feat(card): seven acceptance counterexamples end to end (BUG-1812)

- career layer gets the D1 longitudes, so the paksha has one source (Obama base vs career mismatch fixed) - card carries movable_lagna_moon.reason, Narayana ad_status / ad_reason and the mahadasha cycle_index; running line says "sub-period undecided" - engine keeps the MD when the AD is undecided; timingFollowing skips zero_duration rows explicitly - timingKeys allowlist: adstatus / adreason / cycleindex / zeroduration - fixtures + capture script, 16 engine tests, 7 card tests, PROGRESS, BUG-1812 Co-Authored-By: Grok (agent loop)
docs: task brief for #169 counterexamples (2026-11-09)
2026-11-09 08:38:56 +08:00 · 2026-11-09 07:18:12 +08:00
25 changed files with 7481 additions and 63232 deletions
@@ -24398,21 +24398,30 @@
 - 关联记录：无
 - 修复版本：无

-## BUG-1809 | 第七轮整合上 staging 后，生时校正三组测试集仍封存在 round5 标签：重封 round7_integration_2026_11_09
+## BUG-1812 | 验收清单七个反例：事业层与基础层「月相」不一致，数据卡丢掉月相原因 / 小运未决状态 / 第几轮大运
+
+- 状态：fixed-pending-verify（本地分支 `codex/counterexamples-20261109`，未推送）
+- 首次发现 / 最近更新：2026-11-09（占星师 2026-11-08《近期六轮问题闭环与网站验收清单》§7）/ 2026-11-09
+- 影响面：`scripts/consultation_native_layers.py`（`build_career_analysis`、`build_dasha_systems`）、`scripts/narayana_legacy_label.py`（`_period`、`rath_narayana_view`）、`frontend/src/lib/consultation-evidence-card.ts`（`functionalProfileFields`、`NARAYANA_MD_CARD_KEYS`、`DASHA_ROW_KEYS`、`baseTiming`、`narayanaRunningLine`、`timingFollowing`）、`frontend/src/mastra/consultation-workflow.ts`（`timingKeys` 白名单）。
+- 现象（三处，同一批验收反例）：
+  1. **同一张盘两层月相不一致。** 一张动星座上升、月相可从 D1 经度算出的公开盘，基础层 `functional_benefic_malefic.movable_lagna_moon` 是 `derived_from_table_30_note`（waning / neutral），事业层 `consultation_native_layers.career_analysis…movable_lagna_moon` 却是 `blocked` / `moon_paksha_missing`。
+  2. **卡上说不出「为什么没月相」。** `functionalProfileFields` 只抄 `status` / `moon_paksha` / `role`，丢掉 `reason`，「未给月相」与「表里没这一格」在卡上同形。
+  3. **小运未决被写成「没算」。** 小运定不出时 `current_dasha` 整段没有 `ad`，卡上分不开「子运未决」与「子运没算」；`narayanaRunningLine` 只写主运。**第几轮大运**（`cycle_index`）和 **0 年段**（`zero_duration`）同样没有上卡，卡上无法区分第一轮与第二轮、也无法说明 0 年段不是当前段。
+- 触发条件：(1) 动星座（Aries / Libra / Capricorn）上升且事业层从 `_affliction_planets` 的行拿行星；(2) 任何缺月相输入；(3) 该大运的 AD 起点比较未决（例如土星与计都同在运星座，`saturn_ketu_priority_unresolved`）、年龄进入第二轮、或盘里存在 0 年段。
+- 根因：
+  1. `build_career_analysis` 交给 `analyze_career` 的行来自 `_affliction_planets`，只有 `house` / `sign`，没有绝对经度；`analyze_career` 内的 `moon_paksha_from_planets` 因此判不出盈亏，而基础层用的是真实的 D1 经度 —— 同一个判定有两个数据源。
+  2. 卡上的 `movable_lagna_moon` 字段清单是手写白名单，加了 `status` 没加 `reason`。
+  3. `rath_narayana_view` 只在 AD 选定后写 `current_dasha`，AD 未决时整段省略；`_period` 的字段清单里没有 `cycle_index` / `zero_duration`；卡的 `DASHA_ROW_KEYS` / `NARAYANA_MD_CARD_KEYS` 也没有对应键，且 `timingKeys` 白名单同样没有，新字段即使引擎给了也会在投给模型的 `timing` 投影里被裁掉。
+- 修复：
+  1. 事业层与基础层同源：`build_career_analysis` 把 D1 的 `lon` 并入交给 `analyze_career` 的行（缺经度时仍返回 `moon_paksha_missing`，不猜）。
+  2. `functionalProfileFields` 在 `movable_lagna_moon` 上照抄 `reason`；缺月相时卡上明确 `reason = moon_paksha_missing`，只有月亮不进功能吉凶三档。
+  3. `rath_narayana_view` 在 AD 未决时保留大运并写 `ad_status = blocked` / `ad_reason`；`_period` 带 `cycle_index`（有值时）与 `zero_duration`（为真时）；`build_dasha_systems` 把这四项原样带到 `dasha_systems` 行；卡上 `NARAYANA_MD_CARD_KEYS` 加 `cycle_index`、`DASHA_ROW_KEYS` 与 `timingKeys` 加 `ad_status` / `ad_reason` / `cycle_index` / `zero_duration`、`narayanaRunningLine` 在 AD 未决时写「子运未决（原因）」、`timingFollowing` 显式跳过 `zero_duration: true` 的行。
+- 验证：
+  - 引擎：`tests/test_counterexamples_engine_20261109.py`（16 条）。七个反例各一条，外加「引擎当前输出 == fixture」与「fixture 覆盖七个反例」「fixture 不含个人资料」。断言了：缺月相只阻塞月亮且两层一致；未决步骤被点名且 `used_for_scoring` 为假、没有补出来的当前大运；AD 未决时 MD 星座 / 轮次 / 日期保留、`ad_reason` 被点名、无 `ad`；自参照格局 `independent_condition_count = 0`；出生小运同时留父段起点与可见起点且 9 条边界不移动；第二轮 `cycle_index = 2`、年数 11；0 年段留在数据里带 `zero_duration`、当前段是 Aquarius。
+  - 数据卡：`frontend/tests/consult-counterexamples-card-20261109.test.ts`（7 条），读同一份 fixture 经真实链路 `parse → toAgentConsultationContext → toModelOutput → buildEvidenceCard`。修前同一批测试在基线 `b453505d` 上 3 条失败（(a)、(c)、(f)），修后 7/7 通过；连同 `consult-self-reference-pipeline-20261108`、`consult-evidence-card-20260927`、`consult-evidence-card-v2-20260927` 共 36/36 通过、0 cancelled。
+  - `tsc --noEmit` 0 错；`npm run lint` 0 error。
+- 防复发：上述两套测试；fixture 由 `scripts/research/capture_counterexample_fixtures_20261109.py` 生成，引擎行为一变 `test_fixture_is_fresh_for_every_counterexample` 立即失败。卡上字段清单（`NARAYANA_MD_CARD_KEYS` / `DASHA_ROW_KEYS` / `timingKeys`）与引擎字段一一对应，改一处必须同时改另一处，否则对应用例失败。
+- 未处理：数据卡上的大运（MD）仍拿不到 `parent_period_start` / `visible_period_start` —— 唯一生产者是冻结文件 `scripts/jyotish_api_server.py::_consultation_dasha_sub_periods` / `_consultation_birth_moment_mahadashas`（只返回 `lord` / `start` / `end`），本单按任务书不碰冻结文件；出生余额的 MD 裁剪改为在引擎层断言（`sub_periods.current.mahadasha.start` 早于出生）。要上卡需另开一单。
+- 关联记录：BUG-1732（Narayana 逐步未决）、BUG-1735（运星座与日期上卡）、BUG-1741（五步 profile）、BUG-1783（`profile_config`）、BUG-1812 所在验收清单 §3 / §5 / §7。
+- 修复版本：本地分支 `codex/counterexamples-20261109`

- 状态：fixed-pending-verify（本地分支 `codex/rect-reseal-r7-1109`，基于 staging b453505d，未推送；产品负责人 2026-11-09 批准「agree new result」）
- 首次发现 / 最近更新：2026-11-09（第七轮整合 b453505d 合入 staging 后，封存记录仍是 `round5_pvr_narayana_2026_11_08`）/ 2026-11-09（产品负责人批准，重封）
- 影响面：只加研究记录与测试，产品代码不改。新文件 `docs/research/sealed_holdout_rerun_round7_integration_2026_11_09.{freeze.json,json}`、`docs/research/reported_offset_round7_integration_2026_11_09.{freeze.json,json}`、`docs/research/live_astrologer_v5_round7_integration_2026_11_09.json`（只存相对真值的分钟数，不含日期）、`docs/research/round7_integration_2026_11_09_comparability.json`、`docs/research/round7_integration_2026_11_09_replay_overlay.patch`；`tests/test_rectification_reseal_round7_bug1809.py`。
- 现象：b453505d 带来 R1（小运方向默认 p50）、R2 局部传播（默认强弱范围 `seed_only_v0`，BUG-1784）、可比领先（`common_evidence_coverage_v1`，BUG-1801）等改动，回放结果已和 round5 标签不同，但标签没有更新。
- 重跑（与 BUG-1720 / BUG-1759 同一协议）：回放树 = `git archive c1a69445^` + `float_split_2026_11_08_replay_overlay.patch` + `round5_pvr_narayana_2026_11_08_replay_overlay.patch` + b453505d 的 `scripts/`、`references/` 整体覆盖（退役文件保留回放树版本）+ 新补丁 `round7_integration_2026_11_09_replay_overlay.patch`。新补丁只有三处：两份回放脚本的 FREEZE / REPORT 指向新标签；退役打分器 Narayana 无证据行时按现行引擎「不计分」补 0 分（偏移回放触发 3351 / 1316340 行，封存回放不加载该文件）；回放副本里 `birth_at` 守卫改成 `RuntimeError`。回放实际加载的模块里，仍在 b453505d 的 57 个除两个守卫文件外与 b453505d 逐字节相同。按补丁头部步骤从零重建，`scripts/`、`references/` 与运行用的树完全相同。脚本 `--freeze` 新建记录后回放（`PYTHONHASHSEED=0`）。现行阶段用 `scripts/research/live_astrologer_v5_sensitivity.py`（多记 `leader_coverage`、`duplicate_event_ids`），直接跑 b453505d 工作树。`round5_pvr_narayana_2026_11_08`、`float_split_2026_11_08`、`astrologer_166d_2026_11_05` 及更早标签原样保留。
- 与整合代理 seed_only 列核对：封存 20/20、偏移 900/900、现行 231/231 逐条相同，覆盖统计两份逐字节相同，没有差异。整合代理报的「偏移 57 次 / 4 例」「现行 115 次」都是对整合前的 staging（未封存）比的，不是对 round5 标签比的；对 round5 标签见下。整合前的 staging 本来就和 round5 标签不完全相同：偏移头名平均误差 20.65 对 20.49，现行 9/231 次不同，来自两次重封之间合入 staging 的其他改动。
- 结果（旧 `round5_pvr_narayana_2026_11_08` → 新 `round7_integration_2026_11_09`）：
-  - 封存 20 例：top-1 0.35→0.30、top-3 0.45→0.45、平均分钟误差 6.80→6.75；2/20 例变：1 例失去头名（true_rank 1→11，分钟误差 9→8），1 例 true_rank 10→3；误确认 0→0。
-  - 报时偏移 900 次：top-1 0.0156→0.0100、top-3 0.113→0.106、头名平均误差 20.49→20.63 分、交付覆盖率 0.900→0.883；164/900 次变（7/20 例、44/45 格）：true_rank 变好 107 / 变差 39，头名误差变 57 次，头名命中变 5 次，交付覆盖变 15 次，交付宽度变 18 次。对整合前的 staging：57/900 次变（4 例），true_rank 变差 39 / 变好 0。
-  - 现行占星师阶段（holdout v5 77 × 3 = 231）：最佳 = 真值 8→8、真值在前三 19→18、平均最佳误差 10.38→10.29 分、最佳 = 报时 17→16；124/231 次变（49/77 例）：前十 105、最佳分钟 31、匹配档 13、报告置信度 13、引擎置信度 77、头部簇 38。对整合前的 staging：115/231 次变（45 例）。
-  - 可比性（`common_evidence_coverage_v1`）：封存 40 次评分（20 全量 + 20 稀疏）全部 comparable；偏移 900 次 comparable 863、暂列（`provisional_unequal_coverage`）35（全在 1 例）、无单一领先 2（覆盖相同但并列，无可比领先），不可拆 0；现行 231 次 comparable 226、暂列 5（3 例）；重复事件 0。偏移样本同一例重复，不按 900 次独立计。
- 判断：这些只是敏感度，不是准确率。事件打分接近随机水平（BUG-1644），三组结果早被看过，升降不能当变好或变差，也不能当产品指标。重封理由是记录与产品口径一致。
- 验证：新测试 `tests/test_rectification_reseal_round7_bug1809.py`：新标签封存 / 偏移 / 现行数字、可比性策略与领先计数、旧标签数字不变（带原值 / 新值 / 原因）。之前没有测试按标签钉封存结果，所以没有要改写的旧测试。`test_active_rectification_api.py::test_long_real_conversation_reaches_vedastro_after_local_range_is_narrow` 已知失败（见 BUG-1720），与本单无关。
- 防复发：上述测试；以后重封必须新建标签，不覆盖旧标签。
- 关联记录：BUG-1720、BUG-1759、BUG-1784、BUG-1801、BUG-1644
- 修复版本：本地分支 `codex/rect-reseal-r7-1109`
@@ -1,118 +0,0 @@
-{
-  "record_version": "reported-offset-native-candidate-date-v3",
-  "frozen_at_utc": "2026-11-08T23:15:54.895716+00:00",
-  "ayanamsa": "raman",
-  "node_mode": "mean",
-  "radii_minutes": [
-    15,
-    30,
-    60
-  ],
-  "offsets_minutes": [
-    -30,
-    -20,
-    -15,
-    -10,
-    -8,
-    -5,
-    -3,
-    0,
-    3,
-    5,
-    8,
-    10,
-    15,
-    20,
-    30
-  ],
-  "minute_step": 1,
-  "dataset": "references/real_case_calibration/minute_rectification_holdout_v3.json",
-  "dataset_sha256": "45aa76ab79dfc6c6404b6252083e1aa5c7e2e2b7de811fc11042a8ad2da3f0ea",
-  "dataset_benchmark_id": "minute_rectification_fact_ranker_v4_holdout_v3",
-  "implementation_sha256": "b32140f91dd8c65a4380ad31a29d77902b2ec2cd06495b9ce387eb7e60b68825",
-  "implementation_sha256_prefix": "b32140f91dd8c65a",
-  "historical_artifacts_manifest_sha256": "6102a26a840be207b5858b3a4c0509274468ae9071e86d308cbbac7371d6864e",
-  "production_scoring_files": [
-    "scripts/active_rectification_event_engine.py",
-    "scripts/active_rectification_events.py",
-    "scripts/ashtakavarga.py",
-    "scripts/dasha_analyzer.py",
-    "scripts/divisional_charts_extended.py",
-    "scripts/domain_calculation_service.py",
-    "scripts/functional_benefics.py",
-    "scripts/jaimini.py",
-    "scripts/minute_rectification_fact_ranker_v4.py",
-    "scripts/minute_rectification_feature_facts_v4.py",
-    "scripts/narayana_dasha.py",
-    "scripts/rectification/api_service.py",
-    "scripts/rectification/candidate_contrast.py",
-    "scripts/rectification/candidate_window.py",
-    "scripts/rectification/case_holdout.py",
-    "scripts/rectification/contracts.py",
-    "scripts/rectification/dasha_transition_proximity.py",
-    "scripts/rectification/decision_policy.py",
-    "scripts/rectification/event_probes.py",
-    "scripts/rectification/refinement_packet.py",
-    "scripts/rectification/scoring_service.py",
-    "scripts/shadbala.py",
-    "scripts/varga.py"
-  ],
-  "production_scoring_sha256": "9575adabfc825fe515d571ea2a8e8550b85ccbf089f9e4b185d330821d37ace8",
-  "research_files": [
-    "scripts/research/reported_offset_sweep.py",
-    "scripts/research/cluster_width_lib.py",
-    "scripts/research/probe_supply_after_six.py",
-    "scripts/research/sealed_holdout_rerun.py",
-    "scripts/minute_rectification_blind_eval.py",
-    "scripts/minute_rectification_fact_blind_eval_v4.py",
-    "scripts/minute_rectification_holdout_validator.py"
-  ],
-  "research_implementation_sha256": "b7130fd8a9dd0a734d1c10aa0a4900a19702ecf75044ad1740e596f20cada0b5",
-  "file_sha256": {
-    "scripts/active_rectification_event_engine.py": "3bda6fcf8855bec90dd62e64fae167ce3e6afe5956549bbd26587ea59eba3d94",
-    "scripts/active_rectification_events.py": "3d8a625d24aac6f7f05a3265a23b488153daa03b5b81c6b993160011cc07eb1c",
-    "scripts/ashtakavarga.py": "c524da3f54b2ed9c080b78b33854854a0ca3d5d79b7e536bb1c3bc2be3e02852",
-    "scripts/dasha_analyzer.py": "3949b96813217fe624899856afec9f0b787c4cb0fea3a9a0d961990dbd11943a",
-    "scripts/divisional_charts_extended.py": "b13ad68c45685568d6eb17f87017624a9288e31fad74662735cd6d6564f7755d",
-    "scripts/domain_calculation_service.py": "66923a2bc69daf65c965d2973c4b9234edbec5147909eaf3609b6b377ce3c7ca",
-    "scripts/functional_benefics.py": "d34dd7745ba9db0a61e3f7a2aa1e6acd1f819961442cb3412b5dbe22c02ef032",
-    "scripts/jaimini.py": "e474b3a1d5c50de87fafdbdcc3df9bd724de004aa2495463763f5b2688b8f5e2",
-    "scripts/minute_rectification_blind_eval.py": "07aa30d8bad3278669495d5fea8be937aaf6367c8af980ff4f2fe9aa84ccc6ff",
-    "scripts/minute_rectification_fact_blind_eval_v4.py": "bdf460e3467208e8b87345c5beb2128f3e22de7eeffd68f72b281b3ad9affc92",
