Memory in projects
It's great that memory was added in Lumo 2.0, but based on description it's applied only to general chats, not projects. It makes sense to isolate them, but what about using memory inside projects? Instructions inside projects are a good start, but what if you have multiple complex chats, and then in a new one you want to reference something that was discussed in a previous chat inside the project?
I have been using Claude across multiple chats discussing stuff related to a complex project, and its memory is quite useful in this regard (since I do not use it for anything else), since it can get extra context related to my question, that I did not even think about. Seeing that Lumo 2.0 has gotten much better with certain tupes of questions, that I get while coding, I would love to switch to it for this, and rely on memory in similar way.
The problem is that I have other chats not related to the project, so enabling memory and using it same way as Claude, would, probably, "contaminate" the memory with, at least unneeded stuff, or potentially something that would derail the whole context. Manual update of instructions in a project is an option, but the point of memory is automation of that process.
A simpler alternative would be a way to refer specific project chats, but project-isolated memory still feels more convinient for this.