Definitely acting as an attenuverter with 12 o’clock being close to 0 output, clockwise being the signal attenuated, and counter-clockwise being the signal attenuated and inverted.
I assume the the input of 1 is normalled to the input of 2 if no cable is inserted on input 2 but it’s not clear. If the output of 1 is normalled to input of 2 then you would have to make sure that you don’t want the amplitude of 2 to be greater than that of 1, which might be… well… you might find yourself just using a multiple instead
No mention of DC offsets when no input cable is connected, though the signal flow diagram sort of implies this more than it does any normaling between channels. I suppose it’s possible only input 1 is buffered which is why it’s normalled to 2 and 3 but that is just a wild guess (and maybe ignorant on a technical understanding level!)
Gotta hand it to makers like addac, though, still rolling out these big granular modules knowing clouds 2 is likely imminent, and has been for some time (different features, etc. etc. I know…)