Any machine you reach is also an always-on automation backend. Schedule a script, keep its credentials encrypted on that machine, call any API, and manage it all from your laptop or your AI tool.
Every command works on a remote machine. The name comes first.
A job runs on a machine you own, on a schedule you set, whether or not your laptop is open. The daemon runs it, not your terminal.
API keys and tokens are encrypted at rest on the machine that uses them, keyed from its own identity. Readable from the command line, from scripts, and by your AI tool, never in plaintext on disk.
Scripts get a fetch-style HTTP client with bearer tokens pulled straight from the secrets store. No curl, no dependencies, no third-party automation service holding your keys.
Ready-made jobs you switch on with one command: a health monitor, a security baseline, fleet-wide log search, Gmail triage with a morning briefing. The list.
A key-value store and time series on every machine, so a job can remember what it saw last time and you can query the result later.
From zero to AI-triaged email, every morning, on a machine you own.
Encrypted on the remote machine. They never leave it in plaintext.
Registers the schedule, wires up the script, and starts it.
Unread mail sorted into urgent, action and FYI. Routine mail marked read. Or run it now.
The Gmail side needs a one-time OAuth setup. The walkthrough and the script itself are in the docs: email monitor setup.
Work that keeps running on your own machine after you close the laptop.
A scheduled job runs on a machine you own, on a schedule you set, using credentials that never leave it. It can call a model, act on the answer, and store the result, without a session open and without a third-party service holding your keys. Your AI tool can create, inspect and change schedules directly.