It would be interesting to see if LLM's can scan and surface such issues. Might have been probable with parsing tools pre-LLM also? Not sure.