A rejected upload costs you a day. The file goes in, the review comes back, and you are reading an error message about a dimension you cannot see in your PDF viewer. This validator exists to catch the mechanical problems before Amazon does.
It inspects the file itself: page dimensions and whether they are consistent, page count, embedded fonts, image resolution, colour space and whether the document opens cleanly at all. These are the failures that are objectively checkable from the file.
It is deliberate about its limits. A validator can tell you a page is the wrong size. It cannot tell you the cover text sits too close to the trim, that your chapter headings are inconsistent, or that page 143 is upside down. Those need a human and a printed proof.
No, and it cannot. This checks the mechanical properties of the file. Amazon also reviews content, rights and metadata, and applies its own review that no third-party tool can predict or guarantee.
Properties that can be read from the file itself: page dimensions and their consistency, page count, whether fonts are embedded, image resolution, colour space, and whether the document parses.
Anything that requires judgement or context. Margin safety against your specific design intent, typographic consistency, image placement, whether the content reads correctly, and anything about your metadata or rights.