Windows Secures AI Agents: MXC SDK & Agent 365 Revealed

Windows is addressing new security challenges as AI agents become more autonomous and take actions across systems. These agents introduce new risks to control and trust. Developers are building agents that read files, invoke services, and modify environments. This raises the question of how to keep these systems trustworthy when they operate at scale. Security for agents must be built into the foundation by design. Microsoft Agent 365 is expanding its capabilities to discover and manage local agents on Windows. This includes OpenClaw agents, and soon GitHub Copilot CLI and Claude Code. Organizations will also be able to apply policy-based controls to set guardrails for what agents are allowed to do. At Build 2026, an update will be shared on how Agent 365 and Windows are working together with the introduction of the Microsoft Execution Containers, or MXC, SDK. For developers, Windows will provide the building blocks for more secure agents on consumer and enterprise systems. For IT teams, Agent 365 and Windows offer observability, governance, and security capabilities to manage risks. The MXC SDK is being introduced as an early preview. It is a cross-platform, policy-driven execution layer for agents on Windows and WSL. This allows developers to define what to constrain in their apps and agents. This focus on foundational security is crucial for safely integrating AI agents into daily operations.