Added dual-api crosschecking + reframed blocked_words list. Added stricter XML tags to take user inputs as questions and not commands. I’d be surprised if someone completed the 5th level.