Re‑architecting AI Agent Security: Tool‑Layer Zero Trust and Audit‑Ready Controls2 min read·7h ago·via InfoQ AI/ML/Data