multi-agent python code

Question

I would like to run a multi-agent setup using a named workflow via the Python SDK. The code uses openai_client.conversations.create() and openai_client.responses.create() with streaming, and relies on ResponseStreamEventType to handle streamed events. But I got Error message:

cannot import name 'ResponseStreamEventType' from 'azure.ai.projects.models'

(d:\my_env\lib\site-packages\azure\ai\projects\models_init_.py)

Answer 1

While I am foundry owner but
I still got HttpResponseError: (UserError) Identity(object id: d32d8048-dccd-4e16-8ffb-fe14bdc4f634) does not have permissions for Microsoft.MachineLearningServices/workspaces/agents/action actions. Please refer to https://aka.ms/azureml-auth-troubleshooting to fix the permissions issue. Code: UserError Message: Identity(object id: d32d8048-dccd-4e16-8ffb-fe14bdc4f634) does not have permissions for Microsoft.MachineLearningServices/workspaces/agents/action actions.

Answer 2

ResponseStreamEventType is not part of azure.ai.projects.models, which is why the import fails. The Python SDK surface for multi-agent and workflow scenarios in the provided context uses:

• AIProjectClient from azure.ai.projects for multi-agent setups.
• Streaming patterns that do not rely on ResponseStreamEventType from azure.ai.projects.models.

To build a multi-agent setup with the current SDK surface, use the Connected Agents pattern instead of trying to import ResponseStreamEventType:

1. Initialize the project client:

      import os
      from azure.ai.projects import AIProjectClient
      from azure.ai.agents.models import ConnectedAgentTool, MessageRole
      from azure.identity import DefaultAzureCredential
      
      project_client = AIProjectClient(
          endpoint=os.environ["PROJECT_ENDPOINT"],
          credential=DefaultAzureCredential(),
      )
   

2. Create specialized agents and connect them via ConnectedAgentTool:

      stock_price_agent = project_client.agents.create_agent(
          model=os.environ["MODEL_DEPLOYMENT_NAME"],
          name="stock_price_bot",
          instructions=(
              "Your job is to get the stock price of a company. "
              "If you don't know the realtime stock price, return the last known stock price."
          ),
      )
      
      connected_agent = ConnectedAgentTool(
          id=stock_price_agent.id,
          name=stock_price_agent.name,
          description="Gets the stock price of a company",
      )
      
      main_agent = project_client.agents.create_agent(
          model=os.environ["MODEL_DEPLOYMENT_NAME"],
          name="my-agent",
          instructions="You are a helpful agent, and use the available tools to get stock prices.",
          tools=connected_agent.definitions,
      )
   

3. Create a thread, add a message, and run the main agent. The main agent orchestrates the connected agents and aggregates their responses.

For workflow-style multi-agent orchestration with streaming, use the Microsoft Agent Framework with FoundryChatClient, FoundryAgent, and workflow builders. Streaming is handled via AgentResponseUpdate events rather than ResponseStreamEventType:

• Create a shared FoundryChatClient.
• Create specialized agents.
• Connect them in a WorkflowBuilder.
• Run the workflow with stream=True and handle AgentResponseUpdate chunks (for example, reading update.text).

This approach avoids importing ResponseStreamEventType entirely and uses the supported streaming and multi-agent patterns in the current SDK and Agent Framework.

---

References:

• Build collaborative, multi-agent systems with Connected Agents (classic) (python)
• Agents in Workflows (programming-language-python)
• Microsoft Agent Framework Workflows - Using Workflows as Agents (programming-language-python)
• AutoGen to Microsoft Agent Framework Migration Guide