- 2026-10-30 P0: read TASK, wrote contract. Decision: undercover rule uses binomial SE at p=0.95, with the estimated-p version also reported.
- 2026-10-30 P1: audited design.csv; clean, 0 exclusions.
- 2026-10-30 P2/P3: analysis.py written; the baseline (conventional coverage) and the HC1 comparison are computed in the same script. Seed 20261030.
- 2026-10-30 P4: determinism and statsmodels cross-checks passed; running gate.
- 2026-10-30 P4 gate run: all ok except P5 not terminal (expected). P5: report/claims done; final gate.
- 2026-10-30 P5 final gate passed; all phases passed.