Skip to content
#

AI Sandbox

ai-sandbox logo

An AI sandbox is a secure, isolated runtime environment designed to execute AI-generated code, run autonomous agents, and evaluate large language model (LLM) outputs without risk to the host system. Sandboxes prevent untrusted code from accessing the network, filesystem, or other sensitive resources. Common approaches include container-based isolation (Docker, gVisor), WebAssembly runtimes, and MicroVM hypervisors. AI sandboxes are foundational to agentic AI systems where LLMs write and execute code on behalf of users.

Here are 38 public repositories matching this topic...

Lightweight MicroVM engine built on Cloud Hypervisor. Features include OCI and cloud image support, instant snapshot and clone via reflink, Windows 11 guest support, CNI networking with TC redirect, memory balloon, hugepages, and a Docker-like CLI. Designed for AI sandboxing, cloud desktops, and ephemeral dev environments.

  • Updated Aug 2, 2026
  • Go

Open-source agent sandbox infrastructure for parallel coding agents: isolated workspaces, MCP/CLI control, observability, and atomic publication.

  • Updated Aug 2, 2026
  • Rust

Agent Reference Stack for Kubernetes (kars) - an open source stack from Microsoft for running AI agents safely on Kubernetes. Multi-runtime, Foundry-aware, hardened per-agent sandboxes, governed egress, end-to-end encrypted inter-agent mesh.

  • Updated Aug 2, 2026
  • Rust

A modular, distributed AI runtime for personal automation, research, and autonomous workflows. Built with Node.js and Python, it connects Android (Termux) edge instances with Proxmox core servers. Features multi-project context memory, a modular tool sandbox, and an organic, trace-based reasoning policy layer.

  • Updated Aug 1, 2026
  • JavaScript