system and how to perform complex tasks that utilize it. In addition, we will look at the practicality of Multi-Agent systems through a variety of applications and implementation examples. 

## 2. Overview of the Multi-Agent system 
The Multi-Agent system consists of several autonomous agents, each of which has the ability to make decisions independently. These systems are suitable for solving problems in a decentralized environment, and each agent performs a specific role or function. Interaction between agents is essential to increasing the efficiency of the system and solving complex problems. Multi-Agent systems are mainly utilized in a variety of fields, including distributed computing, robotics, and network management. The design and implementation of these systems is centered around the cooperation and coordination mechanisms between agents. 

## 3. Advantages of the Multi-Agent system 
One of the main advantages of the Multi-Agent system is scalability. By adding a new agent to the system, it can be easily extended, which increases the flexibility of the system. In addition, parallel processing between agents allows you to maximize the efficiency of your work. These systems have high fault tolerance, minimizing the impact of some agent errors on the entire system. The Multi-Agent system can be solved by dividing complex problems into small sub-problems, reducing the complexity of problem solving. Finally, different agents can have different expertise, allowing comprehensive access to a variety of issues. 

## 4. Components of the Multi-Agent system 
Multi-Agent systems usually consist of agents, environments, and interaction mechanisms. Agents act autonomously, and the environment provides space for agents to interact with. The interaction mechanism enables communication and cooperation between agents, which achieves the goals of the system. Each agent performs three main functions: detection, planning, and action. These components play an important role in the design and implementation of the system, and the efficient integration of each element depends on the performance of the system. 

## 5. Applications of Multi-Agent systems 
Multi-Agent systems are being applied in a variety of fields, among which are representative areas: robotics, finance, logistics, smart grids, etc. In robotics, many robots are used to collaborate to perform tasks. In the financial sector, agent-based trading systems conduct market analysis and automatic trading. In the logistics sector, it is used for logistics optimization and supply chain management. In the smart grid, Multi-Agent systems are used for energy management and optimization. These applications demonstrate the flexibility and efficiency of the Multi-Agent system. 

## 6. Implementation examples of Multi-Agent systems 
Examples of implementation of Multi-Agent systems include autonomous vehicle systems, smart home automation systems, and distributed network management systems. In an autonomous vehicle system, each vehicle acts as an agent to optimize traffic flow. In smart home automation systems, consumer electronics act as agents to increase energy efficiency. In distributed network management systems, network equipment acts as agents to optimize network performance. These implementation cases demonstrate the practicality and effectiveness of the Multi-Agent system. 

## 7. conclusion 
The Multi-Agent system is a powerful tool for efficiently solving complex problems. These systems are applicable in a variety of fields, and their flexibility and scalability serve as a great advantage in system design. Effective interaction and cooperation mechanisms between agents are essential for the successful implementation of the Multi-Agent system. In the future, the Multi-Agent system is expected to further develop and contribute to solving various complex problems. 

## 8. references 
-Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg. 
-Weiss, G. (Ed.). (1999). Multiagent systems: a modern approx to distributed material intelligence. MIT press. 
-Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons. 

This report was written to help you understand the overall structure and application of the Multi-Agent system. Each section covers various aspects of the Multi-Agent system and highlights the importance of these systems in performing complex tasks. 
================================================== 

================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
ResearchTeam 
================================================== 

================================================== 
🔄 Node: Supervisor in [ResearchTeam] 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
FINISH 
================================================== 

================================================== 
🔄 Node: ResearchTeam 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
# How to perform complex tasks using Multi-Agent structures 

## Table of Contents 
One. Introduction 
2. Overview of the Multi-Agent system 
3. Advantages of the Multi-Agent system 
4. Components of the Multi-Agent system 
5. Applications of Multi-Agent systems 
6. Implementation examples of Multi-Agent systems 
7. conclusion 
8. references 

## One. Introduction 
The Multi-Agent system refers to a distributed artificial system in which multiple agents work together to solve complex problems. These systems achieve the goals of the entire system through mutual operation, while each agent works independently. Recently, the importance of Multi-Agent systems is emerging to efficiently perform complex tasks. This report discusses the structure of the Multi-Agent system and how to perform complex tasks that utilize it. In addition, we will look at the practicality of Multi-Agent systems through a variety of applications and implementation examples. 

