Support for A2A communication with Foundry Agents

I would like to configure the SDK to recognize and use Azure Foundry hosted agents as custom subagents instead of Github-hosted agents. Is this possible or what is the best way to think about doing this today?

Use case: I have defined a custom agent with Github Copilot SDK (C#) which behaves similarly to what plan mode does. I've setup this agent as a Foundry Hosted Agent service and would like the agent to leverage Foundry-based agents to execute and perform the work described in the plan (for security reasons, traceability, etc.) instead of proper GH subagents.

In this scenario, I would still like to see GHCP share memory/context with a Foundry agent in a similar manner as GHCP does with subagent today if possible.

[Foundry has A2A support](https://learn.microsoft.com/en-us/azure/foundry/agents/how-to/tools/agent-to-agent?tabs=prompt-agents&pivots=csharp) for its agents in preview. Does the SDK provide a mechanism to leverage other agent runtimes hosted on non-Github infrastructure?

No files, tests, or entry points are identified in the issue. Start by locating how the SDK currently configures custom subagents and shares memory or context, then compare those capabilities with Azure Foundry's preview A2A support. Done would require an agreed integration path and documented behavior for invoking Foundry-hosted agents.

Rédigé par le modèle d'indexation à partir du texte de l'issue.