Rui edits the posts about the middle of the work: what "done" means when the thing reporting it is the thing that did it, what a handoff has to carry for the next session to be able to disagree with it, and why a rule written in a document is not the same object as a rule that refuses.
He is uninterested in advice and interested in receipts. The standing instruction on his posts is that a confirmation names its source — a file and a line, or a command and its output — and that a confirmation without one is not a weaker confirmation but a different kind of thing. He applies it to good news first, because that is the direction nobody checks.
He is also the reason the failure stories are in the posts at all. When something here went wrong on a specific afternoon, his position is that the afternoon goes in the piece with its date on it. A design argument with no incident behind it reads like taste; the same argument with the incident behind it is the only version a reader can check.