## 2. Overview of the Multi-Agent system 
The Multi-Agent system consists of several autonomous agents, each of which has the ability to make decisions independently. These systems are suitable for solving problems in a decentralized environment, and each agent performs a specific role or function. Interaction between agents is essential to increasing the efficiency of the system and solving complex problems. Multi-Agent systems are mainly utilized in a variety of fields, including distributed computing, robotics, and network management. The design and implementation of these systems is centered around the cooperation and coordination mechanisms between agents. 

## 3. Advantages of the Multi-Agent system 
One of the main advantages of the Multi-Agent system is scalability. By adding a new agent to the system, it can be easily extended, which increases the flexibility of the system. In addition, parallel processing between agents allows you to maximize the efficiency of your work. These systems have high fault tolerance, minimizing the impact of some agent errors on the entire system. The Multi-Agent system can be solved by dividing complex problems into small sub-problems, reducing the complexity of problem solving. Finally, different agents can have different expertise, allowing comprehensive access to a variety of issues. 

## 4. Components of the Multi-Agent system 
Multi-Agent systems usually consist of agents, environments, and interaction mechanisms. Agents act autonomously, and the environment provides space for agents to interact with. The interaction mechanism enables communication and cooperation between agents, which achieves the goals of the system. Each agent performs three main functions: detection, planning, and action. These components play an important role in the design and implementation of the system, and the efficient integration of each element depends on the performance of the system. 

## 5. Applications of Multi-Agent systems 
Multi-Agent systems are being applied in a variety of fields, among which are representative areas: robotics, finance, logistics, smart grids, etc. In robotics, many robots are used to collaborate to perform tasks. In the financial sector, agent-based trading systems conduct market analysis and automatic trading. In the logistics sector, it is used for logistics optimization and supply chain management. In the smart grid, Multi-Agent systems are used for energy management and optimization. These applications demonstrate the flexibility and efficiency of the Multi-Agent system. 

## 6. Implementation examples of Multi-Agent systems 
Examples of implementation of Multi-Agent systems include autonomous vehicle systems, smart home automation systems, and distributed network management systems. In an autonomous vehicle system, each vehicle acts as an agent to optimize traffic flow. In smart home automation systems, consumer electronics act as agents to increase energy efficiency. In distributed network management systems, network equipment acts as agents to optimize network performance. These implementation cases demonstrate the practicality and effectiveness of the Multi-Agent system. 

## 7. conclusion 
The Multi-Agent system is a powerful tool for efficiently solving complex problems. These systems are applicable in a variety of fields, and their flexibility and scalability serve as a great advantage in system design. Effective interaction and cooperation mechanisms between agents are essential for the successful implementation of the Multi-Agent system. In the future, the Multi-Agent system is expected to further develop and contribute to solving various complex problems. 

## 8. references 
-Jennings, N. R., & Wooldridge, M. (1998). Applications of intelligent agents. In Agent technology (pp. 3-28). Springer, Berlin, Heidelberg. 
-Weiss, G. (Ed.). (1999). Multiagent systems: a modern approx to distributed material intelligence. MIT press. 
-Wooldridge, M. (2009). An introduction to multiagent systems. John Wiley & Sons. 

This report was written to help you understand the overall structure and application of the Multi-Agent system. Each section covers various aspects of the Multi-Agent system and highlights the importance of these systems in performing complex tasks. 
================================================== 

================================================== 
🔄 Node: Supervisor 🔄 
- - - - - - - - - - - - - - - - - - - - - - - - - - - -  
next: 
FINISH 
================================================== 

Copy

output = run_graph(
    super_graph,
    """Topic: How to perform complex tasks using a multi-agent architecture

Detailed guidelines:
- Create a report in the form of an Arxiv paper on the topic
- Create an outline 
- Write more than 5 sentences for each outline 
- If a chart is needed when writing the details, create and add a chart 
- Write a report in Korean 
- Write the sources in APA format - Save the final result as a .md file""",

    recursive_limit=150,
)