I don't think you're outing yourself exactly; this is a scale-dependent question. People will give different answers because their experience is coloured by the size of the systems and organisations they have experience with.
At small scale it's fine. Plenty of useful and valuable stuff runs on a handful of servers that have consistent identities etc and people manage them interactively. Process lives in people's heads or (better) a wiki.
At large scale it's a completely bonkers thing to do. There are tons of nodes; it doesn't make sense to mutate just one unless something has gone badly wrong. Interactive login to a production system should be triggering an incident or at least linked to one. Really you shouldn't be mutating systems in any kind of "manual" way because that kind of change is supposed to be locked down to authorised deployment methods. The current state of the system has to be legible to other team members and other teams, and the easiest way to do that is by looking at what's gone through the deployment process, which is usually navigable via a dashboard.
In the middle, it's possible you could have a deployment system that relies on "git clone" with a key on the instance. That would be a little weird because git is not a great way to store or distribute build artefacts. Not crazy though - could make sense in some situations.
At small scale it's fine. Plenty of useful and valuable stuff runs on a handful of servers that have consistent identities etc and people manage them interactively. Process lives in people's heads or (better) a wiki.
At large scale it's a completely bonkers thing to do. There are tons of nodes; it doesn't make sense to mutate just one unless something has gone badly wrong. Interactive login to a production system should be triggering an incident or at least linked to one. Really you shouldn't be mutating systems in any kind of "manual" way because that kind of change is supposed to be locked down to authorised deployment methods. The current state of the system has to be legible to other team members and other teams, and the easiest way to do that is by looking at what's gone through the deployment process, which is usually navigable via a dashboard.
In the middle, it's possible you could have a deployment system that relies on "git clone" with a key on the instance. That would be a little weird because git is not a great way to store or distribute build artefacts. Not crazy though - could make sense in some situations.