-    "scripts/minute_rectification_fact_ranker_v4.py": "f1a0d8d2482cd754d2e7f253a9606b0373519435a2bf58cfc3753e2b9ae0fb76",
-    "scripts/minute_rectification_feature_facts_v4.py": "4661d725d4f04935169d9d8a86ce8d0ef4061915a3a985c4912b93a6e035689e",
-    "scripts/minute_rectification_holdout_validator.py": "b60b1c752e58672beaee8b36eb3b2348169726cda070ec969db0f13e7ed715af",
-    "scripts/narayana_dasha.py": "5a924f06f76426ab587163fd8c0b5e32a6a8c6917e8670029bd84a2eb6a6172e",
-    "scripts/rectification/api_service.py": "089730097d3872a4f7b5d7dd5574bb7e9ab4d2e246d4a380800bc824d7979b95",
-    "scripts/rectification/candidate_contrast.py": "eee7da71c0574a08876404b78ca111153fa95fb83ace82b9fdca141d1992e273",
-    "scripts/rectification/candidate_window.py": "2fbc5058b12049b15fca375cdc125b8283e19f9f3e4bedcc291e344501891bef",
-    "scripts/rectification/case_holdout.py": "fe0e699b617d9e216acdfeff87af4221c1d6f4c101bf33e1dbf14e2bcd75a8b5",
-    "scripts/rectification/contracts.py": "c6dc6bd230e38f9f2126c224aae9ddbd6415304c15340f8c7cd3f06f6ef3160f",
-    "scripts/rectification/dasha_transition_proximity.py": "66d29c0b182309d17264dd2645b5cb426e28b4066d6c7004fa66f3334286753f",
-    "scripts/rectification/decision_policy.py": "80e8de4680d19a052213bb821d2a9c94143a2962235aeb27f6ad81374d2e60d5",
-    "scripts/rectification/event_probes.py": "283ac525efa5ba5b12e804d9b87e6725741ede3365659002224fffe61722bc3f",
-    "scripts/rectification/refinement_packet.py": "dbeea5aa780889785ce850f3881e8b7f67bb44594e9bc382c65b2e730884a2f4",
-    "scripts/rectification/scoring_service.py": "d1530ea3569f86747c4e6d7bccf18639e50c870526c06702bc1d9b3fec07ed9b",
-    "scripts/research/cluster_width_lib.py": "1d8f2ffa7795d62069deebb2de8e16749317037d4ed3f24a2c0092676e5cfcc6",
-    "scripts/research/probe_supply_after_six.py": "098a5b4398d6f8d996917d809c2fc922ad64031d3b70660f8ef7c876aba92438",
-    "scripts/research/reported_offset_sweep.py": "1ef4645a954da5d8374efc807aeeb2daf6e3431db771ed1e346b46ebbe83de62",
-    "scripts/research/sealed_holdout_rerun.py": "388b8af89fb89da9c3f2ce0e2bb6351ed20fd54fa19b8be34438540eed128617",
-    "scripts/shadbala.py": "cf1127cb60ce9aedc01050e7e391257d1c354b9ca735c7a2293baefb7d57e343",
-    "scripts/varga.py": "c1c8521320176e31b1f897807e50322f147654251c323731666482fc98f6df0a"
-  },
-  "hash_scope": "explicit_identity_file_sets_not_a_transitive_dependency_lock",
-  "scorer": "native_event_contribution_matrix_not_shadow_fact_ranker",
-  "delivery": "initial_signature_clusters_peak_gap_lt_8_envelope_no_answers_no_elimination",
-  "rank": "score_desc_then_sha256(benchmark_id:case_id:candidate_time)",
-  "grid": "every_minute_inclusive_not_production_two_minute_sampling",
-  "cross_midnight": "date_aware_candidates_distance_and_native_single_matrix",
-  "evaluator_sha256": "1ef4645a954da5d8374efc807aeeb2daf6e3431db771ed1e346b46ebbe83de62",
-  "replay_revision": "native_candidate_date_v3",
-  "supersedes": "candidate_date_grouped_v2_identity_and_path_not_assumed_numerically_wrong",
-  "is_blind_evaluation": false,
-  "truth_hidden_from_ranker": true,
-  "official_valid_independent_blind": false,
-  "official_blind_trial_count": 0,
-  "results_previously_seen": true,
-  "must_not_use_for_tuning": true
-}
@@ -1,42 +0,0 @@
-{
- "tag": "round7_integration_2026_11_09",
- "source_commit": "b453505d50f8d9c5ff4a472eb225a0c60a68e607",
- "policy": "common_evidence_coverage_v1",
- "policy_summary": "evidence unit = (event, layer); equal coverage -> comparable (comparable_leader = provisional_leader = winning_segment); unequal -> provisional_unequal_coverage, re-rank on units every candidate has; merged layers that cannot be split -> not_reducible (BUG-1801)",
- "narayana_strength_scope": "seed_only_v0",
- "scope": "coverage counts only, not accuracy (BUG-1644)",
- "probe": "product adjudicator compute_event_candidate_result on the replay script's own requests (sealed: full + sparse per case; offset: every shifted window); live: chart_choice astrologer leader_coverage",
- "sealed_20": {
-  "evaluations": 40,
-  "leader_status": {
-   "comparable": 40
-  },
-  "coverage_comparison": {
-   "comparable/raw_totals": 40
-  }
- },
- "reported_offset_900": {
-  "evaluations": 900,
-  "leader_status": {
-   "comparable": 863,
-   "provisional_unequal_coverage": 35,
-   "no_single_leader": 2
-  },
-  "coverage_comparison": {
-   "comparable/raw_totals": 865,
-   "provisional_unequal_coverage/common_coverage": 35
-  },
-  "not_reducible": 0,
-  "provisional_cases": 1,
-  "no_comparable_leader": 2
- },
- "live_v5_231": {
-  "evaluations": 231,
-  "leader_status": {
-   "comparable": 226,
-   "provisional_unequal_coverage": 5
-  },
-  "provisional_cases": 3,
-  "duplicate_event_runs": 0
- }
-}
@@ -1,128 +0,0 @@
-# BUG-1809 回放树：round7_integration_2026_11_09（staging b453505d）
-#
-# 构建步骤（与 BUG-1720 / BUG-1759 同一协议）：
-#   1. git archive c1a69445^ （两份回放脚本与退役打分器随 BUG-1644 删除，从这里取）
-#   2. patch -p1 < docs/research/float_split_2026_11_08_replay_overlay.patch
-#   3. patch -p1 < docs/research/round5_pvr_narayana_2026_11_08_replay_overlay.patch
-#   4. git archive b453505d scripts references | tar -x  （staging 的 scripts/、references/ 整体覆盖；
-#      仓库已删的退役文件保留回放树版本。两份回放实际加载的 67 个模块里，仍在 b453505d 的 57 个除 (c) 两个守卫文件外逐字节相同）
-#   5. 本补丁（三部分）：
-#      (a) 两份回放脚本 FREEZE / REPORT 指向新标签（r5tree → r7tree）；
-#      (b) 退役打分器 scoring_service.py 的回放适配：Narayana 定不出、该候选该事件无证据行时按现行引擎「不计分」补 0 分
-#          （BUG-1769 已记录的 IndexError 路径；本次偏移回放 3351 / 1316340 行触发，封存回放不加载该文件）；
-#      (c) 回放副本里 birth_at 守卫由 TypeError 改为 RuntimeError（REPLAY_GUARD），避免任何调用方 except TypeError 吞掉日期字符串；
-#          staging → r7guard。
-#   6. python scripts/research/sealed_holdout_rerun.py --freeze；reported_offset_sweep.py --freeze；再各自回放（PYTHONHASHSEED=0）。
-diff -ruN r5tree/scripts/rectification/scoring_service.py r7tree/scripts/rectification/scoring_service.py
--- r5tree/scripts/rectification/scoring_service.py	2026-11-09 07:17:58.857906592 +0800
-+++ r7tree/scripts/rectification/scoring_service.py	2026-11-09 07:17:58.861906532 +0800
-@@ -254,6 +254,17 @@
-     return {**request, "events": [event for event in request["events"] if is_scoreable_event(event)]}
- 
- 
-+import atexit as _atexit, sys as _sys
-+_REPLAY_EMPTY = [0]
-+_REPLAY_TOTAL = [0]
-+_atexit.register(lambda: print(f"REPLAY_ADAPTER empty_evidence_rows={_REPLAY_EMPTY[0]} of {_REPLAY_TOTAL[0]}", file=_sys.stderr))
-+
-+
-+def _replay_empty(event):
-+    _REPLAY_EMPTY[0] += 1
-+    return {"event_id": event["id"], "points": 0.0, "rule_ids": []}
-+
-+
- def build_event_contribution_matrix(
-     request: RectificationRequest,
-     row_provider: Callable[[dict[str, Any]], Sequence[CandidateScoreRow]] | None = None,
-@@ -277,10 +288,14 @@
-         sample_rows = []
-         for sampled in samples:
-             rows = list(provider(_legacy_request(scoring_request, event, sampled)))
-+            _REPLAY_TOTAL[0] += len(rows)
-             sample_rows.append([
-                 {**row, "score": adjusted["points"], "evidence": [adjusted]}
-                 for row in rows
-                for adjusted in [_kind_adjusted_evidence(event, row["evidence"][0])]
-+                # REPLAY ADAPTER (round9 integration, 2026-11-09): a candidate whose mandatory layer is missing for this
-+                # event (e.g. Narayana blocked) has no evidence row; the product adds nothing for it, so it counts 0 here.
-+                for adjusted in [_kind_adjusted_evidence(event, row["evidence"][0]) if row["evidence"]
-+                                 else _replay_empty(event)]
-             ])
-         grids = [[row["time"] for row in rows] for rows in sample_rows]
-         if any(grid != grids[0] for grid in grids[1:]) or (candidate_grid is not None and grids[0] != candidate_grid):
-diff -ruN r5tree/scripts/research/reported_offset_sweep.py r7tree/scripts/research/reported_offset_sweep.py
--- r5tree/scripts/research/reported_offset_sweep.py	2026-11-09 07:17:58.857906592 +0800
-+++ r7tree/scripts/research/reported_offset_sweep.py	2026-11-09 07:17:58.861906532 +0800
-@@ -66,8 +66,11 @@
- # 原值: docs/research/reported_offset_float_split_2026_11_08.{freeze.json,json}
- # 新值: docs/research/reported_offset_round5_pvr_narayana_2026_11_08.{freeze.json,json}
- # 原因: 产品 2026-11-08 批准；占星师第五轮裁定 1、3；与报告/卡口径一致（BUG-1759：校正功能吉凶改用默认 PVR 表 30，Narayana 改用来源版 narayana_rath、出生时刻起算；回放结果变了，另存新标签）
-FREEZE = ROOT / "docs/research/reported_offset_round5_pvr_narayana_2026_11_08.freeze.json"
-REPORT = ROOT / "docs/research/reported_offset_round5_pvr_narayana_2026_11_08.json"
-+# 原值: docs/research/reported_offset_round5_pvr_narayana_2026_11_08.{freeze.json,json}
-+# 新值: docs/research/reported_offset_round7_integration_2026_11_09.{freeze.json,json}
-+# 原因: 产品负责人 2026-11-09 批准；第七轮整合（R1 默认 p50、R2 局部传播 seed_only、可比领先）；BUG-1809（staging b453505d 全部 scripts/references 叠到回放树，回放结果变了，另存新标签）
-+FREEZE = ROOT / "docs/research/reported_offset_round7_integration_2026_11_09.freeze.json"
-+REPORT = ROOT / "docs/research/reported_offset_round7_integration_2026_11_09.json"
- LEGACY_REPORT = ROOT / "docs/research/reported_offset_2026_09_20.json"
- 
- 
-diff -ruN r5tree/scripts/research/sealed_holdout_rerun.py r7tree/scripts/research/sealed_holdout_rerun.py
--- r5tree/scripts/research/sealed_holdout_rerun.py	2026-11-09 07:17:58.857906592 +0800
-+++ r7tree/scripts/research/sealed_holdout_rerun.py	2026-11-09 07:17:58.861906532 +0800
-@@ -63,8 +63,11 @@
- # 原值: docs/research/sealed_holdout_rerun_float_split_2026_11_08.{freeze.json,json}
- # 新值: docs/research/sealed_holdout_rerun_round5_pvr_narayana_2026_11_08.{freeze.json,json}
- # 原因: 产品 2026-11-08 批准；占星师第五轮裁定 1、3；与报告/卡口径一致（BUG-1759：校正功能吉凶改用默认 PVR 表 30，Narayana 改用来源版 narayana_rath、出生时刻起算；回放结果变了，另存新标签）
-FREEZE = ROOT / "docs/research/sealed_holdout_rerun_round5_pvr_narayana_2026_11_08.freeze.json"
-REPORT = ROOT / "docs/research/sealed_holdout_rerun_round5_pvr_narayana_2026_11_08.json"
-+# 原值: docs/research/sealed_holdout_rerun_round5_pvr_narayana_2026_11_08.{freeze.json,json}
-+# 新值: docs/research/sealed_holdout_rerun_round7_integration_2026_11_09.{freeze.json,json}
-+# 原因: 产品负责人 2026-11-09 批准；第七轮整合（R1 默认 p50、R2 局部传播 seed_only、可比领先）；BUG-1809（staging b453505d 全部 scripts/references 叠到回放树，回放结果变了，另存新标签）
-+FREEZE = ROOT / "docs/research/sealed_holdout_rerun_round7_integration_2026_11_09.freeze.json"
-+REPORT = ROOT / "docs/research/sealed_holdout_rerun_round7_integration_2026_11_09.json"
- LEGACY_REPORT = ROOT / "docs/research/sealed_holdout_rerun_2026_09_20.json"
- ARCHIVE = ROOT / "docs/research/history/rectification_pre_cross_midnight_2026_09_20"
- PRODUCTION_FILES = [
-diff -ruN stg/scripts/active_rectification_event_engine.py r7g/scripts/active_rectification_event_engine.py
--- staging/scripts/active_rectification_event_engine.py	2026-11-09 07:17:58.869906413 +0800
-+++ r7guard/scripts/active_rectification_event_engine.py	2026-11-09 07:17:58.869906413 +0800
-@@ -107,7 +107,7 @@
-     and started every period from midnight of that day.
-     """
-     if not isinstance(birth_at, datetime):
-        raise TypeError("birth_at must be the candidate birth datetime")
-+        raise RuntimeError("REPLAY_GUARD birth_at must be the candidate birth datetime")
-     timeline, _, _, _ = birth_moment.mahadashas(birth_at, float(moon_longitude))
-     return timeline
- 
-@@ -125,7 +125,7 @@
-     with ``birth_moment.split_period`` at both levels.
-     """
-     if not isinstance(birth_at, datetime):
-        raise TypeError("birth_at must be the candidate birth datetime")
-+        raise RuntimeError("REPLAY_GUARD birth_at must be the candidate birth datetime")
-     if timeline is None:
-         timeline = _vimshottari_timeline(birth_at, moon_longitude)
-     _, major = dasha_analyzer.find_current(timeline, event_at)
-@@ -200,7 +200,7 @@
-     step; there is no fallback to the old algorithm.
