Started using the Cursor "projects" more lately and really liking them. Doesn't feel like as big of an "oh this is clearly where things are going" as Grok Bot did, but I do think it's vague shape will get more common over time The big single coordinator thread with deep context on the project and "subagents" that are just cloud agents works way better than I expected It's kinda weird to get used to and I expected to loose more work into the void but I really haven't. It's one of the best systems for NOT loosing work (T3 Code's sidebar is another excellent system for this kind of thing). Between the project view, list of prs at the bottom, the list of running agents, it's surprisingly easy to get a clear high level look at what "work" is currently happening It's another layer of abstraction which takes a lot of getting used to. You kinda have to commit to getting it setup for a project to really see the benefits (get the cloud agent env setup correctly), but once u do it's impressive It's like most of the SpaceXAi stuff right now: they have the best products of any lab (cursor, grok bot, origin, etc.) while being stuck with a pretty mid model. They are definitely pushing Grok to it's absolute limit which I respect, but man it's gonna be nice when the model catches up to the products...