This Phase-2 preprint tests Reality-Bounded Improvement and the Grounding-Efficiency Engine beyond the original constructed witness. It reports a controlled hidden-environment success lane, a public digits negative/mixed shadow lane, and a named OpenCV Haar public-model smoke lane that is null/weak. The central claim is asymmetric: GEE wins inside the frozen controlled simulator under the Phase-2 threshold, while public shadow failures are retained as no-credit receipts. No real robot deployment, third-party replication, task-relevant VLA/world-model public validation, hard 0.05 margin, or unconditional GEE optimality is claimed. Public navigation and challenge routes:Main site: https://mianzhang.org/Public paper index: https://mianzhang.org/papers/Concept index: https://mianzhang.org/concepts/GitHub source and public issue routes: https://github.com/mmjbds/mianzhang.org and https://github.com/mmjbds/mianzhang.org/issues/new/chooseHugging Face technical mirror: https://mmjbds-mianzhang-org.static.hf.space/HF Space repository: https://huggingface.co/spaces/MMJBDS/mianzhang-org Use this Zenodo DOI landing page as citation authority. The linked public routes provide navigation, issue routing, evidence boundaries, and challenge reports.