-     """
-     if not isinstance(birth_at, datetime):
-        raise TypeError("birth_at must be the candidate birth datetime")
-+        raise RuntimeError("REPLAY_GUARD birth_at must be the candidate birth datetime")
-     if periods is None:
-         periods = _narayana_periods(ascendant_index, planet_longitudes)
-     if not periods:
-diff -ruN stg/scripts/rectification/event_probes.py r7g/scripts/rectification/event_probes.py
--- staging/scripts/rectification/event_probes.py	2026-11-09 07:17:58.869906413 +0800
-+++ r7guard/scripts/rectification/event_probes.py	2026-11-09 07:17:58.873906354 +0800
-@@ -509,7 +509,7 @@
-     if birth_at is None:
-         return None
-     if not isinstance(birth_at, datetime):
-        raise TypeError("birth_at must be the candidate birth datetime")  # a date string would silently drop the track
-+        raise RuntimeError("REPLAY_GUARD birth_at must be the candidate birth datetime")  # a date string would silently drop the track
-     periods = _narayana_periods(ascendant_index, planet_longitudes)
-     if not periods:
-         return None
@@ -1,116 +0,0 @@
-{
-  "record_version": "exposed-v3-fixed-protocol-cross-midnight-rerun-v2",
-  "extended_identity": {
-    "historical_artifacts_manifest_sha256": "6102a26a840be207b5858b3a4c0509274468ae9071e86d308cbbac7371d6864e",
-    "production_scoring_files": [
-      "scripts/active_rectification_event_engine.py",
-      "scripts/active_rectification_events.py",
-      "scripts/ashtakavarga.py",
-      "scripts/dasha_analyzer.py",
-      "scripts/divisional_charts_extended.py",
-      "scripts/domain_calculation_service.py",
-      "scripts/functional_benefics.py",
-      "scripts/jaimini.py",
-      "scripts/minute_rectification_fact_ranker_v4.py",
-      "scripts/minute_rectification_feature_facts_v4.py",
-      "scripts/narayana_dasha.py",
-      "scripts/rectification/api_service.py",
-      "scripts/rectification/candidate_contrast.py",
-      "scripts/rectification/candidate_window.py",
-      "scripts/rectification/case_holdout.py",
-      "scripts/rectification/contracts.py",
-      "scripts/rectification/dasha_transition_proximity.py",
-      "scripts/rectification/decision_policy.py",
-      "scripts/rectification/event_probes.py",
-      "scripts/rectification/refinement_packet.py",
-      "scripts/rectification/scoring_service.py",
-      "scripts/shadbala.py",
-      "scripts/varga.py"
-    ],
-    "production_scoring_sha256": "9575adabfc825fe515d571ea2a8e8550b85ccbf089f9e4b185d330821d37ace8",
-    "research_files": [
-      "scripts/research/reported_offset_sweep.py",
-      "scripts/research/cluster_width_lib.py",
-      "scripts/research/probe_supply_after_six.py",
-      "scripts/research/sealed_holdout_rerun.py",
-      "scripts/minute_rectification_blind_eval.py",
-      "scripts/minute_rectification_fact_blind_eval_v4.py",
-      "scripts/minute_rectification_holdout_validator.py"
-    ],
-    "research_implementation_sha256": "b7130fd8a9dd0a734d1c10aa0a4900a19702ecf75044ad1740e596f20cada0b5",
-    "file_sha256": {
-      "scripts/active_rectification_event_engine.py": "3bda6fcf8855bec90dd62e64fae167ce3e6afe5956549bbd26587ea59eba3d94",
-      "scripts/active_rectification_events.py": "3d8a625d24aac6f7f05a3265a23b488153daa03b5b81c6b993160011cc07eb1c",
-      "scripts/ashtakavarga.py": "c524da3f54b2ed9c080b78b33854854a0ca3d5d79b7e536bb1c3bc2be3e02852",
-      "scripts/dasha_analyzer.py": "3949b96813217fe624899856afec9f0b787c4cb0fea3a9a0d961990dbd11943a",
-      "scripts/divisional_charts_extended.py": "b13ad68c45685568d6eb17f87017624a9288e31fad74662735cd6d6564f7755d",
-      "scripts/domain_calculation_service.py": "66923a2bc69daf65c965d2973c4b9234edbec5147909eaf3609b6b377ce3c7ca",
-      "scripts/functional_benefics.py": "d34dd7745ba9db0a61e3f7a2aa1e6acd1f819961442cb3412b5dbe22c02ef032",
-      "scripts/jaimini.py": "e474b3a1d5c50de87fafdbdcc3df9bd724de004aa2495463763f5b2688b8f5e2",
-      "scripts/minute_rectification_blind_eval.py": "07aa30d8bad3278669495d5fea8be937aaf6367c8af980ff4f2fe9aa84ccc6ff",
-      "scripts/minute_rectification_fact_blind_eval_v4.py": "bdf460e3467208e8b87345c5beb2128f3e22de7eeffd68f72b281b3ad9affc92",
-      "scripts/minute_rectification_fact_ranker_v4.py": "f1a0d8d2482cd754d2e7f253a9606b0373519435a2bf58cfc3753e2b9ae0fb76",
-      "scripts/minute_rectification_feature_facts_v4.py": "4661d725d4f04935169d9d8a86ce8d0ef4061915a3a985c4912b93a6e035689e",
-      "scripts/minute_rectification_holdout_validator.py": "b60b1c752e58672beaee8b36eb3b2348169726cda070ec969db0f13e7ed715af",
-      "scripts/narayana_dasha.py": "5a924f06f76426ab587163fd8c0b5e32a6a8c6917e8670029bd84a2eb6a6172e",
-      "scripts/rectification/api_service.py": "089730097d3872a4f7b5d7dd5574bb7e9ab4d2e246d4a380800bc824d7979b95",
-      "scripts/rectification/candidate_contrast.py": "eee7da71c0574a08876404b78ca111153fa95fb83ace82b9fdca141d1992e273",
-      "scripts/rectification/candidate_window.py": "2fbc5058b12049b15fca375cdc125b8283e19f9f3e4bedcc291e344501891bef",
-      "scripts/rectification/case_holdout.py": "fe0e699b617d9e216acdfeff87af4221c1d6f4c101bf33e1dbf14e2bcd75a8b5",
-      "scripts/rectification/contracts.py": "c6dc6bd230e38f9f2126c224aae9ddbd6415304c15340f8c7cd3f06f6ef3160f",
-      "scripts/rectification/dasha_transition_proximity.py": "66d29c0b182309d17264dd2645b5cb426e28b4066d6c7004fa66f3334286753f",
-      "scripts/rectification/decision_policy.py": "80e8de4680d19a052213bb821d2a9c94143a2962235aeb27f6ad81374d2e60d5",
-      "scripts/rectification/event_probes.py": "283ac525efa5ba5b12e804d9b87e6725741ede3365659002224fffe61722bc3f",
-      "scripts/rectification/refinement_packet.py": "dbeea5aa780889785ce850f3881e8b7f67bb44594e9bc382c65b2e730884a2f4",
-      "scripts/rectification/scoring_service.py": "d1530ea3569f86747c4e6d7bccf18639e50c870526c06702bc1d9b3fec07ed9b",
-      "scripts/research/cluster_width_lib.py": "1d8f2ffa7795d62069deebb2de8e16749317037d4ed3f24a2c0092676e5cfcc6",
-      "scripts/research/probe_supply_after_six.py": "098a5b4398d6f8d996917d809c2fc922ad64031d3b70660f8ef7c876aba92438",
-      "scripts/research/reported_offset_sweep.py": "1ef4645a954da5d8374efc807aeeb2daf6e3431db771ed1e346b46ebbe83de62",
-      "scripts/research/sealed_holdout_rerun.py": "388b8af89fb89da9c3f2ce0e2bb6351ed20fd54fa19b8be34438540eed128617",
-      "scripts/shadbala.py": "cf1127cb60ce9aedc01050e7e391257d1c354b9ca735c7a2293baefb7d57e343",
-      "scripts/varga.py": "c1c8521320176e31b1f897807e50322f147654251c323731666482fc98f6df0a"
-    },
-    "hash_scope": "explicit_identity_file_sets_not_a_transitive_dependency_lock"
-  },
-  "production_identity_scope": "context_only_shadow_scorer_does_not_call_transition_proximity",
-  "frozen_at_utc": "2026-11-08T23:15:54.758056+00:00",
-  "dataset_path": "references/real_case_calibration/minute_rectification_holdout_v3.json",
-  "dataset_sha256": "45aa76ab79dfc6c6404b6252083e1aa5c7e2e2b7de811fc11042a8ad2da3f0ea",
-  "algorithm_version": "birth-time-event-fact-ranker-v4-shadow",
-  "implementation_sha256": "b32140f91dd8c65a4380ad31a29d77902b2ec2cd06495b9ce387eb7e60b68825",
-  "files": [
-    "scripts/active_rectification_event_engine.py",
-    "scripts/active_rectification_events.py",
-    "scripts/ashtakavarga.py",
-    "scripts/dasha_analyzer.py",
-    "scripts/divisional_charts_extended.py",
-    "scripts/domain_calculation_service.py",
-    "scripts/jaimini.py",
-    "scripts/minute_rectification_fact_ranker_v4.py",
-    "scripts/minute_rectification_feature_facts_v4.py",
-    "scripts/narayana_dasha.py",
-    "scripts/shadbala.py",
-    "scripts/varga.py"
-  ],
-  "historical_frozen_sha256": "f41c298dd6cdcebe7a93e632f7191954be7987f6012ca2a34cecb6e447fbf196",
-  "evaluator_sha256": "388b8af89fb89da9c3f2ce0e2bb6351ed20fd54fa19b8be34438540eed128617",
-  "ayanamsa": "raman",
-  "node_mode": "mean",
-  "candidate_radius_minutes": [
-    10
-  ],
-  "minute_step": 1,
-  "release_metrics": {
-    "top_1_rate_minimum": 0.6,
-    "top_3_rate_minimum": 0.85,
-    "mean_absolute_minute_error_maximum": 2.0,
-    "false_confirmation_rate_maximum": 0.05,
-    "correct_insufficient_evidence_rejection_rate_minimum": 0.9
-  },
-  "results_previously_seen": true,
-  "official_valid_independent_blind": false,
-  "must_not_use_for_tuning": true,
-  "tie_breaker": "sha256(benchmark_id:case_id:candidate_time); published minute is never passed to the ranker",
-  "metric_rank_definition": "competition_rank_1_plus_strictly_higher_scores_legacy_protocol",
-  "extra_metric_rank_definition": "score_desc_then_existing_opaque_sha256_total_order"
-}
@@ -0,0 +1,74 @@
+# PROGRESS：占星师验收清单的七个反例（#169 / BUG-1812）
+
+- 分支：`codex/counterexamples-20261109`（基于任务书提交，代码基线 `b453505d`）
+- 任务书：`docs/tasks/TASK-counterexamples-20261109.md`
+- 开工时 `docs/BUG_HISTORY.md` 最大编号：BUG-1804；按任务书用 BUG-1812
+- 完成日期：2026-11-09
+
+## 结论
+
+七个反例各有引擎侧与数据卡侧测试，共 16 条 Python + 7 条 TS，全部通过。三处卡上漏字段（月相 `reason`、小运 `ad_status` / `ad_reason`、大运 `cycle_index`）已补，Obama 盘事业层与基础层「月相缺失」不一致已修。
+
+## 七个反例
+
+| 反例 | 引擎函数 | 卡上字段 | 修前结果 | 修后结果 | 测试名 |
+| --- | --- | --- | --- | --- | --- |
+| (a) 缺月相 | `build_career_analysis` → `analyze_career` / `derive_functional_benefic_malefic` | `base.natal.functional_benefic_malefic.movable_lagna_moon`、`domain_sections.career.career_analysis.*.movable_lagna_moon` | Obama 基础层 `derived_from_table_30_note` / waning / neutral，事业层 `blocked` / `moon_paksha_missing`；卡上 `movable_lagna_moon` 没有 `reason`，两种情形分不开 | 两层同源（事业层拿到 D1 经度）；缺月相时两层都 `blocked` + `moon_paksha_missing`，卡上带 `reason`，只有月亮被阻塞、其余六颗照常分类 | `test_a_missing_moon_paksha_blocks_only_the_moon_and_both_layers_agree` / `(a) missing moon paksha` |
+| (b) 低档强弱未决 | `narayana_rath.compare_signs` / `calc_narayana_rath` / `shadbala.calc_shadbala` + `attach_parameter_sensitivity` | `base.timing.narayana.method_status`、`domain_sections.career.shadbala` | 未决通过 `method_status` 能到卡，但没有一条测试证明「未定步骤被点名且不补当前大运」；Shadbala 的 `parameter_sensitive` 也没有端到端用例 | `comparison_status=unresolved`、`blocked_step=seed`、`unresolved_steps=["seed"]`、`chat_use=no_judgment`、`md` 缺失进 gaps；Jupiter 保留 `minimum_check=parameter_sensitive` 与开放问题，不写达标 / 不达标 | `test_b_low_grade_comparison_undecided_names_the_step_and_scores_nothing` / `(b) undecided low-grade comparison` |
+| (c) 小运未决、大运已定 | `narayana_rath.antardasha` / `branch_under` / `product_narayana` / `build_dasha_systems` | `base.timing.narayana.ad_status`、`ad_reason`；`dasha_systems.systems.narayana.ad_status`、`ad_reason`；`method_status.running` | `current_dasha` 在 AD 未决时干脆没有 `ad`，卡上「没有 ad」与「没算」分不开；`running` 只写主运 | 大运 Taurus（第 2 轮）与日期原样保留；`ad_status=blocked`、`ad_reason=saturn_ketu_priority_unresolved`；`running` 写「子运未决（saturn_ketu_priority_unresolved）」；没有任何补出来的 AD | `test_c_ad_undecided_keeps_the_md_and_names_the_reason` / `(c) undecided sub-period…` |
+| (d) 月亮自参照 | `neecha_bhanga_rows` / `build_yoga_conditions` | `domain_sections.career.independence.self_reference_only`、`.self_reference` | 已有单测与 2026-11-08 的管线测试，但没有「引擎行 → fixture → 卡」这一环 | 自参照格局进 `self_reference_only`、带自参照说明；注入这一行与不注入时 `support` 逐项相同（一票都不加） | `test_d_moon_self_reference_is_never_independent_confirmation` / `(d) Moon-only self-reference…` |
+| (e) 出生余额 | `chart_dasha.birth_moment.antardashas` / `_consultation_dasha_sub_periods` | `base.timing.vimshottari.antardashas_in_mahadasha[].parent_period_start` / `visible_period_start`、`base.timing.vimshottari.mahadasha` | golden 三张盘都是成年人，卡上的裁剪分支从未被真实数据覆盖 | 合成年轻盘（2020-05-01，余额 Mercury 剩 10.72 年）：出生小运同时带父段起点 `2020-04-12` 与可见起点 `2020-05-01`，9 条子运边界与未裁剪版本逐日相同 | `test_e_birth_balance_keeps_the_full_parent_start_beside_the_visible_one` / `(e) birth balance…` |
+| (f) 第二轮大运 | `narayana_rath.mahadasha_sequence` / `product_narayana` / `birth_moment.next_cycle_mahadashas` | `base.timing.narayana.md.cycle_index` | `_period` 不抄 `cycle_index`，卡上说不出「第几轮」 | 当前大运 Aquarius、`cycle_index=2`、`years=11` 与引擎行一致；24 行里只有一行覆盖参考年龄；Vimshottari 跨 120 年后当前段 `cycle=2`（Moon，10 年） | `test_f_second_cycle_is_named_and_the_running_row_is_unique` / `(f) second cycle…` |
+| (g) 0 年大运 | `narayana_rath.mahadasha_sequence`（`zero_duration`） | `base.timing.narayana.md`、`narayana.next` | 卡上只靠 `years > 0` 间接排除 0 年段，引擎的 `zero_duration` 标记没被用上 | 0 年段（Sagittarius 第 1 轮、`start_age == end_age == 38`）留在 `mahadasha` 数据里并带 `zero_duration`；参考日恰在段起点时当前大运是 Aquarius（第 2 轮，11 年）；`next` 不含 0 年段 | `test_g_zero_year_period_is_kept_in_the_data_and_never_current` / `(g) zero-year period…` |
+
+## 修前 / 修后
+
+修前用基线 `b453505d` 另开一次性工作树（`counterexamples-before`，跑完已删除），把同一批测试与 fixture 放进去：
+
+- 数据卡 7 条：3 条失败 —— `(a)`、`(c)`、`(f)`，正好是卡上三处漏字段。
+- 引擎 16 条：`test_fixture_is_fresh_for_every_counterexample` 失败（引擎输出与 fixture 不一致），其余通过（断言读 fixture，不重跑引擎）。
+
+修后同一条命令全绿（见下）。工作树已删，`git worktree list` 不再有 `counterexamples-before`。
+
+## 改了什么
+
+- `scripts/consultation_native_layers.py`
+  - `build_career_analysis`：把 D1 经度并入交给 `analyze_career` 的行（原值只给 `house` / `sign`），事业层与基础层同源。
+  - `build_dasha_systems`：Narayana 大运带 `cycle_index` / `zero_duration`，整行带 `ad_status` / `ad_reason`（都照抄引擎）。
+- `scripts/narayana_legacy_label.py`
+  - `_period` 带 `cycle_index`（有值时）与 `zero_duration`（为真时）。
+  - `rath_narayana_view`：AD 未决时 `current_dasha` 照给大运，另带 `ad_status=blocked` / `ad_reason`；AD 选定时 `ad_status=selected`。
+- `frontend/src/lib/consultation-evidence-card.ts`
+  - `functionalProfileFields`：`movable_lagna_moon` 带 `reason`。
+  - `NARAYANA_MD_CARD_KEYS` 加 `cycle_index`。
+  - `DASHA_ROW_KEYS` 加 `ad_status` / `ad_reason`。
+  - `baseTiming`：`narayana` 带 `ad_status` / `ad_reason`。
+  - `narayanaRunningLine`：AD 未决时写「子运未决（原因）」。
+  - `timingFollowing`：显式跳过 `zero_duration: true` 的行。
+- `frontend/src/mastra/consultation-workflow.ts`：`timingKeys` 白名单加 `adstatus` / `adreason` / `cycleindex` / `zeroduration`（否则这些字段在投给模型的 `timing` 里被裁掉）。
+- 新增：`scripts/research/capture_counterexample_fixtures_20261109.py`、7 份 fixture、`tests/test_counterexamples_engine_20261109.py`、`frontend/tests/consult-counterexamples-card-20261109.test.ts`。
+
+## 偏离与原因
+
+1. **卡上 MD 没有 `parent_period_start` / `visible_period_start`。** 任务书要求「卡上 MD 与 AD 同时带」，但 MD 这两个字段的唯一生产者是冻结文件 `scripts/jyotish_api_server.py`：`_consultation_birth_moment_mahadashas` 只返回 `lord` / `start` / `end`，`boundary()` 因此拿不到父段起点。任务书把该文件列为只读，所以卡上只断言出生小运的两个字段，MD 的裁剪在引擎层断言（`sub_periods.current.mahadasha.start` 早于出生、`end` 晚于出生；`birth_moment.antardashas` 的边界不移动）。若要 MD 也上卡，需要另开一单改冻结文件。
+2. **改了 `scripts/narayana_legacy_label.py`。** 任务书「要改的文件」没列它，但它是 `current_dasha` 的唯一生产者：`ad_status` / `cycle_index` 不在这里带出，卡上就没有来源。只加字段，不动 `narayana_rath` 的公式、规则与年数。
