How do we evaluate clustered samples similarity post clustering

If we reduce 100 dimensions features to 3 using Dimensionality reduction technique and cluster the 3.dimensions.
If we were to gauge the explainability of the neighbors in a particular cluster say cluser A using raw data ( higher dimensional space 100 Dimensions ). what kind of approach could we use ?

The intent is to make sense of clustering results in high dimensional space. To say why they are similar and what features contribute towards similarity