Agent DailyAgent Daily
releaseintermediate

[Release] microsoft/autogen python-v0.7.5: python-v0.7.5

By ekzhugithub
View original on github

Microsoft AutoGen releases python-v0.7.5 with 18 bug fixes and improvements including support for linear memory in RedisMemory, thinking mode support for Anthropic client, fixes for streaming response handling, GraphFlow cycle detection, and security enhancements with DockerCommandLineCodeExecutor. The release addresses issues with Bedrock streaming, JSON schema conversion, token counting, and MCP session management.

Key Points

  • Fixed Bedrock streaming response handling for tool usage with empty arguments, improving compatibility with AWS Bedrock models
  • Added linear memory support in RedisMemory, enabling more efficient memory management for agent state persistence
  • Implemented thinking mode support for Anthropic client, allowing extended reasoning capabilities with Claude models
  • Fixed message ID correlation in streaming chunks to ensure proper message tracking and reconstruction
  • Resolved GraphFlow cycle detection recursion state cleanup to prevent infinite loops in agent workflows
  • Added security warnings and defaulted to DockerCommandLineCodeExecutor for safer code execution environments
  • Fixed Redis caching logic that was incorrectly returning False for string values, restoring proper cache functionality
  • Added reasoning_effort parameter support for OpenAI GPT-5 models, enabling advanced reasoning configurations
  • Fixed JSON schema conversion for nested objects in array items, improving tool definition handling
  • Resolved OllamaChatCompletionClient component loading by adding provider to WELL_KNOWN_PROVIDERS registry

Found this useful? Add it to a playbook for a step-by-step implementation guide.

Workflow Diagram

Start Process
Step A
Step B
Step C
Complete
Quality

Concepts