The cockpit: one screen per question
§ 1 — One question per screen
The cockpit is the screen of a person who runs a company of agents. Each view answers one question asked during the day: what is waiting for me? is the machine healthy? what is broken, and since when?
§ 2 — What is waiting for me
On the evening of 15 November the top sentence said « 38 things are waiting for you. Everything else moves without you. » It counts what waits for a person, not what the agents are doing: what limits the company is the queue of decisions the agents cannot take. Fig. 1. When nothing waits, it says so.
§ 3 — Can I trust this figure?
Every figure carries its date. Each source has its gathering cadence (15 minutes for the board, 5 for the teams, one hour for the releases); a block turns « old » once it has gone past twice its cadence; if the last gathering failed, the badge says « Failed » with the date of the data still shown. The word « old » reads in black and white too: colour alone does not carry the information. Fig. 2.
§ 4 — Is the machine healthy?
The page answers first with one sentence, yes or no, then six rows whose detail stays folded. That evening: 13 agent sessions at work, 16 open alerts, 2 of them serious. The system also slows itself down when the machine is overloaded: one launch waited for that reason.
§ 5 — What it does not do
The cockpit reads the operations database and GitHub. It can run only a short, fixed list of local commands, from the machine itself. It never writes to GitHub.
The full story: one local cockpit to run the whole company →