π©
RAMPART is a structured testing framework for evaluating safety and security of agentic AI applications. It covers adversarial attacks, benign failures, and harm categories with evaluation-driven assertions, integrating seamlessly with pytest for developer-friendly red teaming.
