Rebuff

An open-source prompt injection detection framework that employs a self-hardening mechanism combining heuristic rules with LLM-based dual verification. It is designed to detect and block prompt injection attacks targeting AI applications.