[Release] openai/openai-agents-python v0.15.2: v0.15.2

OpenAI Agents Python v0.15.2 introduces context management model settings and fixes critical issues with conversation session replay, tool execution, and error handling. The release includes 11 bug fixes addressing assistant conversation item IDs, function tool tracing, MCP tool metadata isolation, and stream terminal rejection. Comprehensive test coverage improvements ensure reliability across realtime tool behaviors, handoff mechanisms, and tool identity helpers.

Key Points

• •Add context management model setting for improved configuration flexibility
• •Fix assistant conversation item ID replay in OpenAIConversationsSession to prevent duplicate processing
• •Redact sensitive function tool trace span errors and MCP invalid JSON errors when logging is disabled
• •Reject failed response stream terminals to prevent invalid state propagation
• •Implement session suffix matching for precise rewind operations
• •Block disabled function tools before execution to enforce tool access controls
• •Isolate merged MCP tool metadata to prevent cross-tool contamination
• •Make ToolContext hashable for consistency with RunContextWrapper
• •Filter custom_tool_call types in remove_all_tools handoff filter
• •Expand test coverage for realtime tool timeouts, serialization edge cases, and RunHooks callbacks

Workflow Diagram

Concepts