Foundations & actual-engine laboratory
Experiments execute the game rather than a substitute model.
Foundation modules provide typed canon, dependency/semantic checks, bounded variation, scaling checks and a release-evidence gate. The laboratory runs combat, encounter variations, professional cases, synthetic economies, coherence checks and earned journeys through actual engine commands with deterministic replay.
Verification: All 11 selected quick gates pass with recorded case counts. Empty, stale or indeterminate evidence cannot establish release approval.
Remaining: Calibrated human behavior, complete Floor 1 journeys, longitudinal cohorts and future-floor balance remain unproven.