The wall-function models (log_law, werner, cabot) now declare their unit-test coverage (make unit-runtime, make unit-boundaries), so review tooling routes a wall-model change to its tests. The Case Reference now documents the first-cell y+ guard that was already in place: a run stops after ten consecutive steps with the wall-face mean y+ outside the selected law's range. The master template's wall-function comments now state correctly that roughness_height is rejected for werner and cabot, and that cabot uses the resolved pressure gradient.