+3. **改了 `frontend/src/mastra/consultation-workflow.ts` 的 `timingKeys`。** 不加白名单，新字段在 `toModelOutput` 的 `timing` 投影里被裁掉，卡上永远看不到。只加四个键名，不加字段、不改结构。
+4. **本工作树的环境。** `frontend/node_modules` 与 `.venv` 不随 worktree 复制：node_modules 用目录联接指向 `.worktrees/agent-loop/frontend/node_modules`（git 不跟踪），Python 用同一工作树的 venv（`.venv\Scripts\python.exe`）。不是代码改动。
+
+## 跑过的命令与结果
+
+| 检查 | 命令 | 结果 |
+| --- | --- | --- |
+| fixture 生成 | `.venv\Scripts\python scripts/research/capture_counterexample_fixtures_20261109.py` | 7 份写入 `frontend/tests/fixtures/counterexamples-20261109/` |
+| 引擎测试 | `.venv\Scripts\python -m pytest tests/test_counterexamples_engine_20261109.py -q` | 16 passed |
+| 数据卡测试 | `node node_modules/tsx/dist/cli.mjs --test tests/consult-counterexamples-card-20261109.test.ts` | 7 pass / 0 fail |
+| 相关卡片回归 | 同上再带 `consult-self-reference-pipeline-20261108`、`consult-evidence-card-20260927`、`consult-evidence-card-v2-20260927` | 36 tests / 36 pass / 0 fail / 0 cancelled，退出码 0 |
+| 类型检查 | `frontend\node_modules\.bin\tsc --noEmit -p frontend` | 0 错 |
+| lint | `npm run lint`（frontend） | 0 error / 62 warning（既有警告，未新增） |
+
+前端全量测试与 Python 快速门按任务书留给 Claude 验收。
+
+## 没做 / 留给验收
+
+- 回答阶段（模型读卡后怎么写）的盲测由 Claude 另做。
+- 生产 / staging 部署核对由 Claude 在合入后做。
@@ -0,0 +1,170 @@
+# TASK：占星师验收清单的七个反例，从引擎到数据卡端到端测试（2026-11-09）
+
+- 分支：`codex/counterexamples-20261109`（基于 staging `b453505d`）
+- Bug 编号：BUG-1812（开工时核对 `docs/BUG_HISTORY.md` 的最大编号；1809–1830 已预留给本批任务）
+- 负责验收：Claude（jyotisha-2a）
+
+## 背景
+
+占星师 2026-11-08《近期六轮问题闭环与网站验收清单》§7 原文：“另加入缺月相、低档强弱未决、AD未决但MD已定、Moon自参照、出生余额、第二周期与0年等反例。”
+
+§5 的要求：每个问题沿“计算 → 数据卡 → 模型输入 → 回答 → 导出”五个阶段验收。原文：“不要只测第一阶段，再从最后一阶段‘读起来不错’推断中间过程正确。”数据卡阶段必须检查：“是否丢失必要分盘、父段、限制、冲突；不能把未送达写成未计算。”
+
+§3 P0 的通过标准：“只阻塞依赖项；旧候选不可冒充主结果；缺失不能产生评分；独立确定的MD保留。”
+
+**本任务把七个反例都做成从计算一路到数据卡的测试。** 测试暴露的数据卡漏字段，在本任务内补上。回答阶段的盲测由 Claude 另做。
+
+## 测试怎么做（现有机制）
+
+仓库里没有在同一次运行里同时调用 Python 和 TS 数据卡的测试。现有做法分两层：
+
+1. **Python 层**：调用真实引擎函数，断言状态字段。
+2. **TS 层**：用真实引擎输出抓取的 golden，经过真实前端链路生成数据卡。链路是 `consultationWorkflowResponseSchema.parse` → `toAgentConsultationContext` → `toModelOutput` → `buildEvidenceCard`。
+   - 模板：`frontend/tests/consult-self-reference-pipeline-20261108.test.ts`。它对 golden 做 `structuredClone`，把形状与引擎一致的行注入 `chart.consultation_native_layers.*`。
+   - golden 文件：`frontend/tests/fixtures/consult-evidence-card-golden.json`。生成脚本是 `scripts/research/capture_consult_evidence_card_golden.py`，含 Jobs、Obama、Taylor 三张公开盘。
+
+**端到端的要求**：每个反例都要有一个 Python 测试。这个测试调用真实引擎函数，把输出的那一段 JSON 写成 fixture，放在 `frontend/tests/fixtures/counterexamples-20261109/<case>.json`。
+
+- 生成方式写成脚本 `scripts/research/capture_counterexample_fixtures_20261109.py`，可重复运行。
+- Python 测试断言：“现在引擎的输出 == fixture”，防止 fixture 过期。
+
+TS 测试读同一份 fixture，经真实前端链路生成数据卡，再断言卡上的字段。**不得在 TS 里手造引擎形状。**
+
+只用公开名人盘（golden 里已有的那些）或合成输入（`tests/neutral_subject.py`，或像 `tests/test_round6_r1_r2_narayana.py` 那样构造的 lagna + 经度字典）。
+
+## 七个反例（b453505d 实测位置，行号以符号为准）
+
+### (a) 缺月相
+
+- **引擎**：`scripts/functional_benefics.py` 的 `_pvr_roles`（约 :276–301）。动宫上升、`moon_paksha=None` 时，结果为 `movable_lagna_moon = {status:"blocked", reason:"moon_paksha_missing", ...}`，`blocked_planets:["Moon"]`。只阻塞月亮。
+- **已知真实缺陷，本任务要修**：Obama 盘（golden 内）出现两层结果不一致。
+  - `chart.modules.functional_benefic_malefic.movable_lagna_moon` 为 waning / neutral。
+  - `consultation_native_layers.career_analysis.functional_benefic_malefic.movable_lagna_moon` 却是 `blocked / moon_paksha_missing`。
+  - 原因：`scripts/consultation_native_layers.py` 的 `build_career_analysis`（约 :1280）传入的是 `_affliction_planets`（约 :1210）的行。这些行没有度数，`scripts/career_analysis.py`（约 :93）因此算不出月相。
+  - 修法：让事业层拿到月亮与太阳的经度，或直接拿到基础层已经算好的月相。修复点只在 `consultation_native_layers.py` 或 `career_analysis.py`。
+- **数据卡**：`frontend/src/lib/consultation-evidence-card.ts` 的 `functionalProfileFields`（约 :1325–1345）只抄了 `{status, moon_paksha, role}`，丢掉了 `reason`。要补上 `reason`。
+- **断言**：
+  - 真缺月相（合成盘不给月亮经度）时：卡上为 `blocked` 且 `reason` 为 `moon_paksha_missing`；只有月亮被阻塞，其他行星的功能吉凶照常；月亮不参与吉凶计票。
+  - Obama 盘修复后，两层月相一致。
+
+### (b) 低档强弱未决
+
+- **Narayana**：`scripts/narayana_rath.py` 的 `compare_signs`（约 :562–614）返回 `status:"undecided"`、`comparison_status:"unresolved"`、`blocked_level`。
+  - `calc_narayana_rath`（约 :1001）在种子判不出时为 `status:"blocked"`、`blocked_step:"seed"`；年数判不出时为 `status:"partial_blocked"`、`blocked_step:"years"`。
+  - 产品视图：`scripts/narayana_legacy_label.py` 的 `rath_narayana_view`（约 :252）与 `narayana_method_status`（约 :194）。
+  - 卡上：`consultation-evidence-card.ts` 的 `narayanaMethodStatus`（约 :1703）与 `baseTiming`（约 :1786、:1834–1846）。
+- **Shadbala**：
+  - `scripts/shadbala_service/sensitivity.py` 的 `attach_parameter_sensitivity`（约 :237）；
+  - `scripts/shadbala_display.py` 的 `row_fields`（约 :47–63），输出 `minimum_check:"parameter_sensitive"`；
+  - 卡上：`SHADBALA_CARD_KEYS` 与 `shadbalaCardRows`（约 :1438–1449）。
+- **断言**：
+  - 卡上 Narayana 的未决步骤被点名（种子 / 年数 / 双宫主），不出现“已确认”字样的状态，`used_for_scoring` 为 false。
+  - Shadbala 为 `parameter_sensitive` 的行星，卡上保留这个标记，并且不以“达标 / 不达标”计入支持或压力（第五轮第 4 条：“只能标为 parameter_sensitive，不能用它决定聊天里的吉凶结论”）。
+
+### (c) 小运（AD）未决，大运（MD）已定
+
+- **引擎**：`narayana_rath.py` 的 `antardasha`（约 :921）返回 `{status:"blocked", reason, rows:[]}`；`branch_under`（约 :1362）返回 `ad:None, ad_status:"blocked", ad_reason`。
+  - 默认 `STRENGTH_SCOPE_DEFAULT = seed_only_v0`（BUG-1784）。默认设置下，现实的触发条件是土星与计都同在大运星座（`saturn_ketu_priority_unresolved`），或者像 `tests/test_round6_r1_r2_narayana.py:86` 那样在测试里显式放宽范围。两种都要测。
+- **卡上的缺口**：`scripts/consultation_native_layers.py` 的 `build_dasha_systems`（约 :1676–1760）在 AD 未决时省略了 `nar["ad"]`。`consultation-evidence-card.ts` 的 `baseTiming`（约 :1826–1830）只在有值时才抄 md / ad / pd。所以现在 AD 未决在卡上只表现为“没有 ad”，与“没算”分不开。这违反 §5 的“不能把未送达写成未计算”。
+- **修法**：卡上 Narayana 增加 `ad_status` 与 `ad_reason`（从引擎原样带过来），`narayanaRunningLine`（约 :1769）在 AD 未决时写明“子运未决”。
+- **断言**：
+  - MD 的星座和日期保留，并与引擎相等；
+  - AD 为未决，且原因被点名；
+  - 卡上不出现由旧算法补出的 AD。
+
+### (d) 月亮自参照
+
+- **引擎**：`scripts/main_yoga_detectors.py` 的 `neecha_bhanga_rows`（约 :129–163）输出 `self_reference`、`independent_confirmation_allowed:false`。
+- **卡上**：`consultation-evidence-card.ts` 的 `yogaSelfReference`（约 :1008–1030）与 `evidenceCardIndependence`（约 :1065）。
+- **已有测试**：`frontend/tests/consult-self-reference-pipeline-20261108.test.ts`、`tests/test_round6_self_reference.py`。本任务不重写这两个，只补端到端缺的一环：引擎真输出 → fixture → 卡。
+- **断言**：只由月亮自参照形成的格局，不进入 `support`；同一事实重复引用只算一票（现有规则见 BUG-1773、BUG-1774）。
+
+### (e) 出生余额
+
+- **引擎**：`scripts/chart_dasha/birth_moment.py` 的 `antardashas(md, visible_from=)`（约 :88–112）：`parent_period_start` 是未裁剪的父段起点，`visible_period_start = max(start, birth)`。
+- **服务器**：`scripts/jyotish_api_server.py` 的 `_consultation_dasha_sub_periods` 下的 `boundary()`（约 :1504–1514）。**只读，不改。**
+- **卡上**：`CLIPPED_PERIOD_KEYS`（约 :1349）与 `baseTiming`（约 :1786–1812）。
+- **缺口**：golden 里三位都是成年人，参考日 2026-09-27 时都已不在出生大运内，所以卡上的“裁剪”分支从未被真实数据覆盖。请用一张合成的年轻盘（参考日落在出生 MD 之内）。
+- **断言**：
+  - 卡上 MD 与 AD 同时带 `parent_period_start` 和 `visible_period_start`，二者不同，且都等于引擎值；
+  - 子运边界不因裁剪而改变（第五轮第 5 条）。
+
+### (f) 第二周期
+
+- **引擎**：
+  - `narayana_rath.py` 的 `mahadasha_sequence`（约 :820）设 `cycle_index` 为 1 或 2（第二周期用 12 减去年数）；
+  - `branch_under` 设 `md_cycle`；
+  - Vimshottari 用 `birth_moment.py` 的 `next_cycle_mahadashas`（约 :35）。
+- **缺口**：卡上 `NARAYANA_MD_CARD_KEYS`（约 :1782）没有 `cycle_index`，卡说不出“这是第二周期”。要补上 `cycle_index`。
+- **断言**：
+  - 合成盘的参考年龄落在 Narayana 第二周期内时，卡上当前 MD 带 `cycle_index: 2`，年数等于引擎值；
+  - Vimshottari 跨过 120 年后，当前段等于引擎值（可参照 `tests/test_vimshottari_second_cycle.py`）；
+  - 其他大运 `not_available_yet` 的行，在卡上进入 gaps，不显示为“当前”。
+
+### (g) 0 年大运
+
+- **引擎**：`narayana_rath.py` 的 `mahadasha_sequence`（约 :828/:860/:886）设 `zero_duration`。
+  - 判断“当前”的区间是半开的 `[start, end)`：`_running_age`（约 :1355）、`narayana_legacy_label.py` 的 `_running`（约 :235）、`consultation_native_layers.py` 的 `_running`（约 :1505）。
+- **卡上**：`timingFollowing`（约 :1633、过滤在 :1667–1671）。golden 里没有 `zero_duration:true` 的行。
+- **断言**：
+  - 合成一张会产生 0 年 Narayana 段的盘（可参照 `tests/test_round5_item_reply_n1_k3_20261108.py:181` 的人马座 0 年 / 12 年构造），参考日恰好落在该段起点；
+  - 卡上当前 MD 不是 0 年段；
+  - 0 年段保留在数据里，并带 `zero_duration` 标记（若卡上需要展示时序，就补上这个键）。
+
+## 要改的文件（以实际代码为准）
+
+- **新建**：
+  - `scripts/research/capture_counterexample_fixtures_20261109.py`
+  - `frontend/tests/fixtures/counterexamples-20261109/*.json`
+  - `tests/test_counterexamples_engine_20261109.py`
+  - `frontend/tests/consult-counterexamples-card-20261109.test.ts`
+- **修改**：
+  - `frontend/src/lib/consultation-evidence-card.ts`：只改上面点名的函数和常量，即 `functionalProfileFields`、`NARAYANA_MD_CARD_KEYS`、`baseTiming` 的 Narayana 段、`narayanaRunningLine`、`timingFollowing`。**不做结构性重排**，压缩是 #170 的事。
+  - `scripts/consultation_native_layers.py`：(a) 的事业层月相、(c) 的 `ad_status` 透传。
+  - `scripts/career_analysis.py`：如 (a) 的修复需要。
+- 若卡片新增字段导致 `frontend/tests/consult-evidence-card-20260927.test.ts` 的体积上限超出，只许按实测增量放宽，并写三栏注释。
+
+## 不要碰
+
+- `references/**`（占星师方法文件）、`frontend/src/lib/consult-preset-prompts.ts`、`references/product_runtime_contract.md`。
+- 冻结文件：
+  - `scripts/jyotish_api_server.py`（只读）；
+  - `scripts/functional_benefics.py`、`scripts/narayana_dasha.py`、`scripts/dasha_analyzer.py`、`scripts/domain_calculation_service.py`、`scripts/shadbala.py`、`scripts/ashtakavarga.py`、`scripts/jaimini.py`、`scripts/varga.py`。
+- `scripts/narayana_rath.py`：不改公式与规则，本任务只测。
+- 其他单子的地盘：
+  - `frontend/src/mastra/consultation-tools.ts` 全文（#171/#173 在改）；
+  - `frontend/src/app/api/consult/route.ts`（#172）；
+  - `consultation-evidence-card.ts` 里的 `EVIDENCE_LOOKUP_SECTIONS` 与 `evidenceCardIndependence` 的结构（#170/#171）。
+- 不改大运计算公式，不改计分权重。
+
+## 验收标准
+
+1. 七个反例各有一条 Python 引擎测试，以及一条读同一 fixture 的数据卡测试，共 14 条以上，全部通过。
+2. fixture 由脚本重新生成后与仓内一致。Python 测试断言“引擎当前输出 == fixture”。
+3. (a)、(c)、(f) 三处卡上缺字段已补，Obama 两层月相不一致已修。每处修复各有一条“修前会失败”的回归断言。在 PROGRESS 里写明：先在未修代码上跑出失败，再修。
+4. 所有“未决 / 阻塞”的反例都断言了不计分、不增票：不进入 `support` 或压力计数，`used_for_scoring` 为 false。
+5. 改既有断言时写“原值 / 新值 / 原因”三栏注释。原因写：`占星师 2026-11-08 验收清单 §7 反例：未决不得写成未计算`。
+6. 在 `docs/BUG_HISTORY.md` 末尾追加 BUG-1812。记录 Obama 事业层月相不一致、卡上缺 `reason` / `ad_status` / `cycle_index` 这三处。不写个人出生资料。
+7. 只用公开名人盘或合成数据。
+
+## 自测命令（Windows / PowerShell）
+
+- `frontend\node_modules\.bin\tsc --noEmit -p frontend`
+- 在 `frontend` 下运行 `npm run lint`
+- 在 `frontend` 下运行 `node node_modules/tsx/dist/cli.mjs --test tests/consult-counterexamples-card-20261109.test.ts tests/consult-self-reference-pipeline-20261108.test.ts tests/consult-evidence-card-20260927.test.ts tests/consult-evidence-card-v2-20260927.test.ts`
+- `.venv\Scripts\python scripts/research/capture_counterexample_fixtures_20261109.py`
+- `.venv\Scripts\python -m pytest tests/test_counterexamples_engine_20261109.py tests/test_round6_r1_r2_narayana.py tests/test_round5_item_reply_n1_k3_20261108.py tests/test_round6_self_reference.py -q`
+
+前端全量测试和 Python 快速门由 Claude 在验收时跑。
+
+## 进度记录
+
+`docs/tasks/PROGRESS-counterexamples-20261109.md`。
+
+其中要有一张七行的表，列为：反例 | 引擎函数 | 卡上字段 | 修前结果 | 修后结果 | 测试名。
+
+## 何时停下来选“需拍板”
+
+- 某个反例要修好，就必须改冻结文件（例如 `functional_benefics.py`）或 `narayana_rath.py` 的规则时。
+- 引擎本身给出的状态就与占星师的裁定相反时，例如 AD 未决却把 MD 也一起阻塞了。不要自己改引擎，写清楚现象。
+- 找不到能产生某个反例（例如 0 年段落在当前）的合法合成输入时。写清楚试过哪些输入。
@@ -1340,7 +1340,14 @@ function functionalProfileFields(value: unknown): JsonRecord {
  }
  const moon = record(source.movable_lagna_moon);
  if (typeof moon.status === "string") {
-    out.movable_lagna_moon = { status: moon.status, moon_paksha: copy(moon.moon_paksha) ?? null, role: copy(moon.role) ?? null };
+    // 占星师 2026-11-08 验收清单 §7 反例「缺月相」：卡上原只抄 status / moon_paksha / role，
+    // 丢掉 reason，于是「未给月相」与「表里没这一格」分不开。reason 原样带上（BUG-1812）。
+    out.movable_lagna_moon = {
+      status: moon.status,
+      moon_paksha: copy(moon.moon_paksha) ?? null,
+      role: copy(moon.role) ?? null,
+      ...(typeof moon.reason === "string" ? { reason: moon.reason } : {}),
+    };
  }
  out.functional_profile_note = EVIDENCE_CARD_LABELS.functionalProfileNote;
  return out;
@@ -1668,7 +1675,10 @@ function timingFollowing(
  const index = narayanaRows.findIndex((row) => record(row).sign === md.sign && record(row).start_age === md.start_age);
  if (index < 0) return;
  // A zero-year sign (start age = end age) is not a period anyone lives through.
+  // 占星师 2026-11-08 验收清单 §7 反例「0 年大运」：引擎行带 zero_duration:true 时同样跳过，
+  // 不靠 years>0 这一条间接推断（BUG-1812）。
  const following = narayanaRows.slice(index + 1).filter((row) => {
+    if (record(row).zero_duration === true) return false;
    const years = record(row).years;
    return typeof years !== "number" || years > 0;
  });
@@ -1775,11 +1785,18 @@ export function narayanaRunningLine(row: Record<string, unknown>): string | unde
    return `${label} ${period.sign}${house}${dates}`;
  };
  const parts = [part("主运", row.md), part("子运", row.ad)].filter((item): item is string => Boolean(item));
+  // 占星师 2026-11-08 验收清单 §7 反例「小运未决但大运已定」：大运在、小运未决时，
+  // 这一行写「子运未决（原因）」，不把未决留成空白（BUG-1812）。
+  if (!record(row.ad).sign && row.ad_status === "blocked") {
+    const reason = typeof row.ad_reason === "string" && row.ad_reason ? `（${row.ad_reason}）` : "";
+    parts.push(`子运未决${reason}`);
+  }
  if (!parts.length) return undefined;
  return `${parts.join("；")}（名义计算边界）。写成单体系候选时也照写主运、子运的星座和起止`;
 }

-const NARAYANA_MD_CARD_KEYS = ["sign", "lord", "years", "start_age", "end_age", "dual_lord_basis", "dual_lord_basis_en"] as const;
+// 占星师 2026-11-08 验收清单 §7 反例「第二轮大运」：卡上要说得出这是第几轮大运（cycle_index 照抄引擎行）。
+const NARAYANA_MD_CARD_KEYS = ["sign", "lord", "years", "start_age", "end_age", "cycle_index", "dual_lord_basis", "dual_lord_basis_en"] as const;
 const LEGACY_MD_CARD_KEYS = ["sign", "start_age", "end_age"] as const;

 function baseTiming(timing: Record<string, unknown>, gaps: string[], horizon: TimingHorizon | null = null) {
@@ -1825,6 +1842,11 @@ function baseTiming(timing: Record<string, unknown>, gaps: string[], horizon: Ti
  }
  const remaining = copy(currentNarayana.remaining_years);
  if (remaining !== undefined) narayana.remaining_years = remaining;
+  // 占星师 2026-11-08 验收清单 §7 反例：小运未决时 md 旁边写明 ad_status / ad_reason（引擎原值）。
+  for (const key of ["ad_status", "ad_reason"] as const) {
+    const value = copy(currentNarayana[key]);
+    if (value !== undefined) narayana[key] = value as Json;
+  }
  // BUG-1215 / BUG-1601: say which Narayana this is (the engine's short label).
  const algorithmLabel = copy(narayanaModule.algorithm_label);
  if (typeof algorithmLabel === "string" && algorithmLabel && (narayana.md || narayanaModule.status === "blocked")) {
@@ -1875,7 +1897,9 @@ const DASHA_ROW_KEYS = [
  "applicable", "status", "reason", "profile", "rule", "year_rule_status", "confidence_status", "depth_status",
  "date_basis", "cycle_end", "md", "ad", "pd",
  // 产品 2026-11-07: the next Narayana AD; where a ported upstream row came from and its status.
