Last released Sep 14, 2026
Runtime behavioral security for Agentic Skills — traces what a skill actually does, not what it declares, and pauses consequential actions for human authorization before they execute.