Agents V2 Py

This wraps the Azure AI Projects SDK to spin up container-based agents in Azure AI Foundry using ImageBasedHostedAgentDefinition. You build a Docker image, push it to ACR, set up pull permissions, then call create_version with your container registry path, resource specs (CPU/memory), and environment variables. It handles versioning, protocol configuration (RESPONSES v1), and tool integration like code_interpreter or MCP servers. The setup is heavier than serverless options but gives you full control over the runtime environment. Useful when you need custom dependencies, specific Python versions, or want to package complex agent logic that doesn't fit standard deployment models. The SDK requires 2.0.0b3 minimum and leans heavily on DefaultAzureCredential for auth.