-  "next_ad", "source", "status_note",
+  // 占星师 2026-11-08 验收清单 §7 反例「小运未决但大运已定」：ad_status / ad_reason 原样带上，
+  // 卡上才分得清「子运未决」与「子运没算」（BUG-1812）。
+  "next_ad", "source", "status_note", "ad_status", "ad_reason",
  // 占星师第三轮审查 2026-11-07: a gap / overlap between consecutive periods (absent when continuous).
  "continuity_issue",
 ] as const;
@@ -410,6 +410,9 @@ const timingKeys = new Set([
  // Without these keys the projection kept `current_dasha` as an empty object
  // and dropped the pratyantar dates, although the engine had both.
  "md", "ad", "pd", "years", "startage", "endage", "remainingyears", "pratyantardashatimeline",
+  // 占星师 2026-11-08 验收清单 §7 反例：小运未决的 ad_status / ad_reason、第几轮大运的
+  // cycle_index、0 年段的 zero_duration 要随 current_dasha 一起到达数据卡（BUG-1812）。
+  "adstatus", "adreason", "cycleindex", "zeroduration",
  // BUG-1643: the next Vimshottari mahadasha with its antardashas (engine `dasha_sub_periods.next_mahadasha`).
  "nextmahadasha",
  // BUG-1215: the running Narayana's label travels with it.
@@ -0,0 +1,177 @@
+// 占星师 2026-11-08 验收清单 §7 的七个反例：从引擎到数据卡的端到端断言（BUG-1812）。
+//
+// 引擎切片由 scripts/research/capture_counterexample_fixtures_20261109.py 抓进
+// fixtures/counterexamples-20261109/*.json（同一个脚本也是 Python 测试的期望值来源），
+// 这里把它们注入一张真实 golden workflow，经真实链路
+// parse → toAgentConsultationContext → toModelOutput → buildEvidenceCard，
+// 再断言卡上的状态、日期与「未决不计票」。TS 里不手造引擎形状。
+import assert from "node:assert/strict";
+import { readFileSync } from "node:fs";
+import test from "node:test";
+
+import { buildEvidenceCard, EVIDENCE_CARD_SELF_REFERENCE_NOTE, evidenceCardFactId } from "../src/lib/consultation-evidence-card.ts";
+import { createConsultationPlan } from "../src/lib/consultation-plan.ts";
+import { consultationWorkflowResponseSchema, toAgentConsultationContext, toModelOutput } from "../src/mastra/consultation-workflow.ts";
+
+// eslint-disable-next-line @typescript-eslint/no-explicit-any -- test reads deep engine JSON without a schema
+type Json = Record<string, any>;
+
+const FIXTURES = new URL("./fixtures/counterexamples-20261109/", import.meta.url);
+const golden = JSON.parse(readFileSync(new URL("./fixtures/consult-evidence-card-golden.json", import.meta.url), "utf8")) as {
+  charts: Array<{ id: string; workflow: Json }>;
+};
+
+function fixture(name: string): Json {
+  return JSON.parse(readFileSync(new URL(`${name}.json`, FIXTURES), "utf8")) as Json;
+}
+
+/** A real golden workflow (public AA chart) with only the engine slices of one case replaced. */
+function cardFor(name: string, inject: (chart: Json, engine: Json) => void) {
+  const data = fixture(name);
+  const workflow = structuredClone(golden.charts.find((chart) => chart.id === "barack_obama")!.workflow);
+  inject(workflow.chart, data.engine);
+  const context = toAgentConsultationContext(consultationWorkflowResponseSchema.parse(workflow));
+  const packet = toModelOutput(context, createConsultationPlan({ userIntent: "q", theme: "career" }), "career");
+  return { card: buildEvidenceCard([{ domain: "career", packet, context }]), engine: data.engine };
+}
+
+test("(a) missing moon paksha: the card names the reason and both layers agree", () => {
+  const { card, engine } = cardFor("a_missing_moon_paksha", (chart, data) => {
+    chart.modules.functional_benefic_malefic = data.blocked.base;
+    chart.consultation_native_layers.career_analysis.functional_benefic_malefic = data.blocked.career;
+  });
+  const natal = (card.base.natal.functional_benefic_malefic as Json).movable_lagna_moon as Json;
+  assert.deepEqual(natal, { status: "blocked", moon_paksha: null, role: null, reason: "moon_paksha_missing" });
+  // Only the Moon is blocked: the other planets keep their role, the Moon joins no list.
+  assert.ok((card.base.natal.functional_benefic_malefic as Json).functional_benefics.length > 0);
+  for (const key of ["functional_benefics", "functional_malefics", "functional_neutrals"]) {
+    assert.ok(!((card.base.natal.functional_benefic_malefic as Json)[key] as string[]).includes("Moon"), key);
+  }
+  // Career section shows the same verdict as the base layer (engine fix: one paksha source).
+  const career = ((card.domain_sections.career!.career_analysis as Json).functional_benefic_malefic as Json).movable_lagna_moon as Json;
+  assert.equal(career.status, natal.status);
+  assert.equal(career.reason, natal.reason);
+  assert.equal(engine.blocked.career.movable_lagna_moon.reason, "moon_paksha_missing");
+});
+
+test("(b) undecided low-grade comparison: a named blocked step, no invented running period", () => {
+  const { card } = cardFor("b_low_grade_undecided", (chart, data) => {
+    chart.modules.narayana_dasha = data.narayana.product;
+    chart.shadbala = data.shadbala;
+  });
+  const narayana = card.base.timing.narayana as Json;
+  assert.equal(narayana.status, "blocked");
+  assert.equal(narayana.md, undefined);
+  assert.deepEqual(narayana.method_status.unresolved_steps, ["seed"]);
+  assert.equal(narayana.method_status.status, "blocked");
+  assert.equal(narayana.method_status.chat_use, "no_judgment");
+  assert.ok(card.meta.gaps.includes("base.narayana.md"), JSON.stringify(card.meta.gaps));
+  // Shadbala: the flipping planet keeps parameter_sensitive and is never written as pass / fail.
+  const jupiter = (card.domain_sections.career!.shadbala as Json).Jupiter as Json;
+  assert.equal(jupiter.minimum_check, "parameter_sensitive");
+  assert.ok(Array.isArray(jupiter.minimum_open_questions) && jupiter.minimum_open_questions.length > 0);
+  const text = JSON.stringify(jupiter);
+  for (const wording of ["达标", "不达标", "meets_minimum"]) assert.ok(!text.includes(wording), wording);
+});
+
+test("(c) undecided sub-period keeps the mahadasha and says the sub-period is undecided", () => {
+  const { card, engine } = cardFor("c_ad_blocked_md_known", (chart, data) => {
+    chart.modules.narayana_dasha = data.narayana;
+    // The dasha_systems layer the card reads comes from build_dasha_systems (same engine run);
+    // only this one row is replaced so the injected Narayana and the table agree.
+    const layer = chart.consultation_native_layers.dasha_systems as Json;
+    layer.rows = [...layer.rows.filter((row: Json) => row.system !== "narayana"),
+      { ...data.dasha_systems_narayana }];
+  });
+  const narayana = card.base.timing.narayana as Json;
+  assert.equal(narayana.md.sign, engine.md.sign);
+  assert.equal(narayana.md.cycle_index, engine.md.cycle_index);
+  assert.equal(narayana.md.start_age, engine.md.start_age);
+  assert.equal(narayana.ad, undefined);
+  assert.equal(narayana.ad_status, "blocked");
+  assert.equal(narayana.ad_reason, "saturn_ketu_priority_unresolved");
+  // The dasha table row carries the same status and one plain sentence (not "not computed").
+  const row = ((card.base.timing.dasha_systems as Json).systems as Json).narayana as Json;
+  assert.equal(row.md.sign, engine.md.sign);
+  assert.equal(row.ad, undefined);
+  assert.equal(row.ad_status, "blocked");
+  assert.equal(row.ad_reason, "saturn_ketu_priority_unresolved");
+  assert.match(narayana.method_status.running as string, /子运未决/);
+  assert.match(narayana.method_status.running as string, /saturn_ketu_priority_unresolved/);
+});
+
+test("(d) Moon-only self-reference adds no support vote on the card", () => {
+  const { card } = cardFor("d_moon_self_reference", (chart, data) => {
+    // The engine caps yoga rows (YOGA_CONDITION_ROW_CAP); the injected row takes the last slot.
+    const rows = chart.consultation_native_layers.yoga_conditions.rows as Json[];
+    chart.consultation_native_layers.yoga_conditions.rows = [...rows.slice(0, rows.length - 1), data.row];
+    chart.consultation_native_layers.yoga_domains.rows =
+      [...chart.consultation_native_layers.yoga_domains.rows, { name: "Neecha Bhanga", hit: true, domain: "career" }];
+  });
+  const section = card.domain_sections.career! as Json;
+  assert.ok(section.yogas.hits.includes("Neecha Bhanga"));
+  const independence = section.independence as Json;
+  assert.deepEqual(independence.self_reference_only, ["Neecha Bhanga"]);
+  assert.equal(independence.self_reference["Neecha Bhanga"], EVIDENCE_CARD_SELF_REFERENCE_NOTE);
+  const jupiter = evidenceCardFactId(card.base, "Jupiter")!;
+  assert.ok(independence.facts[jupiter].includes("Neecha Bhanga（自参照）"));
+  // 不增票：同一张盘不注入这一行时，support 与注入后逐项相同（自参照格局一票都不加）。
+  const baseline = cardFor("d_moon_self_reference", () => {}).card.domain_sections.career! as Json;
+  assert.deepEqual([...(independence.support ?? [])].sort(), [...((baseline.independence as Json).support ?? [])].sort());
+  assert.deepEqual((baseline.independence as Json).self_reference_only, undefined);
+});
+
+test("(e) birth balance: the clipped sub-period keeps its parent start beside the visible one", () => {
+  const { card, engine } = cardFor("e_birth_balance", (chart, data) => {
+    chart.modules.dasha_sub_periods = data.sub_periods;
+  });
+  const vim = card.base.timing.vimshottari as Json;
+  const md = vim.mahadasha as Json;
+  assert.equal(md.lord, engine.sub_periods.current.mahadasha.lord);
+  assert.equal(md.start, engine.sub_periods.current.mahadasha.start);
+  assert.ok(md.start < engine.birth.date && engine.birth.date < md.end);
+  const boundaries = vim.antardashas_in_mahadasha as Json[];
+  assert.equal(boundaries.length, engine.sub_periods.boundaries.length);
+  const clipped = boundaries.filter((row) => !Array.isArray(row) && typeof row.parent_period_start === "string");
+  assert.equal(clipped.length, 1, JSON.stringify(boundaries));
+  assert.equal(clipped[0]!.parent_period_start, engine.sub_periods.boundaries[0].parent_period_start);
+  assert.equal(clipped[0]!.visible_period_start, engine.birth.date);
+  assert.ok(clipped[0]!.parent_period_start < clipped[0]!.visible_period_start);
+  // The running sub-period is still the engine's, with its own unclipped boundaries.
+  assert.equal((vim.antardasha as Json).lord, engine.sub_periods.current.antardasha.lord);
+  assert.equal((vim.antardasha as Json).start, engine.sub_periods.current.antardasha.start);
+});
+
+test("(f) second cycle: the card names the cycle and keeps the engine's years", () => {
+  const { card, engine } = cardFor("f_second_cycle", (chart, data) => {
+    chart.modules.narayana_dasha = data.narayana;
+  });
+  const md = (card.base.timing.narayana as Json).md as Json;
+  const engineMd = engine.narayana.current_dasha.md as Json;
+  assert.equal(engineMd.cycle_index, 2);
+  assert.equal(md.cycle_index, 2);
+  assert.equal(md.sign, engineMd.sign);
+  assert.equal(md.years, engineMd.years);
+  assert.equal(md.start_age, engineMd.start_age);
+});
+
+test("(g) zero-year period: never the running one, and it never appears as the next sign", () => {
+  const { card, engine } = cardFor("g_zero_duration", (chart, data) => {
+    chart.modules.narayana_dasha = data.narayana;
+  });
+  const narayana = card.base.timing.narayana as Json;
+  const md = narayana.md as Json;
+  const zeroSigns = (engine.narayana.mahadasha as Json[])
+    .filter((row) => row.zero_duration === true)
+    .map((row) => row.sign as string);
+  assert.ok(zeroSigns.length > 0);
+  assert.ok(!zeroSigns.includes(md.sign), `${md.sign} is a zero-year sign`);
+  assert.equal(md.sign, engine.narayana.current_dasha.md.sign);
+  assert.equal(md.start_age, engine.narayana.current_dasha.md.start_age);
+  for (const next of (narayana.next as Json[]) ?? []) {
+    assert.ok(!zeroSigns.includes(next.sign as string), `next sign ${next.sign} is a zero-year sign`);
+    // A zero-year row has start_age === end_age; no "next" row may be one.
+    assert.notEqual(next.start_age, next.end_age);
+    if (typeof next.years === "number") assert.ok(next.years > 0);
+  }
+});
@@ -0,0 +1,537 @@
+{
+ "ayanamsa": "raman",
+ "case": "a_missing_moon_paksha",
+ "engine": {
+  "ascendant": "Capricorn",
+  "blocked": {
+   "base": {
+    "ambiguous_cell_notes": {
+     "Mercury": {
+      "applies_to_profile": "bphs_exception_profile",
+      "basis": "专条明确支持",
+      "basis_en": "sign verse",
+      "pvr_table_30_role": "benefic",
+      "summary": "偏吉，保留 6 宫主职责",
+      "summary_en": "mildly benefic; keeps 6th-house duties"
+     }
+    },
+    "ascendant": "Capricorn",
+    "assertion_label": "parameter_sensitive",
+    "blocked_planets": [
+     "Moon"
+    ],
+    "conditions": [
+     "Neutral/mixed is association-dependent, not an absence of risk.",
+     "Natural nature, strength, maraka roles and dasha activation remain separate.",
+     "Strict yogakarakas do not exhaust the text's broader yoga-producing associations.",
+     "Dual lordship needs an integrated judgment of both houses (PVR §13.2)."
+    ],
+    "dual_lordship_note": "PVR §13.2：管两个星座的行星要把两宫的指示综合判断；表 30 是综合后的摘要，不是只看一个宫位的机械规则。",
+    "dual_lordship_note_en": "PVR §13.2: for a planet owning two rasis the two indications must be judiciously combined; Table 30 is that integrated summary, not a mechanical one-house rule.",
+    "effect_on_confidence": "高严谨模式下必须叠加功能性宫主吉凶与自然吉凶；若功能属性与自然属性冲突，应降低置信度或显式标记冲突。",
+    "functional_benefics": [
+     "Mercury",
+     "Saturn",
+     "Venus"
+    ],
+    "functional_malefics": [
+     "Jupiter",
+     "Mars"
+    ],
+    "functional_neutrals": [
+     "Sun"
+    ],
+    "functional_profile": "pvr_table30_v1",
+    "functional_profile_ruling": "占星师 2026-11-08 第五轮裁定 3：默认 PVR 表 30；BPHS 特殊格另存为具名 profile，只在被明确选中时参与打分",
+    "lagna_lord": "Saturn",
+    "movable_lagna_moon": {
+     "listed_in_table_30": false,
+     "moon_paksha": null,
+     "reason": "moon_paksha_missing",
+     "role": null,
+     "rule": "Moon is not listed for movable rasis, because his functional nature depends on whether he is waxing or waning. Waxing Moon is a natural benefic and he becomes a functional malefic with quadrant ownership. Waning Moon, on the other hand, is a natural malefic and quadrant ownership makes him functionally neutral.",
+     "rule_zh": "PVR 表 30 不列动星座上升的月亮：盈月（自然吉星）管角宫 → 功能凶；亏月（自然凶星）管角宫 → 功能中性",
+     "source": "PVR §13.2 note under Table 30 (printed p. 166-167; PDF p. 178-179)",
+     "status": "blocked"
+    },
+    "owned_houses": {
+     "Jupiter": [
+      3,
+      12
+     ],
+     "Mars": [
+      4,
+      11
+     ],
+     "Mercury": [
+      6,
+      9
+     ],
+     "Moon": [
+      7
+     ],
+     "Saturn": [
+      1,
+      2
+     ],
+     "Sun": [
+      8
+     ],
+     "Venus": [
+      5,
+      10
+     ]
+    },
+    "prior_rulings": [],
+    "profile": "pvr_integrated_approach_13_2_table_30",
+    "profile_divergence": {
+     "cells": [
+      {
+       "bphs_exception_profile": "malefic",
+       "kind": "movable_lagna_moon_paksha",
+       "planet": "Moon",
+       "pvr_table30_v1": "not_classified"
+      }
+     ],
+     "display_rule": "两档不同时并列显示，不合并成一张表；默认档参与判断，另一档只作对照",
+     "other_profile": "bphs_exception_profile",
+     "other_profile_roles": {
+      "functional_benefics": [
+       "Mercury",
+       "Saturn",
+       "Venus"
+      ],
+      "functional_malefics": [
+       "Jupiter",
+       "Mars",
+       "Moon"
+      ],
+      "functional_neutrals": [
+       "Sun"
+      ],
+      "yogakarakas": [
+       "Venus"
+      ]
+     },
+     "other_profile_scored": false,
+     "selected_profile": "pvr_table30_v1",
+     "status": "differs"
+    },
+    "profile_is_not": "bphs_ch34_general_with_sign_exceptions_v3_166q4",
+    "profile_source": "PVR Narasimha Rao, Vedic Astrology: An Integrated Approach, §13.2 Table 30 (p. 167; PDF p. 179)",
+    "role_basis": {
+     "Jupiter": "PVR §13.2 Table 30",
+     "Mars": "PVR §13.2 Table 30",
+     "Mercury": "PVR §13.2 Table 30",
+     "Moon": "PVR §13.2 Table 30 does not list this Moon; paksha unknown: blocked (moon_paksha_missing)",
+     "Saturn": "PVR §13.2 Table 30",
+     "Sun": "PVR §13.2 Table 30",
+     "Venus": "PVR §13.2 Table 30"
+    },
+    "rule_reference": "references/functional-role-profile.md",
+    "source": "strict_functional_benefic_malefic_v2",
+    "status": "used",
+    "yogakaraka_definition": "distinct lordships: one of 4/7/10 and one of 5/9",
+    "yogakarakas": [
+     "Venus"
+    ]
+   },
+   "career": {
+    "ambiguous_cell_notes": {
+     "Mercury": {
+      "applies_to_profile": "bphs_exception_profile",
+      "basis": "专条明确支持",
+      "basis_en": "sign verse",
+      "pvr_table_30_role": "benefic",
+      "summary": "偏吉，保留 6 宫主职责",
+      "summary_en": "mildly benefic; keeps 6th-house duties"
+     }
+    },
+    "ascendant": "Capricorn",
+    "assertion_label": "parameter_sensitive",
+    "blocked_planets": [
+     "Moon"
+    ],
+    "conditions": [
+     "Neutral/mixed is association-dependent, not an absence of risk.",
+     "Natural nature, strength, maraka roles and dasha activation remain separate.",
+     "Strict yogakarakas do not exhaust the text's broader yoga-producing associations.",
+     "Dual lordship needs an integrated judgment of both houses (PVR §13.2)."
