Initial release accompanying the position paper "Golden Dome Latency Governance: Autonomous Operations Model Extended to Boost-Phase Intercept Timelines" (Bilar 2026, DOI: 10.5281/zenodo.19368682 What's included simulate.py — unified falsification suite (v5.0), five scenarios A-E requirements.txt Result figures A through E (results/) Scenarios A: Ghost-in-the-Matrix — staleness check bypass confirmed at Δt ≤ −500 ms B: Quantization-Aware Spoofing — INT8/FP32 gap; 0.06% baseline adversarial success rate at unit-scale features (understates operational risk without feature normalization) C: Lethal Compliance — Check 3 passes physics-valid cold-start injection; detection falls to Check 4 above ~80 m discontinuity D: Per-check analytical sensitivity, all 7 checks E: Monte Carlo, 50,000 trials × 20 levels; combined detection 99.0% at nation-state (s=0.74), 89.2% at maximum sophistication (s=1.0); PINN-zeroed drops to 61.7% Known limitations Per-check parametric models are assumptions, not measured hardware failure rates Feature normalization gap in Scenario B understates operational quantization risk 80 m coherence threshold is an arbitrary default; Phase GD-0 must characterize it Numba optional; scipy fallback active if not installed