Limitations

One way. The library converts ADF to Markdown. It does not convert Markdown, HTML or the older wiki markup of Jira to ADF: build the document with the object model instead.
No API client. The library does not call the Atlassian APIs. Download the documents, and the media they reference, with your own HTTP client; see Media and mentions.
Lossy conversion. Colors, underline, alignment, annotations and the other marks without a Markdown equivalent are dropped, and pipe tables lose cell spans; see Other nodes and marks.
No schema validation. Parsing accepts any tree under a doc root, and building a document does not check that a node is allowed where you put it.
Known nodes keep only their modeled attributes. Writing a parsed document back drops the attributes the classes do not have; unknown nodes and marks are kept whole.