Mitos

Isolated, forkable computers for your AI agents.
Millisecond microVM sandbox forking on Kubernetes.

Mitos SDK: create a microVM sandbox, run code, and fork it into isolated parallel attempts

import mitos

sb = mitos.create("python")              # Ready microVM sandbox (~27 ms warm-claim)
print(sb.exec("echo hello").stdout)      # hello

# Fork into independent siblings to try two approaches at once.
a, b = sb.fork(2)
a.exec("echo conservative > /workspace/plan.txt")
b.exec("echo aggressive  > /workspace/plan.txt")

sb.terminate()

pip install mitos-run (the import stays import mitos), set MITOS_API_KEY from mitos.run, and the SDK talks to the hosted endpoint. No Kubernetes required. The same code runs on your own cluster by setting MITOS_BASE_URL.

Why Mitos

⑂ Live-fork a running VM. N-way copy-on-write fork of a live microVM. Daughters share the parent's memory pages until they write, so each fork lands in a warm, ready environment. Branch one agent into many parallel attempts.
⚡ ~27 ms warm-claim activate. P50 on the bare-metal reference node, reproducible from bench/husk-activate-latency.sh. Full methodology in BENCHMARKS.md.
◎ Open-source, self-hostable, Kubernetes-native. Run it hosted, or on your own KVM cluster where your data never leaves your infrastructure.

Building agents on Hugging Face? Whether you use smolagents, LangGraph, or your own harness: give every agent its own computer, and fork it into parallel attempts in milliseconds. Start with the Mitos cookbook.

→ Get an API key at mitos.run and start building with a $5 credit.

⭐ Star on GitHub · Cookbook · Join the Discord

_{Apache-2.0 · Mitos™ · trademarks}