microsandbox

Cellar helps you run isolated microsandbox VMs on your own infra today.

Sandbox for AI coding agents (Claude Code, opencode, pi): runs each project in an isolated microVM with host-side secret injection. Linux/KVM & macOS.

Run OpenClaw, Hermes and your own agents in microsandbox microVMs. One reviewable TOML role sets image, allowed domains and host-bound secrets. No daemon.

🧱 A curated list of SDKs, integrations, tools, and resources for microsandbox

Community-maintained Ruby bindings for microsandbox — hardware-isolated microVM sandboxes for AI agents and untrusted code.

Benchmarked agent execution runtime for Python. Sub-10ms cold starts, real-time streaming, time-travel debugging, and self-growing tool libraries. Compare 3 sandbox backends: Docker (OpenSandbox), MicroVM, and in-process AST.

Unofficial testbed for display and GPU support in microsandbox microVMs on Apple Silicon, with an Omarchy (Hyprland) desktop as the workload.

A daemon that manages a GitHub Actions runner scale set and runs every job in a fresh, ephemeral microsandbox microVM.

A Docker sbx-style terminal dashboard for microsandbox — manage microVM sandboxes visually.

Computers for your agents. Run Linux sandboxes locally or remotely, with a desktop and computer use.

🥾 base images that bundle an init system

.NET SDK for microsandbox

Sandboxes à la carte

Project-based sandbox manager for coding agents, with a keyboard- and mouse-driven Go TUI.

OpenCode image, for Docker and microsandbox

Microsandbox integration for Deep Agents

A cloud sandbox for coding agents, on your own server. Start a run from the dashboard, from any agentic IDE (Cursor, Claude Code, Codex, VS Code), or from a script — it runs in an isolated KVM microVM and streams back live.

🍻 homebrew formula for microsandbox