Excuse me for coming back to the thread so late! And thanks for considering checking it out. :)

Ok, I should probably rather have said "vibecoded under strict supervision". What I meant was, I didn't really try to debug the processes that are working well enough, nor tried to make the interface make more sense. Other than that, the code is not doing anything that is not listed in the readme, and everything stays local unless you opt into the annotation collection (off by default) to support a future project. A backup is always a good idea, although the plugin is designed not to do anything that threatens how Zotero works (see the now added Safety heading in the readme).

Regarding your questions:

1. It used to import everything at once, but thanks a lot for the suggestion, the new release (0.7.2) now has a review step. You can select/deselect individual highlights, whole titles, and also operation types like removals or cleanups, and only then press Apply.

2. Yes, it does. Imports are incremental, so new highlights and notes get picked up on later runs and already integrated ones are skipped. Annotations you created manually in Zotero are left alone, and text you append to an imported comment is kept when the Kindle note changes. It also merges exact duplicates, which I kept running into with scanned PDFs. I assume there are going to be lots of bugs here, I would be thankful for any kind of feedback.

I hope you still decide to check it out. I am especially happy with now being able to set the highlight color from the Kindle itself. I'll try to keep an eye on this thread, and of course you can also always open an issue on GitHub.