[Release] microsoft/autogen python-v0.5.7: python-v0.5.7

Microsoft AutoGen python-v0.5.7 release introduces improvements to AzureAISearchTool with unified search methods (full-text, vector, and hybrid), enhanced SelectorGroupChat with model_context parameter for better speaker selection, and improved OTEL tracing with additional metadata. The release also includes various bug fixes and improvements to agent runtime, documentation, and integrations with external services like Anthropic Bedrock and MCP servers.

Key Points

• •Unified Azure AI Search Tool API with three methods: `create_full_text_search()` (supporting simple, full, and semantic query types), `create_vector_search()`, and `create_hybrid_search()`
• •Migrate from deprecated `create_keyword_search()` to `create_full_text_search()` with `"simple"` query type for backward compatibility
• •Added client-side embeddings support to Azure AI Search Tool with automatic fallback to service embeddings when client embeddings unavailable
• •Enhanced `SelectorGroupChat` with new `model_context` parameter to customize messages sent to model client for improved long-context speaker selection
• •Improved OTEL tracing in `SingleThreadedAgentRuntime` with new metadata and message content fields for better observability
• •Added ability to register Agent instances in Agent Runtime for improved agent lifecycle management
• •Fixed critical issues: AnthropicBedrockChatCompletionClient import errors, MCP server tool exception serialization, and Mistral name field handling
• •Enhanced accessibility with improved header icons focus and hover styles
• •Added GPT-4o search capability and Chainlit integration sample for Core API

Concepts