Launcher Attrition Dominates Command Architecture: Agent-Based Analysis of IRGC Missile C2 Degradation, with Phase II Residual-Force C2 Observability Extension

1. Artifact claim. Companion code and data for the v4 launcher-attrition paper and the Phase II residual-force C2 observability extension (milcom-2026-iran-attrition 0.6.1). 2. Problem / scientific context. Implements the agent-based model, Phase II residual/provocation extension, claim gates, and locked data_phase2/ table that grounds Table I D_win means. 3. Contents. simulation/ (c2_core.py, phase2_runner.py, leg3_null_sim.py), verify_*.py, claim-manifest.json, integrity-manifest.json, data/, data_phase2/ (including mean_H*_D_win columns), and figures. This deposit is the software zip only. No paper PDF. 4. Reproduce. Python 3.8+, standard library only. python verify_milcom_claims.py; python verify_phase2_claims.py; full grid simulation/phase2_runner.py. 5. Relation to papers. Software concept 10.5281/zenodo.19210120. Phase II paper concept 10.5281/zenodo.21865793, this PDF pin 10.5281/zenodo.21926014. Parent paper 10.5281/zenodo.19210451 / 10.5281/zenodo.19558494. Leg 3 prereg stamp remains 10.5281/zenodo.21443990. 6. Limitations / non-claims. Thin Monte Carlo gates are not the 200-seed paper grid. Empirical 0.64 / 0.500 are hand-coded and are not in this zip as a machine-readable coding table. Not a targeting tool. 7. This version (0.6.1). Adds D_win mean columns to the 200-seed discrimination CSV and binds Table I. 8. Citation. Prefer software concept DOI 10.5281/zenodo.19210120. Pin this zip with 10.5281/zenodo.21925910. Cite the PDF separately.