A minimal, CLI-first Python project that detects prompt injection attacks in LLM applications using a hybrid approach combining rule-based detection, ML scoring, and RAG-specific guards ...