Always running.
Close the laptop. Your agents keep working on your server.
Each daemon is isolated, runs the app it builds, and you reach it from any browser and your mobile.
$ daemons run --forever
Close the laptop. Your agents keep working on your server.
A real terminal in any browser, your phone included.
Your code, prompts, and data stay on your Hetzner server. Nothing is stored on ours.
How it works
Paste your Hetzner API token. daemons.run buys and provisions a server in your account.
Pick a coding agent, name your daemon, and it boots in seconds. It is isolated, with its own disk.
Your daemon works continuously. Terminal, chat, files, logs and a live preview URL, from any browser.
Run several daemons on one cheap Hetzner server you own. They share its CPU and memory. Each keeps its own disk, its own processes, and its own data.
Run several daemons on one server you own. Each has its own container, disk, and network.
Each daemon can expose an Agent API. Your product sends a prompt and streams the answer back.
POST /api/v1/daemons/{daemon}/tasks
{
"prompt": "Add rate limiting to the API.",
"agent": "claude-code",
"limits": { "minutes": 30 },
"artifacts": ["diff", "test-report"]
}
POST a task to the daemons.run API with a prompt, an agent, limits, and the artifacts you expect.
Every port gets an HTTPS preview URL, protected by default. Nothing is public by accident.
Open the daemon workspace in a full editor. The same files the terminal and the agent use.
A real terminal in any browser. Sessions survive disconnects, so you can close the tab.
Agents keep working after you close the laptop. Daemons are persistent processes on your server, not sessions in a tab. Come back hours later and the work moved on.
A real terminal in any browser, desktop or phone. Works even on slow or unstable connections. Preview URLs let you open what the agent built from anywhere.
Everything lives on your own Hetzner server: code, prompts, files. We never look at your data and store none of it. Destroy the server and it is all gone.