A document is the "meat" of the Corpus, or its "building block". What a document node may contain is context-specific. For example, in the context of Markdown, a document would be the actual article. Whereas in the context of a programming language, a document could be a class, a module, or a function.

Documents can be nested beneath each other. This is very useful for representing ancestry chains, like in the case of programming languages that support namespacing like Ruby (::), C++ (::), or even virtual namespaces as can be achieved in JavaScript comments (e.g. the @namespace tag).

A document can have entities which are leaf nodes in the tree. Refer to their documentation for more information.