Agent DailyAgent Daily
releaseintermediate

[Release] openai/openai-agents-python v0.13.0: v0.13.0

By seratchgithub
View original on github

OpenAI Agents Python v0.13.0 is a minor release that updates the default Realtime websocket model to gpt-realtime-1.5, expands MCP server capabilities with resource management and session tracking, and introduces opt-in reasoning content replay for chat completions. The release includes critical runtime stability fixes for concurrent writes, tool removal edge cases, and batch executor races, with no breaking changes.

Key Points

  • Default Realtime websocket model upgraded to gpt-realtime-1.5 for improved performance without requiring configuration changes
  • MCPServer now exposes list_resources(), list_resource_templates(), and read_resource() methods for enhanced resource management
  • MCPServerStreamableHttp exposes session_id to enable resumable HTTP sessions across reconnects and stateless workers
  • New opt-in reasoning_content_replay feature via should_replay_reasoning_content for improved provider-specific reasoning and tool-call continuity
  • Fixed concurrent first-write race condition in SQLAlchemySession for thread-safe memory operations
  • Resolved remove_all_tools() bug that left orphaned MCP and reasoning items in the system
  • Fixed compaction logic to strip orphaned assistant message IDs after reasoning removal
  • Corrected function-tool batch executor race condition that could cause eager-task conflicts
  • Preserved streamed output guardrail tripwires in the run loop to maintain safety constraints
  • Added runtime behavior probe skill and improved CI/CD with parallelized code-change verification

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