The paper introduces AI-Infra-Guard, an open-source framework for red teaming AI agents across four layers: infrastructure, protocol and tools, agent behavior, and the underlying model. It matches different testing paradigms to each layer, including deterministic rules covering more than 75 AI components and 1,400 vulnerability rules, LLM-driven auditing of MCP servers and agent-skill packages, multi-turn black-box agent red teaming, and a jailbreak harness with more than 26 attack operators across 16 datasets. The authors position it as a unified foundation for agent security and supply-chain auditing.
No heat snapshots are available in the last 24 hours.