Type what shipped, separated by commas. The note assembles itself underneath. Nothing you type leaves the page.
Fill anything in above.
What makes a release note worth reading?
Three rules, and the box above follows all three so you do not have to remember them.
Breaking changes go first. This is the only one that is not a preference. Somebody opening a release note is asking what is about to stop working, and putting that at the bottom answers it after they have closed the tab.
Empty sections are dropped. Four headings with nothing under three of them reads like a template somebody abandoned. Leave a box blank and its heading does not appear.
One line per item, in plain words. A release note is not a commit log. The useful line says what a person can now do that they could not do yesterday, without the internal name of the module.
Why these five categories?
Scanning, not filing. They are the same five our own product uses, so a draft made here pastes straight in.
A fixed set lets somebody scanning a year of entries see the shape of the year: how much was new work, how much was repair, how often something broke on purpose. Free text destroys that the first time a heading reads Bugfix instead of Bug fix, and it will.
Is this a release notes generator too?
Yes. The two words describe the same box. Whether you publish the result as one line on a running page or as a document for one release is a decision about where it goes, not about how it is written.
What this is not is a pipeline tool. It does not read a commit range, run in a build or open a pull request. If you want a file to start from instead, the release notes template is plain Markdown.