@ErikJonker @tante There is a difference between AI autocomplete and "letting things rip" I think. I would put the line at commits/PRs by an AI client - as this implies limited or not oversight (and a "developer" just pushing yes on most things). But I'm not sure what the final rules will look like.

@ErikJonker @tante Probably not since that is still with a human evaluating. The prompt-spew-commit churning agents are the target I suppose.

@ErikJonker @tante was asking this myself.

because i do sometime use AI to check basic error (double ;;, or thing difficult to see but easy to hunt down by an AI) does that make my code illegal on your new rules ?@Codeberg ?