You can think of a distance to any galaxy as some function ##r(t)##, where the distance ##r## increases with ##t## as the universe expands. It's a completely kosher way of thinking about distances, but not always the easiest.
Comoving distance is just a different way of expressing ##r(t)##, where you say ##r(t)=a(t)D##. You make ##D## - the comoving distance - have some arbitrarily chosen value with units of distance, and you relegate the change in time to a completely separate, dimensionless factor ##a(t)## - called the scale factor. So instead of saying a galaxy 1 billion ly away will after time ##\Delta t## have receded to 2 billion ly, you can say that the scale factor after time ##\Delta t## will have grown to twice the initial value.
This is handy, because in an expanding universe the scale factor applies to all distances. You don't have to keep in mind that this galaxy receded from 1 to 2 Gly, while that one receded from 5 to 10, and yet another one from 2.745 to 5.49. Whatever their individual values of ##D##, they are all affected by the same scale factor. All you're saying with each of those example distances anyway is that they've grown by a factor of 2.
All that is left is to assign some particular distance to serve as ##D## for each galaxy. By convention the current distance ##r(t_0)## is used, so that ##r(t_0)=D##. ##D## will stay constant forever. Then the scale factor expresses changes in distances w/r to the current state of the universe. More than one means larger, less than one is smaller.
BTW, comoving distance increases with greater redshifts only in the sense that you're looking at farther galaxies. That's what it's supposed to represent, after all - the relative positions of different galaxies. If you were to track the ##D## for each particular galaxy as the universe evolves in time, it'd stay the same - because the relative positions of galaxies don't change with the expansion, only the scale of distances between them.