I’m sure it is. You can probably get the parent-document of a cluster and go from there. But afraid I can’t provide much insight there. I’d probably just get the path on the top-level, and then pass that to the cluster as an input variable.
Edit: Apologies, I totally forgot about this: