offensive-ai

`offensive-ai` is a Python-based toolkit designed for authorized red-team engagements, integrating classic network reconnaissance with AI/LLM vulnerability testing. It specializes in probing AI/LLM endpoints for OWASP LLM vulnerabilities, scanning Model Context Protocol servers for known CVEs, and conducting comprehensive security assessments across various infrastructures. Notable features include a configurable CLI, Docker support, and capabilities for specific attack vectors such as blockchain, Kubernetes, and model context attacks, all requiring explicit authorization for usage.