The integration connects centralized governance in Agent 365 with runtime enforcement for MCP servers and tools through Azure API Management for V2 and AI Gateway tiers.
Updated Oct 09, 2026
Version 2.0Thanks, santiagxf , that clears it up. One idea: besides allow and block, a third outcome of "pause and ask for approval in Teams" for sensitive tool calls, with a clear reason code returned to the agent. Is that possible with the current policies?
The current policy doesn't include an approval workflow handled by the gateway but it's something we are currently evaluating. Thanks for the feedback!