+    ],
+    "dual_lordship_note": "PVR §13.2：管两个星座的行星要把两宫的指示综合判断；表 30 是综合后的摘要，不是只看一个宫位的机械规则。",
+    "dual_lordship_note_en": "PVR §13.2: for a planet owning two rasis the two indications must be judiciously combined; Table 30 is that integrated summary, not a mechanical one-house rule.",
+    "effect_on_confidence": "高严谨模式下必须叠加功能性宫主吉凶与自然吉凶；若功能属性与自然属性冲突，应降低置信度或显式标记冲突。",
+    "functional_benefics": [
+     "Mercury",
+     "Saturn",
+     "Venus"
+    ],
+    "functional_malefics": [
+     "Jupiter",
+     "Mars"
+    ],
+    "functional_neutrals": [
+     "Sun"
+    ],
+    "functional_profile": "pvr_table30_v1",
+    "functional_profile_ruling": "占星师 2026-11-08 第五轮裁定 3：默认 PVR 表 30；BPHS 特殊格另存为具名 profile，只在被明确选中时参与打分",
+    "lagna_lord": "Saturn",
+    "movable_lagna_moon": {
+     "listed_in_table_30": false,
+     "moon_paksha": null,
+     "reason": "moon_paksha_missing",
+     "role": null,
+     "rule": "Moon is not listed for movable rasis, because his functional nature depends on whether he is waxing or waning. Waxing Moon is a natural benefic and he becomes a functional malefic with quadrant ownership. Waning Moon, on the other hand, is a natural malefic and quadrant ownership makes him functionally neutral.",
+     "rule_zh": "PVR 表 30 不列动星座上升的月亮：盈月（自然吉星）管角宫 → 功能凶；亏月（自然凶星）管角宫 → 功能中性",
+     "source": "PVR §13.2 note under Table 30 (printed p. 166-167; PDF p. 178-179)",
+     "status": "blocked"
+    },
+    "owned_houses": {
+     "Jupiter": [
+      3,
+      12
+     ],
+     "Mars": [
+      4,
+      11
+     ],
+     "Mercury": [
+      6,
+      9
+     ],
+     "Moon": [
+      7
+     ],
+     "Saturn": [
+      1,
+      2
+     ],
+     "Sun": [
+      8
+     ],
+     "Venus": [
+      5,
+      10
+     ]
+    },
+    "prior_rulings": [],
+    "profile": "pvr_integrated_approach_13_2_table_30",
+    "profile_divergence": {
+     "cells": [
+      {
+       "bphs_exception_profile": "malefic",
+       "kind": "movable_lagna_moon_paksha",
+       "planet": "Moon",
+       "pvr_table30_v1": "not_classified"
+      }
+     ],
+     "display_rule": "两档不同时并列显示，不合并成一张表；默认档参与判断，另一档只作对照",
+     "other_profile": "bphs_exception_profile",
+     "other_profile_roles": {
+      "functional_benefics": [
+       "Mercury",
+       "Saturn",
+       "Venus"
+      ],
+      "functional_malefics": [
+       "Jupiter",
+       "Mars",
+       "Moon"
+      ],
+      "functional_neutrals": [
+       "Sun"
+      ],
+      "yogakarakas": [
+       "Venus"
+      ]
+     },
+     "other_profile_scored": false,
+     "selected_profile": "pvr_table30_v1",
+     "status": "differs"
+    },
+    "profile_is_not": "bphs_ch34_general_with_sign_exceptions_v3_166q4",
+    "profile_source": "PVR Narasimha Rao, Vedic Astrology: An Integrated Approach, §13.2 Table 30 (p. 167; PDF p. 179)",
+    "role_basis": {
+     "Jupiter": "PVR §13.2 Table 30",
+     "Mars": "PVR §13.2 Table 30",
+     "Mercury": "PVR §13.2 Table 30",
+     "Moon": "PVR §13.2 Table 30 does not list this Moon; paksha unknown: blocked (moon_paksha_missing)",
+     "Saturn": "PVR §13.2 Table 30",
+     "Sun": "PVR §13.2 Table 30",
+     "Venus": "PVR §13.2 Table 30"
+    },
+    "rule_reference": "references/functional-role-profile.md",
+    "source": "strict_functional_benefic_malefic_v2",
+    "status": "used",
+    "yogakaraka_definition": "distinct lordships: one of 4/7/10 and one of 5/9",
+    "yogakarakas": [
+     "Venus"
+    ]
+   }
+  },
+  "paksha_known": {
+   "base": {
+    "ambiguous_cell_notes": {
+     "Mercury": {
+      "applies_to_profile": "bphs_exception_profile",
+      "basis": "专条明确支持",
+      "basis_en": "sign verse",
+      "pvr_table_30_role": "benefic",
+      "summary": "偏吉，保留 6 宫主职责",
+      "summary_en": "mildly benefic; keeps 6th-house duties"
+     }
+    },
+    "ascendant": "Capricorn",
+    "assertion_label": "parameter_sensitive",
+    "blocked_planets": [],
+    "conditions": [
+     "Neutral/mixed is association-dependent, not an absence of risk.",
+     "Natural nature, strength, maraka roles and dasha activation remain separate.",
+     "Strict yogakarakas do not exhaust the text's broader yoga-producing associations.",
+     "Dual lordship needs an integrated judgment of both houses (PVR §13.2)."
+    ],
+    "dual_lordship_note": "PVR §13.2：管两个星座的行星要把两宫的指示综合判断；表 30 是综合后的摘要，不是只看一个宫位的机械规则。",
+    "dual_lordship_note_en": "PVR §13.2: for a planet owning two rasis the two indications must be judiciously combined; Table 30 is that integrated summary, not a mechanical one-house rule.",
+    "effect_on_confidence": "高严谨模式下必须叠加功能性宫主吉凶与自然吉凶；若功能属性与自然属性冲突，应降低置信度或显式标记冲突。",
+    "functional_benefics": [
+     "Mercury",
+     "Saturn",
+     "Venus"
+    ],
+    "functional_malefics": [
+     "Jupiter",
+     "Mars"
+    ],
+    "functional_neutrals": [
+     "Moon",
+     "Sun"
+    ],
+    "functional_profile": "pvr_table30_v1",
+    "functional_profile_ruling": "占星师 2026-11-08 第五轮裁定 3：默认 PVR 表 30；BPHS 特殊格另存为具名 profile，只在被明确选中时参与打分",
+    "lagna_lord": "Saturn",
+    "movable_lagna_moon": {
+     "listed_in_table_30": false,
+     "moon_paksha": "waning",
+     "role": "neutral",
+     "rule": "Moon is not listed for movable rasis, because his functional nature depends on whether he is waxing or waning. Waxing Moon is a natural benefic and he becomes a functional malefic with quadrant ownership. Waning Moon, on the other hand, is a natural malefic and quadrant ownership makes him functionally neutral.",
+     "rule_zh": "PVR 表 30 不列动星座上升的月亮：盈月（自然吉星）管角宫 → 功能凶；亏月（自然凶星）管角宫 → 功能中性",
+     "source": "PVR §13.2 note under Table 30 (printed p. 166-167; PDF p. 178-179)",
+     "status": "derived_from_table_30_note"
+    },
+    "owned_houses": {
+     "Jupiter": [
+      3,
+      12
+     ],
+     "Mars": [
+      4,
+      11
+     ],
+     "Mercury": [
+      6,
+      9
+     ],
+     "Moon": [
+      7
+     ],
+     "Saturn": [
+      1,
+      2
+     ],
+     "Sun": [
+      8
+     ],
+     "Venus": [
+      5,
+      10
+     ]
+    },
+    "prior_rulings": [],
+    "profile": "pvr_integrated_approach_13_2_table_30",
+    "profile_divergence": {
+     "cells": [
+      {
+       "bphs_exception_profile": "malefic",
+       "kind": "movable_lagna_moon_paksha",
+       "planet": "Moon",
+       "pvr_table30_v1": "neutral"
+      }
+     ],
+     "display_rule": "两档不同时并列显示，不合并成一张表；默认档参与判断，另一档只作对照",
+     "other_profile": "bphs_exception_profile",
+     "other_profile_roles": {
+      "functional_benefics": [
+       "Mercury",
+       "Saturn",
+       "Venus"
+      ],
+      "functional_malefics": [
+       "Jupiter",
+       "Mars",
+       "Moon"
+      ],
+      "functional_neutrals": [
+       "Sun"
+      ],
+      "yogakarakas": [
+       "Venus"
+      ]
+     },
+     "other_profile_scored": false,
+     "selected_profile": "pvr_table30_v1",
+     "status": "differs"
+    },
+    "profile_is_not": "bphs_ch34_general_with_sign_exceptions_v3_166q4",
+    "profile_source": "PVR Narasimha Rao, Vedic Astrology: An Integrated Approach, §13.2 Table 30 (p. 167; PDF p. 179)",
+    "role_basis": {
+     "Jupiter": "PVR §13.2 Table 30",
+     "Mars": "PVR §13.2 Table 30",
+     "Mercury": "PVR §13.2 Table 30",
+     "Moon": "PVR §13.2 Table 30 note: movable-lagna Moon by paksha",
+     "Saturn": "PVR §13.2 Table 30",
+     "Sun": "PVR §13.2 Table 30",
+     "Venus": "PVR §13.2 Table 30"
+    },
+    "rule_reference": "references/functional-role-profile.md",
+    "source": "strict_functional_benefic_malefic_v2",
+    "status": "used",
+    "yogakaraka_definition": "distinct lordships: one of 4/7/10 and one of 5/9",
+    "yogakarakas": [
+     "Venus"
+    ]
+   },
+   "career": {
+    "ambiguous_cell_notes": {
+     "Mercury": {
+      "applies_to_profile": "bphs_exception_profile",
+      "basis": "专条明确支持",
+      "basis_en": "sign verse",
+      "pvr_table_30_role": "benefic",
+      "summary": "偏吉，保留 6 宫主职责",
+      "summary_en": "mildly benefic; keeps 6th-house duties"
+     }
+    },
+    "ascendant": "Capricorn",
+    "assertion_label": "parameter_sensitive",
+    "blocked_planets": [],
+    "conditions": [
+     "Neutral/mixed is association-dependent, not an absence of risk.",
+     "Natural nature, strength, maraka roles and dasha activation remain separate.",
+     "Strict yogakarakas do not exhaust the text's broader yoga-producing associations.",
+     "Dual lordship needs an integrated judgment of both houses (PVR §13.2)."
+    ],
+    "dual_lordship_note": "PVR §13.2：管两个星座的行星要把两宫的指示综合判断；表 30 是综合后的摘要，不是只看一个宫位的机械规则。",
+    "dual_lordship_note_en": "PVR §13.2: for a planet owning two rasis the two indications must be judiciously combined; Table 30 is that integrated summary, not a mechanical one-house rule.",
+    "effect_on_confidence": "高严谨模式下必须叠加功能性宫主吉凶与自然吉凶；若功能属性与自然属性冲突，应降低置信度或显式标记冲突。",
+    "functional_benefics": [
+     "Mercury",
+     "Saturn",
+     "Venus"
+    ],
+    "functional_malefics": [
+     "Jupiter",
+     "Mars"
+    ],
+    "functional_neutrals": [
+     "Moon",
+     "Sun"
+    ],
+    "functional_profile": "pvr_table30_v1",
+    "functional_profile_ruling": "占星师 2026-11-08 第五轮裁定 3：默认 PVR 表 30；BPHS 特殊格另存为具名 profile，只在被明确选中时参与打分",
+    "lagna_lord": "Saturn",
+    "movable_lagna_moon": {
+     "listed_in_table_30": false,
+     "moon_paksha": "waning",
+     "role": "neutral",
+     "rule": "Moon is not listed for movable rasis, because his functional nature depends on whether he is waxing or waning. Waxing Moon is a natural benefic and he becomes a functional malefic with quadrant ownership. Waning Moon, on the other hand, is a natural malefic and quadrant ownership makes him functionally neutral.",
+     "rule_zh": "PVR 表 30 不列动星座上升的月亮：盈月（自然吉星）管角宫 → 功能凶；亏月（自然凶星）管角宫 → 功能中性",
+     "source": "PVR §13.2 note under Table 30 (printed p. 166-167; PDF p. 178-179)",
+     "status": "derived_from_table_30_note"
+    },
+    "owned_houses": {
+     "Jupiter": [
+      3,
+      12
+     ],
+     "Mars": [
+      4,
+      11
+     ],
+     "Mercury": [
+      6,
+      9
+     ],
+     "Moon": [
+      7
+     ],
+     "Saturn": [
+      1,
+      2
+     ],
+     "Sun": [
+      8
+     ],
+     "Venus": [
+      5,
+      10
+     ]
+    },
+    "prior_rulings": [],
+    "profile": "pvr_integrated_approach_13_2_table_30",
+    "profile_divergence": {
+     "cells": [
+      {
+       "bphs_exception_profile": "malefic",
+       "kind": "movable_lagna_moon_paksha",
+       "planet": "Moon",
+       "pvr_table30_v1": "neutral"
+      }
+     ],
+     "display_rule": "两档不同时并列显示，不合并成一张表；默认档参与判断，另一档只作对照",
+     "other_profile": "bphs_exception_profile",
+     "other_profile_roles": {
+      "functional_benefics": [
+       "Mercury",
+       "Saturn",
+       "Venus"
+      ],
+      "functional_malefics": [
+       "Jupiter",
+       "Mars",
+       "Moon"
+      ],
+      "functional_neutrals": [
+       "Sun"
+      ],
+      "yogakarakas": [
+       "Venus"
+      ]
+     },
+     "other_profile_scored": false,
+     "selected_profile": "pvr_table30_v1",
+     "status": "differs"
+    },
+    "profile_is_not": "bphs_ch34_general_with_sign_exceptions_v3_166q4",
+    "profile_source": "PVR Narasimha Rao, Vedic Astrology: An Integrated Approach, §13.2 Table 30 (p. 167; PDF p. 179)",
+    "role_basis": {
+     "Jupiter": "PVR §13.2 Table 30",
+     "Mars": "PVR §13.2 Table 30",
+     "Mercury": "PVR §13.2 Table 30",
+     "Moon": "PVR §13.2 Table 30 note: movable-lagna Moon by paksha",
+     "Saturn": "PVR §13.2 Table 30",
+     "Sun": "PVR §13.2 Table 30",
+     "Venus": "PVR §13.2 Table 30"
+    },
+    "rule_reference": "references/functional-role-profile.md",
+    "source": "strict_functional_benefic_malefic_v2",
+    "status": "used",
+    "yogakaraka_definition": "distinct lordships: one of 4/7/10 and one of 5/9",
+    "yogakarakas": [
+     "Venus"
+    ]
+   }
+  }
+ },
+ "reference_date": "2026-11-09",
+ "source": "scripts/research/capture_counterexample_fixtures_20261109.py"
+}
@@ -0,0 +1,35 @@
+{
+ "ayanamsa": "raman",
+ "case": "d_moon_self_reference",
+ "engine": {
+  "row": {
+   "category": "neecha_bhanga",
+   "conditions_met": [
+    "该星入旺星座的主星月亮在月亮起第1宫（角宫）"
+   ],
+   "formation_evidence": {
+    "cancellation_reasons": [
+     {
+      "condition": "exaltation_sign_lord_kendra_from_moon",
+      "independent_confirmation_allowed": false,
+      "planet": "Moon",
+      "reference": "Moon",
+      "relative_house": 1,
+      "self_reference": true
+     }
+    ],
+    "independent_condition_count": 0,
+    "self_reference_only": true
+   },
+   "name": "Neecha Bhanga",
+   "planets": {
+    "Jupiter": {}
+   },
+   "running_dasha_levels": [],
+   "status": "computed"
+  },
+  "yoga_conditions_status": "executed"
+ },
+ "reference_date": "2026-11-09",
+ "source": "scripts/research/capture_counterexample_fixtures_20261109.py"
+}
@@ -0,0 +1,202 @@
+{
+ "ayanamsa": "raman",
+ "case": "e_birth_balance",
+ "engine": {
+  "balance": {
+   "elapsed_years": 6.28,
+   "lord": "Mercury",
+   "remaining_years": 10.72
+  },
+  "birth": {
+   "date": "2020-05-01",
+   "hour": 10,
+   "minute": 30,
+   "second": 0
+  },
+  "dasha_periods": [
+   {
+    "end": "2031-01-18",
+    "lord": "Mercury",
+    "start": "2014-01-18",
+    "years": 17
+   },
+   {
+    "end": "2038-01-18",
+    "lord": "Ketu",
+    "start": "2031-01-18",
+    "years": 7
+   },
+   {
+    "end": "2058-01-18",
+    "lord": "Venus",
+    "start": "2038-01-18",
+    "years": 20
+   },
+   {
+    "end": "2064-01-18",
+    "lord": "Sun",
+    "start": "2058-01-18",
+    "years": 6
+   },
+   {
+    "end": "2074-01-18",
+    "lord": "Moon",
+    "start": "2064-01-18",
+    "years": 10
+   },
+   {
+    "end": "2081-01-17",
+    "lord": "Mars",
+    "start": "2074-01-18",
+    "years": 7
+   },
+   {
+    "end": "2099-01-18",
+    "lord": "Rahu",
+    "start": "2081-01-17",
+    "years": 18
+   },
+   {
+    "end": "2115-01-19",
+    "lord": "Jupiter",
+    "start": "2099-01-18",
+    "years": 16
+   },
+   {
+    "end": "2134-01-19",
+    "lord": "Saturn",
+    "start": "2115-01-19",
+    "years": 19
+   }
+  ],
+  "moon_lon": 111.5956938195655,
+  "sub_periods": {
+   "boundaries": [
+    {
+     "end": "2021-02-17",
+     "lord": "Sun",
+     "parent_period_start": "2020-04-12",
+     "start": "2020-05-01",
+     "visible_period_start": "2020-05-01"
+    },
+    {
+     "end": "2022-07-19",
+     "lord": "Moon",
+     "start": "2021-02-17"
+    },
+    {
+     "end": "2023-07-16",
+     "lord": "Mars",
+     "start": "2022-07-19"
+    },
+    {
+     "end": "2026-02-02",
+     "lord": "Rahu",
+     "start": "2023-07-16"
+    },
+    {
+     "end": "2028-05-10",
+     "lord": "Jupiter",
+     "start": "2026-02-02"
+    },
+    {
+     "end": "2031-01-18",
+     "lord": "Saturn",
+     "start": "2028-05-10"
+    }
+   ],
+   "boundary_count": 6,
+   "current": {
+    "antardasha": {
+     "end": "2028-05-10",
+     "lord": "Jupiter",
+     "start": "2026-02-02"
+    },
+    "mahadasha": {
+     "end": "2031-01-18",
+     "lord": "Mercury",
+     "start": "2014-01-18"
+    }
+   },
+   "method": "vimshottari_antardasha_proportional",
+   "next": {
+    "end": "2031-01-18",
+    "lord": "Saturn",
+    "start": "2028-05-10"
+   },
+   "next_mahadasha": {
+    "boundaries": [
+     {
+      "end": "2031-06-16",
+      "lord": "Ketu",
+      "start": "2031-01-18"
+     },
+     {
+      "end": "2032-08-15",
+      "lord": "Venus",
+      "start": "2031-06-16"
+     },
+     {
+      "end": "2032-12-21",
+      "lord": "Sun",
+      "start": "2032-08-15"
+     },
+     {
+      "end": "2033-07-22",
+      "lord": "Moon",
+      "start": "2032-12-21"
+     },
+     {
+      "end": "2033-12-18",
+      "lord": "Mars",
+      "start": "2033-07-22"
+     },
+     {
+      "end": "2035-01-06",
+      "lord": "Rahu",
+      "start": "2033-12-18"
+     },
+     {
+      "end": "2035-12-12",
