Duologs Observatory — Control Pattern
Microsoft moves agent control into a policy-enforced execution layer
Microsoft made Microsoft Execution Containers (MXC) generally available. MXC creates policy-driven containment boundaries for agent workloads, with policy maintained outside the agent's own control. Microsoft frames agent security around containment, identity and manageability.
Duologs Lens
MXC and Duologs address adjacent questions. Containment limits the resources an agent workload can access. Execution authorisation asks whether a specific consequential action against a specific target and parameters carries valid organisational authority before execution. Microsoft has not validated or endorsed Duologs.
Board-level question
Could this happen inside your stack?