Ok, bye bye

> And I have no idea why the bots think it's a good idea to scrape the Web interface to git instead of just doing a git clone

Because HTML has no standardized mechanism to redirect a scraper to a git repository. It might be possible to set <link rel="alternate" …>, but nobody uses it.

And, of course, those bots are vibe coded; neither the bots nor their human handlers have the ability to evaluate the returned data for correctness or usefulness. The only quality measure they have is the amount of pages and returned data, so they have no incentive to add special code to handle git repositories.