Owner-edge runner

An owner-edge runner is a machine you own and operate that NOME dispatches build, test, and (when scoped) deploy jobs to. It runs your toolchain on your network. NOME provides scheduling, cache, review, and the gate; you provide the hardware. Owner-edge runs consume no NOME compute credits.

Install the NOME runner on the machine, then pair it to your account: the machine mints a one-time pairing token, you claim it from a signed-in NOME client, and you confirm on the machine. A host-scoped device token authorizes that one machine as a runner — it is not your account password and carries no provider credentials.

To keep an owner-edge runner available across reboots, run the agent under a process supervisor (for example systemd on Linux) with a restart-on-failure policy. Inspect health and logs to confirm it is online. For a machine behind NAT, reach it over a private network (Tailscale) or SSH.

Build and test jobs run in ephemeral workspaces that are cleaned up at the end of the job, with cache restored from the project-isolated store. Deploy-scoped runners are protected and tagged separately so a build/test runner can never run a deploy. See Security.

When you would rather not operate a machine, select a NOME cloud runner: managed, isolated per job, torn down when the job ends. Cloud runners bill in NOME credits; owner-edge runners do not. You can mix both — owner-edge for routine builds, cloud for burst.