Skip to documentation

Getting started

Install the runner

Register the workspace and prepare the trusted local runtime.

Generate the one-time token

Open Runner in the workspace navigation. Owners and admins can generate a token and copy it before leaving the page. The hosted app stores only a SHA-256 hash, not the raw value.

Prepare the runtime

Run the install command and paste the token when prompted. The CLI validates Docker, kind, and kubectl, creates a usable cluster, prepares persistent storage, and pulls the website-specified trusted inference image when it is missing.

brew install interesta-ai/tap/orchestrator
orchestrator install

Start the runner

The run command validates the cached token, prepares the website-specified images, refreshes provider catalogs, opens an outbound protocol-v9 WebSocket, and waits for workspace jobs.

orchestrator providers status
orchestrator run
Install the CLI runtime, inspect provider status, and confirm the runner is online.
Read transcript

The demo shows the Runner page, the Homebrew installation commands, provider status, the run command, and the website changing to an online runner state.

Frequently asked questions

Where is the runner token cached?

The CLI stores it in an owner-only file under the local .orchestrator directory after successful validation.

Does the runner open an inbound port?

No. It connects outbound to the hosted control plane and waits for pushed jobs.

Can one runner serve multiple repositories?

Yes. One workspace runner serves the enabled repositories in that workspace.

What if the runner shows offline?

Restart orchestrator run on the registered machine, confirm network access, and verify that the cached token is still valid.