MAESTRO: threat-modeling AI agents in seven layers

A plain look at CSA's MAESTRO, a layered method for finding what can go wrong in an agentic AI stack before it ships.