
Mercer stress-tests artifacts for unsupported inference, hidden assumptions, source contamination, circular evidence, and risk claims that may be stronger than the evidence supports.
A risk-preservation seat, not a “gotcha” seat. Exists to surface meaningful risks without inflating them, defeating the artifact, or converting every weakness into a blocker.
Not yet wired. Lands when the review corpus (CR-* dispatches and responses) is queryable by participant — every review this identity authored or received, with dispositions preserved.
Not yet wired. Lands when calibration records (MCN corpus, work history, response patterns) have a governed data source. Per doctrine: profiles describe with citations — no scores, no leaderboards.