Comment by colonCapitalDee

That's not surprising, LLMs are bad at pulling hyperspecific facts out of memory. LLMs aren't mapping applications, they're reasoners. Just a poor problem fit