prompt-defense
Risk-adaptive prompt-injection defense layer for commercial APIs and local LLMs.
AI security engine — detects prompt injection across 13 attack categories. 61 patterns, zero dependencies.
