Sistemo docs¶

Run AI agents and untrusted code in real Firecracker microVMs on Sistemo Cloud — each sandbox is a full VM with its own kernel, not a shared-kernel container.

from sistemo import Sandbox

with Sandbox() as sb:                 # real microVM, booted on demand
    print(sb.run("uname -a").stdout)

Start here¶

Cloud quickstart

Get an API key, run code in minutes. Nothing to operate.

Cloud quickstart
SDKs

Zero-dependency clients for Python and JavaScript/TypeScript.

SDKs
REST API

Any language. The SDKs are thin wrappers over this.

API reference
Usage & billing

Prepaid wallet — €5 free, pay only for what runs.

Billing

Build something¶

Run an AI agent's code

Execute LLM-generated code safely, capture the output, throw the VM away.

Cookbook
Connect an LLM

Wire a sandbox into Claude or OpenAI so your agent can run what it writes.

Connect an LLM
Snapshots

Point-in-time disk undo, restore, or clone a new machine.

Snapshots
Browser terminal

Interactive shell into a running machine from the dashboard.

Terminal

New here?¶

What is Sistemo — the 10-line version and how the pieces fit.
Core concepts — machine vs sandbox, microVM, lifecycle, volumes.

SDKs are open source (Apache-2.0) on GitHub · Reading this with an AI agent? See /llms.txt.