error occurred while processing message: 'NoneType' object has no attribute 'create' while creating an agent in AutogenStudio

[himanshud2611]

Issue: Error in AutogenStudio Workflow Execution

Steps to Reproduce

1. I set up the environment and other requirements for AutogenStudio.
2. Added a skill named read_invoice with the following code:

import fitz  # Import PyMuPDF

def read_invoice():
    pdf_path = r"D:\Testing\AutogenDemo\sample_invoice_updated.pdf"

    # Open the PDF file
    doc = fitz.open(pdf_path)
    text = ""

    # Extract text from each page
    for page in doc:
        text += page.get_text()

    # Close the document
    doc.close()
    return text

print(read_invoice())

Model Configuration

• Model Config: OpenAI Azure setup with gpt-4o-mini

Agent Configuration

1. read_invoice: Added model and skill as read_invoice (code provided above).
2. write_email: Configured with SYSTEM PROMPT and added model.
3. user proxy: Default configuration.

Workflow Setup

Here is the workflow configuration:

Playground Test

I used the following prompt in the playground:

Read the invoice using the read_invoice agent, and then write a customized email for the customer using the write_email agent.

Error Observed

I encountered the following error:
Error occurred while processing message: 'NoneType' object has no attribute 'create'

Request for Assistance

Could you please help me identify and fix this issue? I can provide more details if required.

Thank you!

---

This formatting makes your message clear, organized, and professional for the GitHub discussion.

[victordibia]

Hi @himanshud2611 ,

Thanks for the note. That is indeed a bug related to the sequential mode implementation for AutoGen Studio v0.2x

At the moment, there is an updated version of AutoGen Studio is being written based on the new v0.4 api (see#4208). It is still early but this is where much of the development effort will be focused going forward.
If you are willing, are you open to building out your use case with the new version and I can provide some support?

For example your use case can be written in code like this...

from autogen_agentchat.agents import AssistantAgent, UserProxyAgent
from autogen_agentchat.task import Console, TextMentionTermination
from autogen_agentchat.teams import RoundRobinGroupChat
from autogen_ext.models import OpenAIChatCompletionClient



async def read_invoice(pdf_url: str = "https://arxiv.org/pdf/2408.15247", max_pages: int = 1) -> str:
    """
    Reads text from a PDF URL. 
    """
    import PyPDF2
    import requests
    import io

    response = requests.get(pdf_url)
    reader = PyPDF2.PdfReader(io.BytesIO(response.content))
    pages_to_read = reader.pages[:max_pages]
    return "".join(page.extract_text() for page in pages_to_read)

async def main() -> None:
    # Define an agent
    invoice_agent = AssistantAgent(
        name="invoice_agent",
        system_message="You are a helpful assistant. Use your read_invoice tool to extract the text from the pdf url",
        model_client=OpenAIChatCompletionClient( model="gpt-4o-mini"),
        tools=[read_invoice],
    )
    email_agent = AssistantAgent(
        name="email_agent",
        system_message="You are a helpful assistant. Please convert the text to a brief email summarizing content.",
        model_client=OpenAIChatCompletionClient( model="gpt-4o-mini")
    )

    user_proxy = UserProxyAgent(name="user_proxy")

    # Define termination condition
    termination = TextMentionTermination("TERMINATE")

    # Define a team
    agent_team = RoundRobinGroupChat([invoice_agent, email_agent, user_proxy], termination_condition=termination)

    # Run the team and stream messages to the console
    stream = agent_team.run_stream(task="Send an email summarizing the pdf https://arxiv.org/pdf/2408.15247")
    await Console(stream)


# NOTE: if running this inside a Python script you'll need to use asyncio.run(main()).
await main()

and in AGS, this can be represented as a declarative specification that can be used in the UI

{
  "version": "1.0.0",
  "component_type": "team",
  "name": "invoice_team",
  "participants": [
    {
      "component_type": "agent",
      "name": "invoice_agent",
      "agent_type": "AssistantAgent",
      "system_message": "You are a helpful assistant. Use your read_invoice tool to extract the text from the pdf url",
      "model_client": {
        "component_type": "model",
        "model": "gpt-4o-mini",
        "model_type": "OpenAIChatCompletionClient"
      },
      "tools": [
        {
          "component_type": "tool",
          "name": "read_invoice",
          "description": "Read text from a PDF URL",
          "content": "async def read_invoice(pdf_url: str = \"https://arxiv.org/pdf/2408.15247\", max_pages: int = 1) -> str:\n    import PyPDF2\n    import requests\n    import io\n    response = requests.get(pdf_url)\n    reader = PyPDF2.PdfReader(io.BytesIO(response.content))\n    pages_to_read = reader.pages[:max_pages]\n    return \"\".join(page.extract_text() for page in pages_to_read)",
          "tool_type": "PythonFunction"
        }
      ]
    },
    {
      "component_type": "agent",
      "name": "email_agent",
      "agent_type": "AssistantAgent",
      "system_message": "You are a helpful assistant. Please convert the text to a brief email summarizing content.",
      "model_client": {
        "component_type": "model",
        "model": "gpt-4o-mini",
        "model_type": "OpenAIChatCompletionClient"
      }
    },
    {
      "component_type": "agent",
      "name": "user_proxy",
      "agent_type": "UserProxyAgent"
    }
  ],
  "team_type": "RoundRobinGroupChat",
  "termination_condition": {
    "component_type": "termination",
    "termination_type": "TextMentionTermination",
    "text": "TERMINATE"
  }
}

cc: @gagb