Not sure where else to put this so desktop request seems like the best place.
For Leo, it would be very helpful to have memories able to be selectively accessed per conversation to save on token use in a conversation. Every token saved is processing power saved, time saved. It’s not necessary for my world building conversation to know about my homelab or vice versa. This would allow for greater conversation length as well. Something along the lines of having global memories (things Leo should always know no matter the context) and selective memories (toggleable by conversation), perhaps with a category feature to be able to mass toggle certain memories.