What got my attention is not that someone built a Claude Code plugin for Kindle highlights. It’s that the plugin’s whole pitch is basically: Amazon hides some of your own notes behind a dumb export limit, so this thing walks around the edges and reconstructs them anyway.
That’s the sort of project I’d half expect to be a brittle weekend hack, not something presented as a “personal collection” with a tidy install path and a little marketplace wrapper. But the README is unusually concrete. It says the hidden text comes back through a mix of the Kindle notebook, the Mac Kindle app’s synced annotation database, and OCR of rendered pages. That part feels plausible. The “we recovered every one” claim is the part I’d want to kick the tires on. Maybe it really does work on the four books they tested. Maybe it gets uglier on a book with weird layout, foreign scripts, or a different kind of truncation. The page is confident in a way that makes me slightly suspicious, which is probably healthy.
I also think the most interesting thing here is not the recovery trick itself. It’s the product shape. This isn’t “here’s a script.” It’s “here’s a Claude Code skill that you can invoke, and Claude will auto-run when the request matches.” That tells you where this ecosystem is going: little task-specific agents bundled as plugins, each one wired to external tools and local state. In this case the external toolchain is comically specific — Chrome with Apple Events enabled, the current Mac Kindle app, Xcode Command Line Tools, Python, a localhost receiver. That specificity is a feature and a warning. Feature, because it makes the thing real. Warning, because this is the kind of pipeline that will break the moment one assumption changes.
The copyright note is the part I’d actually pause on. The author is careful to say it’s your own Amazon account, your own notes, personal use only. That’s the right disclaimer, but it doesn’t magically make the ethics simple. If a system intentionally throttles export, then bypassing that throttle may still be within a moral gray zone even when you’re touching your own annotations. I’m not making a legal call here. I just think the tension is the entire story: users reasonably want durable access to their notes, platforms want control, and tooling like this picks a side.
What I’d try first is the boring version: see whether the plugin reliably exports a couple of messy books with lots of highlights and truncated notebook entries. If it works there, then this is genuinely useful. If it only works on the author’s demo cases, then it’s a neat proof of concept dressed up as a marketplace item. Either way, the README is refreshingly specific, which I trust a lot more than the usual “AI-powered productivity” fog.
