Unsloth adds OS-level code sandboxing on Windows, Linux and Mac; latency claims unverified
Overview
Unsloth now supports OS-level sandboxing on Windows by integrating Microsoft's open-source mxc repository for sandboxed code execution.
The integration adds under 100 ms of overhead, according to the post. A setup guide is available in Unsloth's documentation.
Written by AI from one article, by Unsloth AI; the timeline lists all 3 articles
Check the sources:
Developments
2 developments
- Oct 8, 12:28 PM ET · 1 articleUnsloth adds OS-level sandboxing for tool calls on Linux, Mac and WindowsDaniel Han: Unsloth adds OS-level sandboxing for Linux, Mac, and Windows
- Oct 8, 12:15 PM ET · 2 articlesUnsloth adds Windows OS-level sandboxing using Microsoft's mxcUnsloth AI: Unsloth adds Windows OS-level sandboxing via Microsoft's mxc
Article timeline
The articles in this story. Times are ET.
meng shao@shao__mengUnsloth integrates Microsoft's mxc sandbox for Windows AI agent isolationAIUnsloth has integrated Microsoft's open-source mxc sandboxing system into Windows as an OS-level sandbox for isolating AI agent code execution. Its High mode provides real operating-system isolation that confines tool calls to specified directories, while its Low mode adds language-level checks that block dangerous commands and shell escapes. Both modes also strip secret environment variables and enforce resource limits such as 8GB memory and 600-second CPU time.

Daniel Han@danielhanchenUnsloth adds OS-level sandboxing for Linux, Mac, and WindowsAIUnsloth now supports OS-level sandboxing on Linux via bwrap, on Mac via seatbelt, and on Windows via Microsoft's MXC. Per-tool-call latency is under 100ms across all three, and its software-style sandboxing with regex AST checks adds about 3ms. The Windows integration was built in collaboration with Microsoft.
Unsloth AI@UnslothAIUnsloth adds Windows OS-level sandboxing via Microsoft's mxcAIUnsloth now supports OS-level sandboxing on Windows by integrating Microsoft's open-source mxc repository for sandboxed code execution. The integration adds under 100 ms of overhead, according to the post. A setup guide is available in Unsloth's documentation.

Heat trend
Not enough continuous observations to show a trend yet.