@avrldotdev Id tighten the system prompt based on logs and also id first let the ai write the file names need to be edited, then a deterministic rule/system loops through those files and ensures the agent edits them properly. The system checks if the file has been edited and then proceeds
@D_Is_Purple@NathanielRedmon@yacineMTB Luna on Max performance is as good as 5.6-sol med-high if u use spec driven and multi agent flow that includes reviewer and worker agents. So do add that u suck at english and donโt know shit about ai, basically a dumbfuck vibe coder before saying luna on xhigh is shit
@thsottiaux Could u add multi-window chat in codex nd also the ability to open a temp fork chat? sometimes I wanna keep my main chat nd fork chat side by side.
also for the gpt work, could you lower the render area of chat? only reveal the contents upon scrolling up n down.