Microsoft releases container isolation solution for AI agents
Start here
Microsoft announced general availability of Microsoft Execution Containers (MXC), providing a policy-driven execution isolation layer for AI agents that restricts the files and network resources accessible to model-generated code, plugins, tools, or the entire agent, with policies enforced by the container at runtime and the agent itself unable to escalate its own privileges. MXC offers four isolation backends: process containers (Windows 11, macOS, Linux), session containers supported only on Windows, WSL containers, and experimental MicroVM. Another report says Microsoft released an open-source cross-platform sandbox library, MxC, that provides isolated environments for agents on Windows, macOS, and Linux, using processcontainer, bubblewrap, and seatbelt underneath for sandboxing; the solution targets developers managing multiple sandboxes for agents, deeply integrates with agent ID on Windows, and covers process, session, and micro VM levels.
Generated by AI from reports · Updated 4 days ago
Progress timeline
- X @simonwSelectedMicrosoft open-sources cross-platform sandbox library MxC, providing Windows, macOS, and Linux isolation solutions for agents
Microsoft released an open-source cross-platform sandbox library, MxC, that provides isolated environments for agents on Windows, macOS, and Linux. Simon Willison says it uses processcontainer, bubblewrap, and seatbelt underneath for sandboxing. The cited @kdaigle says the solution targets developers managing multiple sandboxes for agents and deeply integrates with agent ID on Windows, covering process, session, and micro VM levels.
- Microsoft Windows Developer Blog微软发布 Microsoft Execution Containers，为 AI Agent 提供策略驱动的执行隔离
微软宣布 Microsoft Execution Containers（MXC）正式可用，为 AI Agent 提供策略驱动的执行隔离层，限制模型生成代码、插件、工具或整个 Agent 可访问的文件与网络资源，策略由容器在运行时强制执行，Agent 自身无法自行扩权。MXC 提供进程容器（Windows 11、macOS、Linux）、仅 Windows 支持的会话容器、WSL 容器和实验性的 MicroVM 四类隔离后端，开发者通过统一的 JSON 配置和多语言 SDK 声明工作负载所需资源。策略覆盖隔离环境、进程、文件系统、网络和用户界面，并支持 Enforcement、Learning、Permissive 三种运行模式，其中 Learning 模式会生成 JSON 活动报告以帮助编写最小权限策略。
Related discussion
Attention trend
New sources per day
- Oct 8New2 sources (2 outlets,0 accounts)
- Oct 9New0 sources (0 outlets,0 accounts)
- Oct 10New0 sources (0 outlets,0 accounts)
- Oct 11New0 sources (0 outlets,0 accounts)
- Oct 12New0 sources (0 outlets,0 accounts)
- Oct 13New0 sources (0 outlets,0 accounts)