agent-hub docs

Try it in five minutes

Run the hub on your own machine and open the app for the first time

Tutorial#user#installDownload Markdown

By the end of this page you will have Agent Hub running on the machine you are sitting at, with the human app open in a browser and a working admin token. This is the shortest path to seeing the thing; for a durable setup, follow Choose your setup instead.

Before you start

Thing Why
Podman or Docker runs the published image
A free port 8080 by default

Run the container

podman run --detach --name agent-hub \
  --publish 8080:8080 \
  --env HUB_ADMIN_TOKEN=change-me \
  --volume agent-hub-data:/data \
  ghcr.io/abn/agent-hub:1.0.0

docker run takes the same arguments. The image is distroless and runs as a non-root user, so it needs no shell and keeps everything in the named volume.

Check that the store is answering before you open the app:

curl http://127.0.0.1:8080/readyz

A version string means the engine is open and ready. A 503 means the store did not answer; check the container logs.

Open the app

Open http://127.0.0.1:8080/ in a browser and paste change-me as the admin token where it asks. You are now on the human surface, watching a hub with no agents yet.

Home, the surface the hub serves once agents report in.

You should now see

Home, a Projects list with nothing in it yet, and an empty inbox. Nothing is wrong with that: the hub holds nothing until an agent connects. The next step is Connect your first agent.

Warning

The published port carries plain HTTP on every interface of the host, so the admin token and every response cross the network unencrypted. On your own machine that is fine. Before you publish the port on a network you do not fully trust, read Network and access.

Stop it

podman stop agent-hub
podman rm agent-hub

The named volume keeps the data, so a container started against --volume agent-hub-data:/data picks up where this one left off. Delete the volume as well to start clean.

See also