Install wendy-agent

Overview

If you are using a WendyOS device (such as an NVIDIA Jetson flashed with WendyOS), wendy-agent is already installed and running. You do not need to follow this guide.

This guide is for developers who want to deploy to custom hardware that is not running WendyOS, such as:

Intel/AMD NUCs
Remote servers
Raspberry Pi OS (without WendyOS)
Any other Debian-based (Ubuntu) machine

Requirements

A Debian-based Linux machine (Ubuntu 20.04+, Debian 11+, Raspberry Pi OS, etc.)
curl installed

Installation

SSH or log into the target machine and run:

curl -fsSL https://install.wendy.sh/agent.sh | bash

This installs the wendy-agent service, which registers the device on your local network via mDNS so that the Wendy CLI on your developer machine can discover it.

Discovering Your Device

Once wendy-agent is installed and running, go back to your developer machine (where you installed the Wendy CLI) and run:

wendy discover

This uses mDNS to scan your local area network and list all devices running wendy-agent. You should see your newly configured machine appear in the list.

If your device is not on the same local network (e.g., a remote server, cloud instance, or a machine connected via a VPN like Tailscale or Headscale), mDNS discovery will not find it automatically. Instead, you can target the device directly by hostname:

wendy run --device {hostname}

Replace {hostname} with the hostname or IP address of your remote machine. This works with any network topology, including Tailscale, Headscale, WireGuard, or plain SSH tunnels.

Next Steps

Once your device is discoverable, you can deploy applications to it just like any other WendyOS device: