Each health claim on X is extracted, mapped to a GRADE-style evidence card, and scored across six dimensions — evidence alignment, scope, uncertainty calibration, counterevidence, commercial integrity, and safety. Every score is reproducible from the primary literature and the original post.