The general code quality of hacking tools has always been shit. Thanks to LLMs now it's somewhat worse.

@buherator Yeah. And I'm reluctantly realizing I'm doing this too. Not that I usually made tools that were released for others as well, but while I made my tools reusable before I now just throw together whatever I need for the moment with (local) LLM help hard coded for the specific use case.

Helps with #XKCD1319 symptoms though.

@troed I guess it's fine for ad-hoc tools, but I'm looking at code with many stars on GH rn. It works. Sometimes it freezes. Everything is redundant, identifiers don't describe actual purpose but something else, functions are prohibitively large. No one will be willing or able to maintain this shit for more than 6 months.

... the README says "the most useful feature is X". I try X, absolute garbage. I look at the code of X: the only relevant piece of data is simply not included anywhere in the computation. ffs

(no I won't open an Issue/PR, let them figure out themselves...)