●builderYou can use this benchmark to stress-test the security of your tool-using agents in high-stakes financial environments.
●policyIt provides a way to verify if conversational AI adheres to strict authorization and compliance protocols under adversarial conditions.