Bevel modifiers on different sets edges that share vertices

I have a right hexaedron (a “cube”, I guess…) and I wanted to dynamically apply a bevel modifier in different sets of edges. For this reason, I have created two Boolean attributes and I apply each to the corresponding edges, which I can then call using Limit Method: Weight on each Bevel modifier. So, I call four of the edges with the most prominent effect “corner edges” and the rest with the most subtle bevel… “edges (edges)”… yeah.

edit: I forgot to mention that, for illustration purposes alone, I have marked the two different sets of edges. The “corner” edges are cyan and the “edge” ones (sneaky phrasing) are magenta.

It seems to me that, no matter the order of the Bevel modifier operations, I get unexpected results: I get either a discontinuous result (first corner, then edge) or a seemingly random pinch (first edge, then corner). “Random”, because as you can see from the screenshot there doesn’t seem to be a way to tell if the pinch will appear on one side or the other, or both for that matter.

I have uploaded two screenshots which hopefully capture the issue.
(P.S.: Seems like I only have the privilege of uploading one picture)

Is this a known limitation of this workflow? Or is there another way to go about it? I believe applying either of the bevels “destructively” as a mesh deformation would circumvent this, but then again I would like to keep it non-destructive.

Maybe like this with vertex groups?

Hope that helps

Thanks for taking the time to reply.

I believe not. From what I understand, Vertex Groups sort of disregard the edge selection they originated from. So, taking the standard Cube as an example, if you, like, make two cross selections in perpendicular directions (i. all the horizontal edges ii. all the vertical edges) and assign each to a distinct vertex group, from what I can tell you essentially get two identical vertex groups. And that’s exactly because during the selection process for each of the groups, all vertices of the cube are involved. Hence “Vertex groups”, I suppose…