# sim_mornings.py (SIMULATION)
$ exit=0
SIMULATION (not measured on GitHub)
n=200 seed=20261109 p_hotfix=0.18 p_overlap=0.45
strategy mix A/B/C = 0.5/0.2/0.3
counts: {'no_conflict': 189, 'B_theirs': 3, 'C_both': 3, 'A_ours': 5}
mornings with an overlapping hotfix conflict: 11
source-level regressions that would land WITHOUT the hunk guard: 8
source-level regressions that would land WITH the hunk guard: 0
of those unguarded regressions: A=5 C=3 (B=3 stays safe, drops agent work)
