The idealistic answer is that the student must present a novel discovery in their field. This means it has to be something new - you can't just redo an experiment that was already done. However it doesn't have to be significant or impactful at the PhD level. Even higher-tier programs can be pretty forgiving with "insignificant" discoveries for a dissertation. You will also have to demonstrate sufficient technical knowledge related to your discovery and the ability to communicate it clearly. These are all subjective and decided by your dissertation committee.
Also, historically, a PhD means that you are qualified to teach at the highest levels, i.e. that you may now train other PhDs yourself. These days, the job market is tough, and it is rare to get a faculty position with "just" a PhD. But in principle, the ability to "procreate" is the essence of the PhD. In the 70s and before, it was not uncommon to start a faculty position immediately after completing a PhD.
The "real" answer is that typically it is desirable that:
- You publish many papers
- You publish first author papers
- You publish high impact papers
- You are able to secure grants
- You are highly employable as a postdoc in the field and ideally already have tentative job offers
I said earlier that the discovery doesn't have to be impactful. That's true, but your advisor will not want you to attempt defense if you don't have good papers out. For papers, impact matters. And while technically papers are not required to defend a PhD, your advisor's blessing is, so you see how that works.
Moreover, remember that early in the PhD there will be a proposal mechanism such as the qualifying exams. This is where you present a detailed proposal of the research you are planning to do for your PhD. Your committee might not accept your proposal if it seems unlikely to yield interesting results. So if you have already passed your quals, done the research, and the results turned out uninteresting - oh well, you "can" still defend (but assuming your advisor or committee doesn't care about lack of strong papers). But they also try to prevent that very situation at the qual stage.
There may also be some trivial institutional requirements such as must have taken X courses, must have TA'd Y times, etc. These will depend on the institution, they're not universal.
So I am interested in knowing what those are in general for post-graduate and how they differ from Master's or docent preparation.
A Master's dissertation does not require the work to be novel. However, like all degrees, a Master's also requires you to satisfy whatever requirements your advisor has. Master's advisors are generally the same professors that take on PhD students, and it's not unheard of for them to have a strong bias towards novel work. So they may insist that you work on novel things or else refuse to let you defend, whether the degree technically requires the work to be novel or not. Also, not all Master's degrees require a dissertation - some just let you complete coursework. Generally, the degree indicates advanced knowledge of a field, which may be desirable for private employers seeking highly trained specialists. A Master's degree does not normally qualify you to train other Master's students, nor indicate the ability to perform independent research, in contrast to a PhD.
I'm not really familiar with docents since they are a European thing. It also varies a lot with country and isn't much of a universality. Generally, this is an expression of a fresh PhD still not being enough to become a professor straight out of school. Since the PhD is now not enough, the grads are expected to publish some more meaty papers, so that a committee can finally certify them as suitable for professorship. Professorship isn't automatic in any country - even in places like USA with no "docent" rank, you still have to first please a department sufficiently as to get hired as a junior professor, then impress them further at tenure review to become a full professor. The things they look at are the same: Do you have many important papers, and are you getting many good grants. Just some countries give you an official title, others handle it more informally.
Technically, a docent must also demonstrate research that is independent, as opposed to how a PhD student is supervised by someone. However that's sort of a given. There's not many tenure-track post doc positions, and you usually do not get to this stage without already being some sort of faculty.
In sum, I guess you can say that procedural requirements dominate the difficulty of obtaining a PhD, MS or docentship. Therefore, many important skill requirements are implicit. Those that are required explicitly paint a very incomplete picture.