model-supply-chain
Python security gates for LLM systems: prompt-injection filtering, model artifact scanning, and garak-based red-team scoring.