Is this using some sort of machine learning model that runs locally or just strict rule checking ?

I'm used to LanguageTool, and at a glance it seems like Harper covers way fewer rules than LanguageTool does. Not sure if this is actually noticeable in practice, but I run my own LanguageTool server and am not too picky about the performance, so I'm not in a rush to move until someone figures out a good way to compare them. LanguageTool's rules are all open source at least, so it's only a matter of time before Harper gets anything it might be missing.

I like this but I hate that's maintained by automattic. One day Matt gets drunk in a xitter fight and discontinues the product just to prove his point. He's as unstable as elon

Thanks for sharing. I can finally get off LT.

I was thinking 'why not languagetools', but you had me at rust

Great going. Excellent.

Please can you explain how to install it in LibreOffice. I'm not sure that I can as I saw https://github.com/Automattic/harper/issues/1427

Another big problem of languagetool is that they "forgot" to add any kind of authentication in the self hosted version, so even if you have gigabytes of unused memory it can't be published online