alfe setup --managed

alfe setup --managed runs the connection flow without prompts, so it suits cloud-init scripts, Dockerfiles, and other unattended provisioning. It reads its configuration from environment variables and marks the machine as managed.

Terminal

ALFE_API_KEY=<SETUP_TOKEN> alfe setup --managed

Use a setup token from the agent’s Overview tab. The token decides which agent the machine runs.

Environment variables

These variables configure the run:

Variable	Required	Purpose
ALFE_API_KEY	Yes	The API key the agent authenticates with.
ALFE_API_URL	No	Override the API endpoint. By default it’s derived from the key.
ALFE_MANAGED	No	Set to true to enable managed mode without passing --managed.

If ALFE_API_KEY isn’t set, the command falls back to the key in ~/.alfe/config.toml. If neither exists, it exits with an error.

Managed mode changes runtime sign-in

A machine in managed mode never uses a provider CLI login. A terminal runtime there runs on the agent’s stored credential or on Alfe credits. Managed mode also persists: later alfe setup runs on that machine stay non-interactive.

What it does

Managed setup runs these steps in order:

1. Resolves the API key and writes ~/.alfe/config.toml with managed mode on.
2. Asks Alfe which agent and runtime the key belongs to.
3. Installs the runtime if it’s missing, and installs a diagnostics helper on a best-effort basis.
4. Prepares the workspace and applies the agent’s persona.
5. Installs and starts the gateway daemon as a system service.
6. Registers the Alfe platform MCP server. A failure here doesn’t fail setup.

The result is a connected agent that comes back online after a reboot.

Verify

Confirm the agent is healthy:

Terminal

alfe doctor

You can also check its status in the dashboard.