The confidence threshold for triggering guardrail actions based on Automated Reasoning policy violations.
The list of Automated Reasoning policy ARNs to include in the guardrail configuration.