[Release] microsoft/autogen python-v0.6.2: python-v0.6.2

Microsoft AutoGen python-v0.6.2 introduces streaming tools with `BaseStreamTool` and `StreamWorkbench` support, adds `tool_choice` parameter to `ChatCompletionClient`, and enables `AssistantAgent` with an inner tool calling loop via `max_tool_iterations`. Additional features include OpenTelemetry GenAI traces, Mem0 memory extension, `output_task_messages` flag, GraphFlow improvements, and support for latest models like Gemini 2.5 Flash.

Key Points

• •Streaming tools now supported via BaseStreamTool subclassing and run_stream method, enabling real-time access to inner agent/team events through AgentTool and TeamTool
• •New tool_choice parameter added to ChatCompletionClient create and create_stream methods for enhanced tool selection control
• •AssistantAgent now supports inner tool calling loops via max_tool_iterations parameter, automating iterative tool execution until completion or iteration limit
• •OpenTelemetry GenAI traces integrated (create_agent, invoke_agent, execute_tool) with option to disable via AUTOGEN_DISABLE_RUNTIME_TRACING environment variable
• •output_task_messages flag enables customization of whether input task messages are emitted during run_stream operations
• •Mem0 memory extension added as memory backend option for AutoGen agents
• •GraphFlow improved with activation group support for workflows containing multiple cycles
• •uv version limit removed to support latest development versions
• •Support added for Gemini 2.5 flash stable and o1-2024-12-17 function calling
• •Multiple bug fixes including serialization issues, mutable defaults, and workflow self-loop handling

Workflow Diagram

Concepts