Blog | Tech Blog

openai

Explore posts tagged with "openai" below.

Assert Runner Emits Error Output on MCP Failures in OpenAI Agents SDK

Added runner-level regression tests to verify MCP tool failures produce proper tool_call_output_item events in both sync and streaming paths.

OpenAI Agents SDK MCP Testing Python

February 28, 2026

Fix RunConfig Inheritance for Agent-as-Tool in OpenAI Agents SDK

When an agent was invoked as a tool, it silently lost the parent's RunConfig — breaking tracing, model_provider, and privacy settings in nested agent chains. Fixed with ToolContext propagation.

OpenAI Agents SDK RunConfig Python Open Source

February 11, 2026

Suppress Noisy Pydantic Serialization Warnings in OpenAI Agents SDK

High-frequency pydantic serialization warnings were polluting logs in the Agents SDK core path. Fixed with a targeted model_dump wrapper and fallback strategy.

OpenAI Agents SDK Pydantic Python Open Source

February 9, 2026