Long one, grab some a drink of choice. Trying to figure out a best practice for artists that is not too complex and also does some basic amount of reasonable optimization for the atlas. Problem space:

Working art - PSD where the artist creates
Spine Source Art - PSD with flattened working art images ready for spine
Spine export - PNG atlas

We often have working art delivered as source art that are the original size (maybe 2x sometimes 4x the target size.) The spine animator has two options, option A) work at the original size and it happens to fit within the maximum spine canvas. Option B) scale the source art down 2 or 4x and animate at a smaller scale. Option C) ?

Now add in effect images like glows, sparkles, glints, shines to the mix. Those are often created at the original working size too, however they take up so much pixel room that could ideally be given to the actual art that is seen 90% of the time. A trick I'd like to use is scale the effect down by 50, 75, or even 90% in some cases since they aren't on the screen very long we can kinda "get away with it." This however is a bit of work on my part to rescale the attachment and or mesh back to the needed original desired size (but at a lower resolution) in the spine project after the animator has worked with it.

But if I do that (technical artist) and then later the art needs to update or a polish pass happens, they don't know how much I reduced their effect texture down by. If I told them, then they'd have this really tiny layer that doesn't do them much good in context of their working file but does give more pixel density to the best parts of their art and helps reduce atlas size overall.
When they reexport their new changes now the spin source art is back at the original working size and I need to go back through and resize the effects and less important art again.

I don't doubt I'm missing something simple here or over thinking it. It's almost as if I'd want to be able to set a scale per image in the spine texture export settings. Some manifest that lists all exporting textures and a scale value next to each one.

Should artists deliver their working art at it's working size for spine or reduce their work before hand to be more inline with the target resolution and reduce their effects even smaller still? Should a middle-person do this and deal with resizing churn? How does this all play in if we want to release 4k textures later, then all the mesh/attachment slots will need readjusting again.

Thanks for any insights anyone has.