Applied Audit 02: OpenAI Agents SDK — Governance Conformance Audit under the Vertical AI Governance Stack v1.0

This technical report presents Applied Audit 02, an assessment of the OpenAI Agents SDK under the Vertical AI Governance Stack v1.0. It examines framework-native capabilities and the additional controls required to establish enforceable governance boundaries in agentic systems.

The audit evaluates selected execution and authorization scenarios involving guardrail timing, parallel execution, handoff authorization, and OpenTelemetry tracing. It uses defined test vectors and a governance comparison matrix to distinguish observable framework behavior from controls that must be supplied by an implementing organization.

The report identifies implementation-level governance requirements and verification boundaries relevant to high-assurance agentic architectures. Its findings are limited to the assessed framework version, test configurations, and experimental scope. The assessment does not constitute independent certification or establish universal security or regulatory compliance guarantees.