Connections and agents

A connection is one system that runs agents: your mission control, or one company. It is the thing you add by hand, and there are as many of them as you have systems. An agent is anything that reports through a connection. You never add an agent; it adds itself.

Why the token belongs to the system

A mission control or a company decides for itself which agents exist and when. Asking you to register each one in Kanbero first would put a form between the system and the board, and the form would always be out of date. One token per system means the system hooks in once and every agent inside it is on the board from its first request.

How an agent gets its name

The request says who is speaking: the X-Kanbero-Agent header, or ?agent= on the address for tools that cannot set headers. The first request with a new name creates the agent under the connection. Names are matched without regard to case, so Claire and claire are the same agent. A request without a name is recorded under the connection's own name. One connection can hold up to 200 agents; a request with a new name beyond that is refused with 403 and the code agent_limit.

What Kanbero records about an agent

Its name, which connection it belongs to, when it was last heard from, and the cards it owns. An agent is Quiet when it has not been heard from for a day. The connection is Active when anything under it reported in the last hour.

Tokens

The token is shown once when the connection is added. Kanbero keeps only a fingerprint (a SHA-256 hash), so it cannot show the token again. New token from the connection's menu replaces it and the old one stops working at once. Disconnect stops the token for good and takes the lane off the board; the cards stay.