How to evaluate a thesis based on papers with multiple authors

Let me start with a story. At Harvard, before a PhD student is allowed to schedule their oral dissertation defense, which (as is the case in many places) is usually a formality, the advisor is required to explain the significance of the dissertation work at a faculty meeting, and the faculty have to agree that the work is significant enough for a dissertation for the defense to go forward. The possibly apocryphal story goes that, after a contentious faculty meeting at which a dissertation was grudgingly approved, the advisor, a well-known mathematician who has had many students, remarked, "Well, this isn't the worst dissertation I've written."

The point is that, while we claim that a dissertation is the work of the student, in most cases, this claim is really partially true polite fiction. Certainly the student is expected to understand all of it, and to have contributed significantly to it, but, in any sort of collaborative theoretical work, it doesn't even make sense to precisely separate out which ideas came from whom. I've heard mathematicians claim that they correctly decided which idea to pursue to solve a problem based on the quality of the silences during a conversation with their collaborators, and I believe this claim. When ideas are developed with such subtle yet vital contributions, it makes sense to just give each co-author 100% credit for the ideas.

Unless you have a policy that theses should not be based on collaborative work - and, frankly, this policy is unenforceable due to the common but not universal custom for advisors to decline co-authorship with students on papers where they would otherwise be a co-author - you really have no choice but to take at face value the claim that the thesis represents the student's work. At the least, you can expect the student actually wrote the text of the thesis and understands its contents, and if there is an oral defense you can test whether this is true. Beyond this, you will have to trust the student and the advisor that some acceptable portion of the ideas in the work actually came from the student. Someone contemplating hiring the student will get more information in the recommendation letters from the advisors and the other collaborators.

It's generally understood that an advisor usually ends up spending more time advising a student on solving their dissertation problem than it would have taken the advisor to solve the problem themselves. You might want to read between the lines of the following statement about the culture of PhD advising from the American Mathematical Society: http://www.ams.org/about-us/governance/committees/Statement_DirectingPhDTheses.pdf

You don't say which country you are in, but if this were the UK, I'd evaluate the work in thesis on its merits, and then interrogate the student on which parts of the work they consider to be their contributions in the viva. If you are convienced in the viva that the student had made a significant personal contribution, I would work with them, through the corrections process to add a statement/statements to the thesis to make clear what the students contribution to the work was.

French perspective here. My understanding is that the contents of the PhD manuscript handed by the candidate should mostly consist of their own work, or things of which they are a main contributor: they should definitely understand it, probably have written it (or wrote a version of it), etc. If the candidate is an author on a paper to which they only contributed in a minor way, they can mention the paper in the PhD but probably not include the material in the manuscript. And by extension: if the candidate hasn't been a major contributor to any paper to which they are listed an author, then the manuscript should be pretty lightweight indeed. :)

Concretely, if you have doubts I'd advise getting in touch with the thesis advisor to clarify expectations, i.e., ask them to briefly describe the personal contribution of the student to the works they have co-authored, clarify if the contents of the thesis is indeed the candidate's specific contribution, etc. If phrased politely, I think this is a pretty sensible request.