Current state of irredeemably mendacious "AI" slop Try to use TDD and multi-agent workflows to test and...Claude will write shit code and then **change the test** so the code passes. cc @samuelstroschei @caesar0x @doctorow Running out of ideas. Anyone? chattr +i on tests?

so this project with a junior dev doing AI assisted pair programming in an educational manner isn't really going as planned...but forced to put serious effort into finding out in which ways these new tools could be useful...so far looks like a total confidence trick writ large