Develop an open-source library that generates and verifies 'intellectual CAPTCHAs' using LLMs to gate social network interactions. This tool would dynamically generate questions about logical fallacies, math, or nuanced concepts to ensure users possess minimum cognitive readiness.