+      "lord": "Jupiter",
+      "start": "2035-01-06"
+     },
+     {
+      "end": "2037-01-20",
+      "lord": "Saturn",
+      "start": "2035-12-12"
+     },
+     {
+      "end": "2038-01-18",
+      "lord": "Mercury",
+      "start": "2037-01-20"
+     }
+    ],
+    "mahadasha": {
+     "end": "2038-01-18",
+     "lord": "Ketu",
+     "start": "2031-01-18"
+    }
+   },
+   "pratyantar_dasha_timeline": {
+    "boundary": "candidate window inside the running antardasha; not a verified event date",
+    "current": {
+     "end": "2027-01-26",
+     "lord": "Mercury",
+     "start": "2026-10-01"
+    },
+    "next": {
+     "end": "2027-03-16",
+     "lord": "Ketu",
+     "start": "2027-01-26"
+    },
+    "status": "ready"
+   },
+   "source": "chart.dasha.periods + chart_dasha.birth_moment.antardashas",
+   "status": "ready",
+   "summary": "当前 Mercury 大运下的小运为 Jupiter，边界 2026-02-02 至 2028-05-10。更细的 Pratyantardasha 与行运触发不在本层计算范围内。"
+  }
+ },
+ "reference_date": "2026-11-09",
+ "source": "scripts/research/capture_counterexample_fixtures_20261109.py"
+}
@@ -1289,6 +1289,16 @@ def build_career_analysis(engine, chart: dict, modules: dict, body: dict) -> dic
    if asc_sign not in SIGNS:
        return _blocked("ascendant_sign_missing")
    planets = _affliction_planets(engine, chart)
+    # 原值：事业层只拿到 house / sign（_affliction_planets 的行），analyze_career 内的
+    #   moon_paksha_from_planets 读不到经度，动星座上升的 movable_lagna_moon 因此 blocked；
+    #   同一张盘的基础层却按 D1 经度给出 waning / neutral，两层结论不一致。
+    # 新值：把 D1 经度一并交给 analyze_career，两层用同一个盈亏判定。
+    # 原因：占星师 2026-11-08 验收清单 §7 反例「缺月相」：同一张盘两层必须同源（BUG-1812）。
+    d1_planets = _record(chart.get("planets"))
+    career_planets = {}
+    for name, row in planets.items():
+        lon = _planet_lon(d1_planets.get(name))
+        career_planets[name] = {**row, **({"lon": lon} if lon is not None else {})}
    semantic_modules = {**modules, "chart": {"ascendant": chart.get("ascendant"), "planets": chart.get("planets")}}
    western = modules.get("western_evidence_packet") if isinstance(modules.get("western_evidence_packet"), dict) else None
    vimsopaka = modules.get("vimsopaka_semantic_summary") if isinstance(modules.get("vimsopaka_semantic_summary"), dict) else None
@@ -1300,7 +1310,7 @@ def build_career_analysis(engine, chart: dict, modules: dict, body: dict) -> dic
    industry = body.get("user_industry") or body.get("industry")
    industry = str(industry).strip() if isinstance(industry, str) and industry.strip() else None
    analysis = analyze_career(
-        planets, asc_sign,
+        career_planets, asc_sign,
        d10_positions=_record(modules.get("varga_full")).get("D10_Dasamsa"),
        dasha_info=_record(modules.get("dasha")),
        shadbala=_record(modules.get("shadbala")),
@@ -1719,7 +1729,17 @@ def build_dasha_systems(chart: dict, modules: dict, *, birth_payload: dict, refe
                if houses:
                    period["lord_houses"] = houses
            period.update(_narayana_dates(birth, item, zone))
+            # 占星师 2026-11-08 验收清单 §7 反例「第二轮大运」/「0 年大运」：卡上要能说出
+            # 这是第几轮、哪一行是 0 年段；值一律照抄引擎行，不在这里重判（BUG-1812）。
+            for key in ("cycle_index", "zero_duration"):
+                if item.get(key) is not None:
+                    period[key] = item[key]
            nar[level] = period
+    # 占星师 2026-11-08 验收清单 §7 反例「小运未决但大运已定」：小运未决必须被点名，
+    # 不能只表现为「没有 ad」（「没算」与「未决」分开；BUG-1812）。
+    for key in ("ad_status", "ad_reason"):
+        if nar_current.get(key) is not None:
+            nar[key] = nar_current[key]
    # 产品 2026-11-07: the next Narayana AD with its dates (Rath equal ADs, narayana_rath.antardasha).
    md_item = _record(nar_current.get("md"))
    ad_item = _record(nar_current.get("ad"))
@@ -85,7 +85,14 @@ def age_to_date(birth_dt: Optional[datetime], age: Any) -> Optional[str]:

 def _period(row: Dict[str, Any], birth_year: Optional[int],
            birth_dt: Optional[datetime] = None) -> Dict[str, Any]:
-    out = {key: row.get(key) for key in ("sign", "lord", "years", "base_years", "dignity_adjustment", "adjustment_source")}
+    out = {key: row.get(key) for key in ("sign", "lord", "years", "base_years", "dignity_adjustment",
+                                          "adjustment_source", "cycle_index", "zero_duration")}
+    # 占星师 2026-11-08 验收清单 §7 反例：卡上要能说出「第几轮大运」、「哪一行是 0 年段」。
+    # 只在这两项有值时带上（小运行没有轮次，非 0 年段不重复写 false），其余字段不变（BUG-1812）。
+    if out.get("cycle_index") is None:
+        out.pop("cycle_index", None)
+    if not out.get("zero_duration"):
+        out.pop("zero_duration", None)
    # Display ages stay at 0.01 year; dates come from the unrounded ages
    # (原值：咨询层用取整后的年龄换日期，最多偏 1.8 天；三方对照 2026-11-07：产品侧实现缺陷)。
    out["start_age"] = round(float(row["start_age"]), 2)
@@ -341,9 +348,17 @@ def rath_narayana_view(lagna_sign_idx: int, planet_lons: Dict[str, float], *,
    if md is not None and ad is not None:
        current: Dict[str, Any] = {"md": _period(md, birth_year, birth_datetime)}
        rows = ad.get("rows") or []
-        running_ad = _running(rows, current_age)
-        if running_ad is not None:
-            current["ad"] = _period(running_ad, birth_year, birth_datetime)
+        if ad.get("status") == "selected":
+            current["ad_status"] = "selected"
+            running_ad = _running(rows, current_age)
+            if running_ad is not None:
+                current["ad"] = _period(running_ad, birth_year, birth_datetime)
+        else:
+            # 原值：小运未决时整段 current_dasha 不给 ad，卡上「没有 ad」与「没算」分不开。
+            # 新值：大运照常给，另带 ad_status / ad_reason，由卡上点名「子运未决」。
+            # 原因：占星师 2026-11-08 验收清单 §7 反例「小运未决但大运已定」：未决不得写成未计算（BUG-1812）。
+            current["ad_status"] = "blocked"
+            current["ad_reason"] = ad.get("reason") or "antardasha_not_selected"
        if current_age is not None:
            current["remaining_years"] = round(float(md["end_age"]) - float(current_age), 2)
        view["current_dasha"] = current
@@ -0,0 +1,334 @@
+#!/usr/bin/env python3
+"""Capture the seven counterexamples of 占星师 2026-11-08 验收清单 §7 as engine output.
+
+One fixture per counterexample, written under
+``frontend/tests/fixtures/counterexamples-20261109/<case>.json``. Every value in a
+fixture comes from the real engine function the card's producer calls; nothing is
+hand-written here. ``tests/test_counterexamples_engine_20261109.py`` re-runs the
+same functions and compares its output with the file (a stale fixture fails), and
+``frontend/tests/consult-counterexamples-card-20261109.test.ts`` injects the same
+slices into a real golden workflow and reads the evidence card.
+
+Only the public AA chart Barack Obama and synthetic inputs are used.
+
+    .venv\\Scripts\\python scripts/research/capture_counterexample_fixtures_20261109.py
+"""
+
+from __future__ import annotations
+
+import copy
+import json
+import sys
+from datetime import datetime, timedelta
+from pathlib import Path
+from typing import Any
+
+ROOT = Path(__file__).resolve().parents[2]
+sys.path[:0] = [str(ROOT), str(ROOT / "scripts"), str(ROOT / "scripts" / "research")]
+
+import swisseph as swe  # noqa: E402
+
+from scripts import jyotish_engine as engine  # noqa: E402
+from scripts import narayana_legacy_label as label  # noqa: E402
+from scripts import narayana_rath as nr  # noqa: E402
+from scripts.chart_dasha import birth_moment  # noqa: E402
+from scripts.consultation_native_layers import build_career_analysis  # noqa: E402
+from scripts.functional_benefics import derive_functional_benefic_malefic, moon_paksha_from_planets  # noqa: E402
+from scripts.jyotish_api_server import _consultation_dasha_sub_periods  # noqa: E402
+from scripts.jyotish_engine import compute_chart_data  # noqa: E402
+from scripts.main_yoga_detectors import neecha_bhanga_rows  # noqa: E402
+from scripts.narayana_rath import SIGNS  # noqa: E402
+
+OUT = ROOT / "frontend" / "tests" / "fixtures" / "counterexamples-20261109"
+REFERENCE_DATE = "2026-11-09"
+GRAHAS = ("Sun", "Moon", "Mars", "Mercury", "Jupiter", "Venus", "Saturn", "Rahu", "Ketu")
+AYANAMSA = "raman"
+# Public AA chart: Barack Obama, 1961-08-04 19:24 Honolulu (same input as tests/test_round6_r1_r2_narayana.py).
+OBAMA = (1961, 8, 4, 19, 24, 21.3069, -157.8583, -10.0)
+# Synthetic birthplace for the young birth-balance chart (tests/neutral_subject.py coordinates).
+SYNTHETIC_PLACE = (39.9, 116.4, 8.0)
+
+
+def _chart(birth: tuple) -> tuple[dict, int, datetime]:
+    y, mo, d, h, mi, lat, lon, tz = birth
+    chart, asc, *_ = compute_chart_data(y, mo, d, h, mi, lat, lon, tz,
+                                        node_mode="mean", ayanamsa_name=AYANAMSA)
+    return chart, int(asc), datetime(y, mo, d, h, mi)
+
+
+def _lons(chart: dict) -> dict[str, float]:
+    return {name: float(chart["planets"][name]["degree_raw"]) for name in GRAHAS}
+
+
+def _career_layer(chart: dict, base: dict) -> dict:
+    """`build_career_analysis`'s functional layer — the second place the paksha is decided."""
+    analysis = build_career_analysis(engine, chart, {"functional_benefic_malefic": base}, {})
+    return analysis["functional_benefic_malefic"]
+
+
+# ── (a) 缺月相 ────────────────────────────────────────────────────────────────
+
+def case_missing_moon_paksha() -> dict:
+    chart, asc, _birth = _chart(OBAMA)
+    sign = SIGNS[asc]
+    with_lon = derive_functional_benefic_malefic(sign, moon_paksha=moon_paksha_from_planets(chart["planets"]))
+    bare = copy.deepcopy(chart)
+    # 合成输入：月亮保留星座与宫内度数（尊贵判定照常），但没有任何绝对经度，
+    # 于是 moon_paksha_from_planets 判不出盈亏 —— 正是「缺月相」那一刻的输入。
+    for key in ("lon", "degree_raw", "degree"):
+        (bare.get("planets", {}).get("Moon") or {}).pop(key, None)
+    without_lon = derive_functional_benefic_malefic(sign, moon_paksha=moon_paksha_from_planets(bare["planets"]))
+    return {
+        "ascendant": sign,
+        "blocked": {"base": without_lon, "career": _career_layer(bare, without_lon)},
+        "paksha_known": {"base": with_lon, "career": _career_layer(chart, with_lon)},
+    }
+
+
+# ── (b) 低档强弱未决 ──────────────────────────────────────────────────────────
+
+# Aries / Libra carry the same highest occupant tier (both empty of dignities): the
+# seed comparison is `undecided` instead of a made-up winner (round-6 R2).
+UNDECIDED_LONS = {"Sun": 185.0, "Moon": 29.5, "Mars": 190.0, "Mercury": 335.0, "Jupiter": 65.0,
+                  "Venus": 10.0, "Saturn": 275.0, "Rahu": 85.0, "Ketu": 265.0}
+
+
+def _shadbala_rows(birth: tuple, asc_sign: str) -> dict:
+    """`shadbala.calc_shadbala` + `attach_parameter_sensitivity` + `shadbala_display.row_fields`.
+
+    The same three calls the API makes for `chart.shadbala`; only the kept fields differ.
+    """
+    from shadbala import build_shadbala_context, calc_shadbala
+    from shadbala_display import row_fields
+    from shadbala_service.sensitivity import attach_parameter_sensitivity
+
+    y, mo, d, h, mi, lat, lon, tz = birth
+    chart, _asc, _birth = _chart(birth)
+    planets = copy.deepcopy(chart["planets"])
+    hour_decimal = float(h) + float(mi) / 60.0
+    jd_ut = swe.julday(y, mo, d, hour_decimal - float(tz))
+    result = calc_shadbala(
+        planets, asc_sign, hour_decimal, float(planets["Sun"]["degree_raw"]), float(planets["Moon"]["degree_raw"]),
+        context=build_shadbala_context(jd_ut, lat, lon, AYANAMSA, tz),
+    )
+    attach_parameter_sensitivity(result, planets=planets)
+    return {name: {"rupas": round(row["total_rupas"], 2), **row_fields(row)}
+            for name, row in result.get("planets", {}).items()}
+
+
+def case_low_grade_undecided() -> dict:
+    comparison = nr.compare_signs(0, 6, UNDECIDED_LONS, rule3_block=True)
+    calc = nr.calc_narayana_rath(0, UNDECIDED_LONS)
+    product = label.product_narayana(
+        None, 0, UNDECIDED_LONS, current_age=30.0, birth_year=2000, birth_datetime=datetime(2000, 1, 1, 12, 0))
+    return {
+        "narayana": {
+            "comparison": comparison,
+            "calc": {key: calc.get(key) for key in
+                     ("status", "blocked_step", "blocked_sign", "reason", "seed_sign", "seed_basis", "mahadasha")},
+            "product": product,
+        },
+        "shadbala": _shadbala_rows(OBAMA, SIGNS[_chart(OBAMA)[1]]),
+    }
+
+
+# ── (c) 小运未决、大运已定 ───────────────────────────────────────────────────
+
+def _saturn_ketu_in_running_md(lons: dict[str, float], lagna: int, age: float) -> dict[str, float]:
+    """Put Saturn and Ketu in the running MD sign (the engine's `saturn_ketu_priority_unresolved` trigger).
+
+    Moving them changes the strengths the sequence is built from, so the placement is
+    repeated until the running MD sign is the one both grahas sit in.
+    """
+    current = dict(lons)
+    for _attempt in range(12):
+        rows = nr.calc_narayana_rath(lagna, current)["mahadasha"]
+        running = nr._running_age(rows, age) if rows else None
+        if running is None:
+            raise RuntimeError("no running Narayana MD at the reference age")
+        idx = SIGNS.index(running["sign"])
+        if nr.nd._get_planet_sign("Saturn", current) == idx and nr.nd._get_planet_sign("Ketu", current) == idx:
+            return current
+        current = {**current, "Saturn": idx * 30.0 + 10.0, "Ketu": idx * 30.0 + 20.0,
+                   "Rahu": (idx * 30.0 + 200.0) % 360.0}
+    raise RuntimeError("Saturn / Ketu did not settle in the running MD sign")
+
+
+def case_ad_blocked_md_known() -> dict:
+    chart, lagna, birth = _chart(OBAMA)
+    age = label.narayana_age(birth, datetime.fromisoformat(REFERENCE_DATE))
+    lons = _saturn_ketu_in_running_md(_lons(chart), lagna, age)
+    rows = nr.calc_narayana_rath(lagna, lons)["mahadasha"]
+    md_row = nr._running_age(rows, age)
+    ad = nr.antardasha(md_row, lons)
+    product = label.product_narayana(None, lagna, lons, current_age=age, birth_year=birth.year,
+                                     birth_datetime=birth)
+    # 卡上的 dasha_systems 层由 build_dasha_systems 产出；这里用同一张（合成的）盘跑它，
+    # 拿到真实的 narayana 行，而不是在 TS 里手拼。
+    from scripts.consultation_native_layers import build_dasha_systems
+
+    system_chart = {
+        "ascendant": {"sign": SIGNS[lagna], "lon": lagna * 30.0 + float(chart["ascendant"]["degree_in_sign"])},
+        "planets": {name: {"lon": value} for name, value in lons.items()},
+    }
+    layers = build_dasha_systems(
+        system_chart, {"narayana_dasha": product, "dasha_sub_periods": {}},
+        birth_payload={"year": birth.year, "month": birth.month, "day": birth.day,
+                       "hour": birth.hour, "minute": birth.minute, "tz": -10.0},
+        reference=datetime.fromisoformat(REFERENCE_DATE),
+    )
+    return {
+        "age": round(age, 6),
+        "lons": lons,
+        "md": md_row,
+        "antardasha": {key: ad.get(key) for key in ("status", "reason", "comparison_status", "rows")},
+        "narayana": product,
+        "dasha_systems_narayana": next(row for row in layers["rows"] if row.get("system") == "narayana"),
+    }
+
+
+# ── (d) 月亮自参照 ───────────────────────────────────────────────────────────
+
+def case_moon_self_reference() -> dict:
+    from scripts.consultation_native_layers import build_yoga_conditions
+    from scripts.jyotish_api_server import _compact_yoga_entry
+
+    planets = {"Jupiter": {"sign": "Capricorn", "house": 6}, "Moon": {"sign": "Aries", "house": 9}}
+    compact = _compact_yoga_entry(neecha_bhanga_rows(planets)[0])
+    modules = {"raman_support": {"yogas": {"status": "executed", "hits": [], "ungoverned_detected": [compact]}}}
+    layer = build_yoga_conditions({"planets": planets}, modules, {})
+    row = next(item for item in layer["rows"] if item["name"] == "Neecha Bhanga")
+    return {"row": row, "yoga_conditions_status": layer.get("status")}
+
+
+# ── (e) 出生余额 ────────────────────────────────────────────────────────────
+
+def _young_chart() -> dict:
+    """A synthetic chart still inside its birth (balance) mahadasha at the reference date."""
