{
 "P0": {
  "status": "passed",
  "checks": [
   "TASK.md read fully",
   "contract.md written"
  ],
  "notes": ""
 },
 "P1": {
  "status": "passed",
  "checks": [
   "sha256 computed and match bundle.json",
   "row/col counts, NA, duplicates checked",
   "G and N consistent with cluster_sizes for all 9 cells",
   "exclusions.csv written (0 removed)"
  ],
  "notes": "no availability violations"
 },
 "P2": {
  "status": "passed",
  "checks": [
   "naive iid baseline (s^2/N, z=1.96) computed in same draws; stored as baseline_naive_iid_coverage",
   "baseline ~0.95 at ICC=0, 0.30-0.48 at ICC>0"
  ],
  "notes": "baseline ignores clustering"
 },
 "P3": {
  "status": "passed",
  "checks": [
   "analysis.py implemented, reads only data/design.csv",
   "per-cell seeds from design, recorded random_seed=C1 seed",
   "runtime ~2 s"
  ],
  "notes": "WCB quantile = 380th order statistic of 399 |t*|"
 },
 "P4": {
  "status": "passed",
  "checks": [
   "clean re-run twice: results.json and table1.csv byte-identical",
   "17 consistency checks all PASS (see checks.md)",
   "MC rel bias vs analytic E[V_CR0]: max |z|=2.00",
   "figure regenerated after tick fix; results unchanged (cmp)"
  ],
  "notes": ""
 },
 "P5": {
  "status": "passed",
  "checks": [
   "report.md has 6 required sections",
   "every claim's results_keys exist in results.json",
   "report numbers cross-checked against results.json/table1.csv (MC SE ranges, z-distances)"
  ],
  "notes": ""
 }
}