Yep, that looks like it should do the trick. I can confirm the triple "false" options to select_shells is correct (I've done some major redesigning on the logic of that function a while ago when dwi2response got some significant changes during the 0.3.16 development days). All "false" options there basically does no automatic stuff, and fully honours the -shell option without constraints; just the way we want it for a generic MSMT-CSD without assumptions about the scenarios that users want to test. I'll merge this in right now, so it's included in RC2 out of the box.