Documentation in wikis.
Perfect example:
FluidNC which has it's documentation hosted on
wiki.js. The documentation is quite thorough and there is a lot of it, but you need an internet connection to access it. Should wiki.js decide they've had enough, the entire thing is lost.
If only they would give a means to download as html, pdf, something. It would be tempting to try scraping the site, but nowadays you'd be banned as an AI scraping bot or something, and wikis are generally a bit tricky to scrape.