ADF represents a document as a tree of typed nodes: a doc root containing paragraphs, headings, lists, and text nodes carrying marks such as strong or em.

Sending Markdown to a Jira Cloud v3 endpoint does not fail loudly. It renders as literal asterisks and hashes, which is worse than an error because it is only discovered by a human reading a badly formatted ticket.

A minimal converter that handles headings, bullet lists, bold, and paragraphs covers almost everything a feedback report contains. Anything unrecognized should degrade to a plain paragraph rather than being dropped.

Why it matters

It is the single most common reason a Jira integration technically works while producing tickets nobody wants to read.