Remove unsupported pre-approval guardrail option
This documents toolExecution.preApprovalInputGuardrails as a working option, but it is not implemented in the SDK: ToolExecutionConfig in packages/agents-core/src/run.ts only defines maxFunctionToolConcurrency, and a repo-wide search finds preApprovalInputGuardrails only in these new docs. In JavaScript this guidance is silently ignored (and in TypeScript the object is rejected), so function-tool approvals will still be created before input guardrails run despite the documentation saying otherwise.
Useful? React with 👍 / 👎.
👍
Originalmente postado por @chatgpt-codex-connector em lhweiler03-sc/openai-agents-js#3 (comment)
This documents
toolExecution.preApprovalInputGuardrailsas a working option, but it is not implemented in the SDK:ToolExecutionConfiginpackages/agents-core/src/run.tsonly definesmaxFunctionToolConcurrency, and a repo-wide search findspreApprovalInputGuardrailsonly in these new docs. In JavaScript this guidance is silently ignored (and in TypeScript the object is rejected), so function-tool approvals will still be created before input guardrails run despite the documentation saying otherwise.Useful? React with 👍 / 👎.
👍
Originalmente postado por @chatgpt-codex-connector em lhweiler03-sc/openai-agents-js#3 (comment)