[IntelliAvatar]

A runtime layer for AI agents that enforces execution boundaries: traces, replay, and a hard “no” when something unsafe is about to run.