Agent Immune

Agent Immune is an adaptive threat intelligence tool designed to provide AI agent security by assessing, scanning, and hardening agent inputs/outputs. It leverages both regex and semantic memory to detect direct and rephrased attacks, handle input/output scanning, enforce rate limiting, and supports multi-turn escalation and prompt hardening. It's suitable for developers and enterprises deploying AI agents who require security capabilities beyond static governance rules, and can be integrated via Python or as an MCP server.