Yeah to dig into when it’s possible to put the prior in different places, search for various takes on ‘change of variables in statistics’ or something.
There’s stuff in the manual that I like cause I think it makes the computational aspects pretty clear (it’s hard for me to understand the pure statistics stuff sometimes): 10.1 Changes of Variables | Stan Reference Manual
Yeah, I guess this seems crude, but I think this is the state of things. Guess and check with the logit scale until you get some sort of reasonable looking prior-predictives on an interpretable scale.
Nothing more automated than that.