Skip to documentation

Reference

CLI reference

Install, inspect, refresh, and run the outbound local runner.

Install and update

The Homebrew release provides the self-contained macOS or Linux executable. The CLI always connects to https://interesta.ai/orchestrator.

brew install interesta-ai/tap/orchestrator
brew upgrade interesta-ai/tap/orchestrator
brew uninstall orchestrator
orchestrator --version

Commands

  1. orchestrator install

    Validate the runner token and prepare Docker, kind, kubectl, trusted runtime images, agent images, and local storage.

  2. orchestrator run

    Validate the cached token, refresh provider catalogs, connect outbound, and execute pushed jobs.

  3. orchestrator providers status

    Show local provider authentication and catalog state without requiring a runner token.

  4. orchestrator providers refresh [provider]

    Refresh all catalogs or one of openai, anthropic, or openrouter, then synchronize sanitized metadata.

  5. orchestrator --help

    Print the supported command usage.

  6. orchestrator --version

    Print the packaged CLI version without starting runtime dependencies.

Credential environment variables

The runner reads provider environment credentials when it starts. It never prints credential values or writes them to an Orchestrator credential file.

OPENAI_API_KEY=...
ANTHROPIC_API_KEY=...
CLAUDE_CODE_OAUTH_TOKEN=...
OPENROUTER_API_KEY=...

Frequently asked questions

Where does the CLI connect?

The official release connects to https://interesta.ai/orchestrator.

Does provider status require a runner token?

No. It reports local authentication and catalog state without requesting the workspace token.

When should I rerun install?

Run it for initial setup and when new or changed runtime images need to be prepared. The connected runner can also prepare newly configured images for later jobs.

How do I refresh one provider?

Run orchestrator providers refresh followed by openai, anthropic, or openrouter.