An open-source testing framework for machine learning models that identifies hallucinations, bias, and vulnerabilities.