Firecracker microVMs, one POST away
Real Linux VMs behind a typed TypeScript SDK. Create, exec, snapshot, fork — and wake from hibernation in ~0.3 s.
Pro & Max plans · omg_sk_… keys from omg.dev/sandbox/keys
$ bun add @omg-dev/sandbox import { SandboxClient } from "@omg-dev/sandbox" const client = new SandboxClient({ baseUrl: "https://infra.omg.dev", token: process.env.OMG_API_KEY, }) const sandbox = await client.create({ templateId: "react-ts", ports: [5173] }) await sandbox.exec("echo", ["hello from a microVM"]) // → { stdout: "hello from a microVM\n", stderr: "", exitCode: 0 }
Just HTTP underneath — curl the API directly
A whole computer, per request
Sub-second wake
Hibernated VMs resume from memory in ~0.3 s on the next connection.
Snapshot & fork
Snapshot a running VM, fork clones from its lineage.
Baked templates
templateId: "react-ts" boots with deps preinstalled.
Agent presets
claude · codex · opencode · pi · lfg — the agent boots ready.
Preview URLs
Every exposed port gets a public HTTPS URL.
WebSocket shell
Live PTY at wss://ssh-ws.omg.dev/<id>.
Skip the cold start
Templates boot with dependencies preinstalled; agent presets boot with the agent already running. Or bake your own: typed install steps, checks, a start command — one call publishes it.
All templatesconst template = defineTemplate({ id: "my-agent", ports: [3000], install: [run({ command: "bun i -g my-agent" })], start: { command: "my-agent serve", … }, }) await bakeTemplate(client, template) // → "my-agent"
The box
1 vCPU · 1 GB
2 vCPU · 2 GB
4 vCPU · 4 GB
Ubuntu 22.04 · Bun · git · python3 · sshd · pre-warmed package cache