Bug: GroupChatOrchestration times out when using AzureAIAgent

[robinscr]

Issue:
I am using GroupChatOrchestration with mutiple agents of different types (ChatCompletion and AzureAIAgent).
When the group chat manager directs a question to the AzureAIAgent it hangs and times out, other agents are working fine.
AzureAIAgent is created in foundry (it has no tools, just a simple instruction) and I am using the PersistentAgentsClient GetAgentAsync() with the AgentID. The agent that is returned has all required details for instructions and description (I am aware missing description can cause issues with GroupChatOrchestration).

Steps to reproduce the behavior:
Create a GroupChatOrchestration with a combination of AzureAIAgent and ChatCompletionAgent
Hit the InvokeAsyc on the orchestration
When the group chat manager asks the AzureAIAgent to contribute, the conversation hangs and times out

Platform

• Language: C#
• Source:
  -- SemanticKernel and Agents.AzureAI version: 1.65.0-preview
  -- Azure.Agents.Persistent version: 1.2.0-beta.4
• AI model: GPT 4.1-mini
• IDE: Visual Studio
• OS: Windows

[nx-s]

(I am aware missing description can cause issues with GroupChatOrchestration)

Not only that, but anything that causes an agent to fail (wrong API key, wrong modelid, an exception during processing) will be silently ignored, and the orchestrator (handsoff, group, etc.) will time out.

So make sure that the agent works in isolation. To find out the reasons for such failures, load symbols for all involved assemblies and debug the code step by step.

[ghord]

Looks like the same issue as #13040