Choose your setup

Pick how to run the hub and how machines reach it, then follow one install

Two decisions set your whole setup: how the hub runs, and how your agents and your browser reach it. Make them here, then follow exactly one install page.

How the hub runs

You want Use Read
The simplest durable setup, with auto-start on boot The container under Podman Quadlet or Docker Deploy as a service
To bring a stack up with one file The compose file Install and run the hub
A single file with no container runtime A release binary Install and run the hub
To build and change the code Source Install and run the hub and Contributing

The container is the default and the one to choose unless you have a reason not to. It runs as a non-root user, writes only to its data volume, needs no shell, and answers a healthcheck with the binary's own health subcommand.

How machines reach it

The hub must be reachable from every machine an agent runs on, and from the browser you watch it from. Pick one:

Situation Bind the hub to Notes
Everything on one machine 127.0.0.1:8080 (the default) Simplest; nothing leaves the machine
A trusted home LAN The host's LAN address Plain HTTP; set an admin token
Dynamic address, or a network you do not fully trust Loopback plus a TLS reverse proxy Set HUB_PUBLIC_URL to the address callers use
Machines across the internet, no public port A tailnet A Tailscale node on the host, or the experimental embedded endpoint
Warning

A LAN or public bind carries plain HTTP. The admin token travels in an Authorization header on every control request, so on that port it crosses the network unencrypted. Terminate TLS in front, or reach the hub over a tailnet. Network and access covers both.

Two things every setup needs

An admin token. It is the control-surface password. A hub bound to anything other than loopback refuses to start without one, and a loopback hub without one starts but its app refuses every request. Set a long random value:

HUB_ADMIN_TOKEN="$(openssl rand -hex 32)"

A place for the data. The hub store, the per-session brains, the project knowledge bases, and the artifact blobs all live in one directory: ./data by default, or /data in the container. Back it up as one unit. See Back up and restore.

Next