The Fact & Gap Checker reconciles grounded research findings with stored claims by categorizing each fact as confirming, contradicting, or novel relative to parametric recall. When a model's unprompted recall matches live research, claims are validated with supporting citation sources, whereas facts missing from training memory are minted as new gap claims without overriding human rulings.