Managing from the CLI

The Alfe CLI connects an agent from a terminal and keeps it running. It installs a small, always-on gateway daemon that keeps the agent connected to Alfe.

• On a self-hosted agent, you run the CLI on your own machine or server.
• On a managed agent, Alfe runs the same CLI on the machine it provisions, so you rarely need to touch it.

Install it with:

npm

npm i @alfe.ai/cli -- -g

pnpm

pnpm add @alfe.ai/cli -g

bun

bun add @alfe.ai/cli -g

Connect a self-hosted agent

alfe setup connects an agent on the machine you run it on. It needs that agent’s key, which you generate on the agent’s Overview tab.

Terminal

ALFE_API_KEY=<SETUP_TOKEN> alfe setup

It’s safe to re-run at any time to repair the connection or apply a runtime change.

Provision without prompts

alfe setup --managed runs the flow without prompts, reading its configuration from environment variables. Alfe uses it to bring a managed machine online, and it also suits your own unattended provisioning.

Terminal

ALFE_API_KEY=<SETUP_TOKEN> alfe setup --managed

Day-to-day commands

Once an agent is connected, a few commands cover most operation:

Command	What it does
alfe doctor	Check configuration, the API key, the daemon, and the runtime.
alfe whoami	Show the connected identity and runtime.
alfe gateway status	Show whether the daemon is connected.
alfe gateway restart	Restart the daemon.
alfe gateway logs	Tail the daemon logs.
alfe runtime switch <runtime>	Move the agent to another runtime.
alfe remove	Uninstall Alfe’s integrations and revert their config.

Configuration lives at ~/.alfe/config.toml, and setup writes it for you. See the CLI reference for the full command set.

Work without a terminal

You don’t need the CLI to run a managed agent. Creating one in the dashboard handles all of this, and the agent’s Maintenance controls cover restarts. Agents can also work for themselves over MCP and the Agent API.