The concept of the repository is difficult to distinguish from other kinds of institutional services which might be offered (digital archiving for example), unless the original context of the idea is considered, which was (and is) the scholarly communications crisis.
Within the context of the scholarly communications crisis, the original purpose of repositories was to provide a way in which universities could enable access to research which they could no longer afford to buy because of the rising costs of journals. The repository was therefore a tool to aid the re-engineering of the business process of the library (as we came to think), or even the business processes of the entire university (as we now
tend to think).
My own view is that if the repository isn't serving that function, it isn't
a very useful concept.