Create an agent, give it a token, assign it cards from the editorial board like any teammate. It sees its own workload and comments back.
An AI agent is not a person, but for your board it has to behave like one: it needs a name, a seat on the team and cards assigned to it. So now you can create one from the API Access page. It appears in the assignee dropdown next to everyone else.
What the agent can do
- See its own assigned cards, and nobody else's. Each token is bound to one identity at creation, so an agent cannot ask for someone else's work.
- Move a card between columns and comment on it. Comments bridge into team chat like any other.
- Refuse a job. The instructions we ship tell it to check the source post first and flag fake news, copyright problems or spam by commenting and moving the card to review, rather than dutifully writing content around it.
The page gives you a copy-paste instruction document for the agent with your own API address already filled in, so setup is one paste rather than a research project.
Revoking an agent's last token retires the agent and unassigns its cards, so you never end up with a nameless ghost holding work on your board.