I believe there is no complete database that would provide you with a reference to an authoritative review on every chosen topic.
There are obviously journals mentioned by user just-learning that often provide high-quality reviews from people at the frontier of their area. These journals in which the most authoritative reviews independent of the field are usually published are: Reviews of modern physics, Physical review letters and the Physics reports-review section of Physics letters. As for a focus more on experimental progress, excellent reviews are also found in Science and Nature.
If an article in these journal has a very general title or even the words "report" or "review" included, it is most likely the thing you are looking for.
As to the database - even if it existed, I would not use it. When I want to probe a topic I don't know, I type the keywords into an article search tool such as Web of Science or Google scholar and sort the results by citations (Web of Science tends to be more complete, but you need a university login). Amongst the first few of the results, the ones with the general titles will usually be the reviews "everyone knows", implicitly uses their assumptions, notation and copies the typos in their computations.
For example, type "Supersymmetry" into the search - Hans Nilles review is the thing you want. Or you want to know about the so-called "f(R) theories of gravity" - in both search tools you get immediately what you need.
For broader fields and for older reviews this usually works, but for reviews of recent developments in a more narrow field, you will have worse luck. Both the searches allow you to find articles only from a certain date but you will usually just find articles reviewing the most recent problems or lines of research - not an overview of the overall paradigm shift.
Furthermore, if even written, these "minor paradigm shift reviews" are very often included as extended introductions of articles on specific topics. It is hard to imagine how a database of such more or less minor reviews could even be organized. In this finer case, there are usually just a few tens of scientists who actually can identify a good recent review and they, as far as I know, don't create databases in any unified manner. Instead, they mention that this and this article had a nice review in their own articles. I am afraid this is where the detective work kicks in - you have to read at least the introductions of the most influential articles and trace the smaller reviews through references.