Post

they’re pretty much working independently, each one is assigned to review a single api endpoint and write their findings to an md file in one big folder. afterwards 1 agent reviews all the relevant findings