It only outputs indexes that have common first seek columns. Indexes that have common first seek columns are dupes or might be merge-able, or the one of the indexes might be able to get their seek columns rearranged so you can have more index usage possibilities without adding more indexes.
I would only intentially have indexes with common first seek columns on tables with hardly any write activity.
The only other reason to dupe on first seek columns is to dupe on PKs that experience a lot of multi-row up-stream (from many to one; Where is on the many table) joins that cause bookmark lookups. PKs (also UQ constraints) can't have include columns, so you have to create another nonclustered index that has the seem seek as the PK and then add include columns.