+    lat, lon, tz = SYNTHETIC_PLACE
+    for day_offset in range(0, 60):
+        day = datetime(2020, 5, 1) + timedelta(days=day_offset)
+        chart, asc, birth = _chart((day.year, day.month, day.day, 10, 30, lat, lon, tz))
+        moon_lon = float(chart["planets"]["Moon"]["degree_raw"])
+        _timeline, elapsed, remaining, start_lord = birth_moment.mahadashas(birth, moon_lon)
+        if remaining > 9.0:
+            return {"chart": chart, "asc": asc, "birth": birth, "moon_lon": moon_lon,
+                    "periods": birth_moment.compute_vimshottari_timeline(
+                        birth_dt=birth, moon_lon=moon_lon,
+                        current_date=datetime.fromisoformat(REFERENCE_DATE))["periods"],
+                    "balance": {"lord": start_lord, "elapsed_years": elapsed, "remaining_years": remaining}}
+    raise RuntimeError("no synthetic birth kept a long-enough balance mahadasha")
+
+
+def case_birth_balance() -> dict:
+    young = _young_chart()
+    payload = {
+        "birth": {"date": young["birth"].strftime("%Y-%m-%d"), "hour": young["birth"].hour,
+                  "minute": young["birth"].minute, "second": 0},
+        "dasha": {"periods": young["periods"]},
+        "planets": {"Moon": {"lon": young["moon_lon"]}},
+    }
+    sub_periods = _consultation_dasha_sub_periods(payload, {"reference_date": REFERENCE_DATE})
+    return {
+        "birth": payload["birth"],
+        "moon_lon": young["moon_lon"],
+        "balance": young["balance"],
+        "dasha_periods": young["periods"],
+        "sub_periods": sub_periods,
+    }
+
+
+# ── (f) 第二轮大运 ──────────────────────────────────────────────────────────
+
+SECOND_CYCLE_AGE = 38.0
+
+
+def case_second_cycle() -> dict:
+    chart, lagna, birth = _chart(OBAMA)
+    lons = _lons(chart)
+    rows = nr.calc_narayana_rath(lagna, lons)["mahadasha"]
+    running = nr._running_age(rows, SECOND_CYCLE_AGE)
+    product = label.product_narayana(None, lagna, lons, current_age=SECOND_CYCLE_AGE,
+                                     birth_year=birth.year, birth_datetime=birth)
+    moon_lon = lons["Moon"]
+    through = birth + timedelta(days=125 * 365.25)
+    timeline, *_ = birth_moment.mahadashas(birth, moon_lon, through=through)
+    current_vim = next(row for row in timeline if row["start"] <= through < row["end"])
+    return {
+        "age": SECOND_CYCLE_AGE,
+        "narayana": product,
+        "running_md": running,
+        "vimshottari": {
+            "through": through.strftime("%Y-%m-%d"),
+            "current": {"lord": current_vim["lord"], "cycle": current_vim.get("cycle"),
+                        "start": current_vim["start"].strftime("%Y-%m-%d"),
+                        "end": current_vim["end"].strftime("%Y-%m-%d"), "years": current_vim["years"]},
+            "first_cycle_years": round(sum(row["years"] for row in timeline[:9]), 2),
+        },
+    }
+
+
+# ── (g) 0 年大运 ─────────────────────────────────────────────────────────────
+
+def case_zero_duration() -> dict:
+    chart, lagna, birth = _chart(OBAMA)
+    lons = _lons(chart)
+    rows = nr.calc_narayana_rath(lagna, lons)["mahadasha"]
+    zero = next(row for row in rows if row["zero_duration"] and row["cycle_index"] == 1)
+    age = float(zero["start_age"])
+    product = label.product_narayana(None, lagna, lons, current_age=age, birth_year=birth.year,
+                                     birth_datetime=birth)
+    return {
+        "age": age,
+        "zero_row": zero,
+        "zero_rows": [{key: row.get(key) for key in ("sign", "start_age", "end_age", "years", "cycle_index")}
+                      for row in rows if row["zero_duration"]],
+        "narayana": product,
+        "narayana_rows": [{key: row.get(key) for key in
+                           ("sign", "start_age", "end_age", "years", "cycle_index", "zero_duration")}
+                          for row in rows],
+    }
+
+
+CASES = {
+    "a_missing_moon_paksha": case_missing_moon_paksha,
+    "b_low_grade_undecided": case_low_grade_undecided,
+    "c_ad_blocked_md_known": case_ad_blocked_md_known,
+    "d_moon_self_reference": case_moon_self_reference,
+    "e_birth_balance": case_birth_balance,
+    "f_second_cycle": case_second_cycle,
+    "g_zero_duration": case_zero_duration,
+}
+
+
+def build_cases() -> dict[str, dict]:
+    """Every case's engine output, exactly as the fixtures store it."""
+    out: dict[str, dict] = {}
+    for name, builder in CASES.items():
+        out[name] = {
+            "case": name,
+            "source": "scripts/research/capture_counterexample_fixtures_20261109.py",
+            "reference_date": REFERENCE_DATE,
+            "ayanamsa": AYANAMSA,
+            "engine": builder(),
+        }
+    return out
+
+
+def normalise(value: Any) -> Any:
+    """JSON round-trip so the file and a fresh run compare equal."""
+    return json.loads(json.dumps(value, ensure_ascii=False, sort_keys=True, default=str))
+
+
+def main() -> int:
+    OUT.mkdir(parents=True, exist_ok=True)
+    for name, payload in build_cases().items():
+        target = OUT / f"{name}.json"
+        target.write_text(json.dumps(normalise(payload), ensure_ascii=False, sort_keys=True, indent=1) + "\n",
+                          encoding="utf-8")
+        print(f"wrote {target.relative_to(ROOT)} ({target.stat().st_size} bytes)")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
@@ -0,0 +1,217 @@
+"""占星师 2026-11-08 验收清单 §7 的七个反例：引擎侧端到端断言（BUG-1812）。
+
+每个反例一条测试，调用真实的引擎函数（不是包装脚本），断言状态、日期与「不计分 / 不增票」。
+配套的数据卡断言在 `frontend/tests/consult-counterexamples-card-20261109.test.ts`，读同一份
+fixture。fixture 由 `scripts/research/capture_counterexample_fixtures_20261109.py` 生成，
+最后一条测试断言「引擎现在的输出 == fixture」，fixture 过期即失败。
+
+只用公开名人盘（Barack Obama）或合成输入。
+"""
+from __future__ import annotations
+
+import json
+import sys
+from datetime import datetime
+from pathlib import Path
+
+import pytest
+
+ROOT = Path(__file__).resolve().parents[1]
+for entry in (str(ROOT), str(ROOT / "scripts"), str(ROOT / "scripts" / "research")):
+    if entry not in sys.path:
+        sys.path.insert(0, entry)
+
+from capture_counterexample_fixtures_20261109 import (  # noqa: E402
+    CASES,
+    OUT,
+    REFERENCE_DATE,
+    SECOND_CYCLE_AGE,
+    build_cases,
+    normalise,
+)
+
+FIXTURES = {name: json.loads((OUT / f"{name}.json").read_text(encoding="utf-8")) for name in CASES}
+
+
+def engine(case: str) -> dict:
+    return FIXTURES[case]["engine"]
+
+
+def _moon(layer: dict) -> dict:
+    return layer["movable_lagna_moon"]
+
+
+def test_fixture_is_fresh_for_every_counterexample() -> None:
+    """一条覆盖七张 fixture：引擎输出必须与仓内文件逐字段相同。"""
+    fresh = normalise(build_cases())
+    for name in CASES:
+        assert fresh[name] == FIXTURES[name], (
+            f"{name}.json 与引擎当前输出不一致；重跑 "
+            "scripts/research/capture_counterexample_fixtures_20261109.py")
+
+
+def test_a_missing_moon_paksha_blocks_only_the_moon_and_both_layers_agree() -> None:
+    """(a) 缺月相：只有月亮 blocked，原因点名；基础层与事业层同源。"""
+    data = engine("a_missing_moon_paksha")
+    base, career = data["blocked"]["base"], data["blocked"]["career"]
+    for layer in (base, career):
+        moon = _moon(layer)
+        assert (moon["status"], moon["role"], moon["reason"]) == ("blocked", None, "moon_paksha_missing")
+        assert layer["status"] == "used"
+    # 只阻塞月亮：其他行星照常分类，月亮不进任何一档。
+    assert base["blocked_planets"] == ["Moon"]
+    classified = set(base["functional_benefics"]) | set(base["functional_malefics"]) | set(base["functional_neutrals"])
+    assert "Moon" not in classified
+    # 表 30 只列七颗星（Rahu / Ketu 不在功能吉凶表里），月亮之外六颗照常分类。
+    assert len(classified) == 6, sorted(classified)
+    # 盈亏已知时两层一致（修复前事业层是 blocked / moon_paksha_missing）。
+    known = data["paksha_known"]
+    assert _moon(known["base"])["status"] == _moon(known["career"])["status"] == "derived_from_table_30_note"
+    assert _moon(known["base"])["role"] == _moon(known["career"])["role"] == "neutral"
+    assert _moon(known["base"])["moon_paksha"] == _moon(known["career"])["moon_paksha"] == "waning"
+
+
+def test_b_low_grade_comparison_undecided_names_the_step_and_scores_nothing() -> None:
+    """(b) 低档强弱未决：Narayana 点名未定步骤、不计分；Shadbala 保留 parameter_sensitive。"""
+    data = engine("b_low_grade_undecided")
+    comparison = data["narayana"]["comparison"]
+    assert (comparison["status"], comparison["comparison_status"]) == ("undecided", "unresolved")
+    assert comparison["blocked_level"] == 3
+    assert comparison["reason"] == "rule3_highest_tier_below_own_sign_tied"
+    calc = data["narayana"]["calc"]
+    assert (calc["status"], calc["blocked_step"]) == ("blocked", "seed")
+    product = data["narayana"]["product"]
+    assert product["status"] == "blocked"
+    assert product["used_for_scoring"] is False
+    assert "current_dasha" not in product  # 不用旧算法补出一个「已确认」的当前大运
+    method = product["method_status"]
+    assert (method["status"], method["chat_use"]) == ("blocked", "no_judgment")
+    assert method["unresolved_steps"] == ["seed"]
+    assert method["steps"]["seed"]["status"] == "blocked"
+    # Shadbala：翻盘的行星只能是 parameter_sensitive，不写成达标 / 不达标。
+    shadbala = data["shadbala"]
+    assert shadbala["Jupiter"]["minimum_check"] == "parameter_sensitive"
+    assert shadbala["Jupiter"]["minimum_open_questions"]
+    others = [name for name, row in shadbala.items() if name != "Jupiter"]
+    assert others and not any(row["minimum_check"] == "parameter_sensitive" for row in (shadbala[n] for n in others))
+
+
+def test_c_ad_undecided_keeps_the_md_and_names_the_reason() -> None:
+    """(c) 小运未决但大运已定：大运星座与日期保留，子运未决且原因被点名，不补旧 AD。"""
+    data = engine("c_ad_blocked_md_known")
+    md, ad = data["md"], data["antardasha"]
+    assert md["sign"] == "Taurus" and md["cycle_index"] == 2
+    assert (ad["status"], ad["reason"]) == ("blocked", "saturn_ketu_priority_unresolved")
+    assert ad["rows"] == []
+    product = data["narayana"]
+    current = product["current_dasha"]
+    assert current["md"]["sign"] == md["sign"]
+    assert current["md"]["cycle_index"] == md["cycle_index"] == 2
+    # 大运日期原样来自引擎行（未取整的 age 换成日期后与引擎一致）。
+    assert current["md"]["start_date"] and current["md"]["end_date"]
+    assert "ad" not in current  # 没有由旧算法补出的小运
+    assert current["ad_status"] == "blocked"
+    assert current["ad_reason"] == ad["reason"]
+    # 未决层不计分：Narayana 整体仍是单体系候选，used_for_scoring 为 False。
+    assert product["used_for_scoring"] is False
+    assert product["method_status"]["chat_use"] == "single_system_candidate"
+    # 卡上的 dasha_systems 层同样只带未决的原因，不带一个补出来的小运。
+    row = data["dasha_systems_narayana"]
+    assert (row["md"]["sign"], row["md"]["cycle_index"]) == (md["sign"], md["cycle_index"])
+    assert (row["ad_status"], row["ad_reason"]) == ("blocked", ad["reason"])
+    assert "ad" not in row
+
+
+def test_d_moon_self_reference_is_never_independent_confirmation() -> None:
+    """(d) 月亮自参照：只标自参照，独立确认为 0，不当作独立一票。"""
+    row = engine("d_moon_self_reference")["row"]
+    assert row["name"] == "Neecha Bhanga" and row["status"] == "computed"
+    evidence = row["formation_evidence"]
+    assert evidence["self_reference_only"] is True
+    assert evidence["independent_condition_count"] == 0
+    moon = next(item for item in evidence["cancellation_reasons"] if item.get("self_reference"))
+    assert (moon["reference"], moon["relative_house"], moon["self_reference"]) == ("Moon", 1, True)
+    assert moon["independent_confirmation_allowed"] is False
+    non_self = [item for item in evidence["cancellation_reasons"] if not item.get("self_reference")]
+    assert len(non_self) == evidence["independent_condition_count"] == 0
+
+
+def test_e_birth_balance_keeps_the_full_parent_start_beside_the_visible_one() -> None:
+    """(e) 出生余额：子运边界不因裁剪而改变，父段起点与可见起点都留、且不同。"""
+    data = engine("e_birth_balance")
+    balance = data["balance"]
+    assert balance["remaining_years"] > 9, balance
+    sub = data["sub_periods"]
+    assert sub["status"] == "ready"
+    md = sub["current"]["mahadasha"]
+    assert md["lord"] == balance["lord"]
+    # 余额大运在出生前开始：卡上看到的是被出生裁出来的那一段。
+    assert md["start"] < data["birth"]["date"] < md["end"]
+    # 出生那一刻的小运被裁：父段起点（出生前）与可见起点（出生当天）同时留下、且不同。
+    clipped = next(row for row in sub["boundaries"] if row.get("parent_period_start"))
+    assert clipped["parent_period_start"] < clipped["visible_period_start"] == data["birth"]["date"]
+    # 同一条小运的完整父段边界与未裁剪版本逐日相同（边界不移动）。
+    from chart_dasha import birth_moment
+
+    full = {row["lord"]: row for row in birth_moment.antardashas({
+        "lord": md["lord"], "start": datetime.fromisoformat(md["start"]), "end": datetime.fromisoformat(md["end"])})}
+    assert full[clipped["lord"]]["start"].strftime("%Y-%m-%d") == clipped["parent_period_start"]
+    for row in sub["boundaries"]:
+        assert full[row["lord"]]["end"].strftime("%Y-%m-%d") == row["end"], row["lord"]
+    # 参考日所在的小运仍是引擎值，不是按出生比例重算的。
+    assert sub["current"]["antardasha"]["lord"] == next(
+        row["lord"] for row in full.values() if row["start"] <= datetime(2026, 11, 9) < row["end"])
+
+
+def test_f_second_cycle_is_named_and_the_running_row_is_unique() -> None:
+    """(f) 第二轮大运：当前段带 cycle_index 2、年数等于引擎值；Vimshottari 跨 120 年后同样进第二轮。"""
+    data = engine("f_second_cycle")
+    md = data["narayana"]["current_dasha"]["md"]
+    running = data["running_md"]
+    assert md["sign"] == running["sign"] == "Aquarius"
+    assert md["cycle_index"] == running["cycle_index"] == 2
+    assert md["years"] == running["years"] == 11
+    # 24 行里只有一行覆盖参考年龄；没有把第二轮当成「还没算」。
+    rows = data["narayana"]["mahadasha"]
+    live = [row for row in rows if row["start_age"] <= SECOND_CYCLE_AGE < row["end_age"]]
+    assert len(live) == 1 and live[0]["sign"] == md["sign"] and live[0]["cycle_index"] == 2
+    vim = data["vimshottari"]
+    assert vim["first_cycle_years"] == 120
+    assert vim["current"]["cycle"] == 2 and vim["current"]["lord"] == "Moon"
+    assert vim["current"]["years"] == 10
+
+
+def test_g_zero_year_period_is_kept_in_the_data_and_never_current() -> None:
+    """(g) 0 年大运：段本身留在数据里并带 zero_duration，当前大运绝不是 0 年段。"""
+    data = engine("g_zero_duration")
+    zero = data["zero_row"]
+    assert (zero["sign"], zero["years"], zero["cycle_index"]) == ("Sagittarius", 0, 1)
+    assert zero["zero_duration"] is True and zero["start_age"] == zero["end_age"]
+    # 半开区间 [start, end)：0 年段占不到任何年龄，参考日落在它的起点也不是它。
+    product = data["narayana"]
+    assert product["current_dasha"]["md"]["sign"] == "Aquarius"
+    assert product["current_dasha"]["md"]["years"] == 11
+    assert product["current_dasha"]["md"]["start_age"] == data["age"]
+    signs = [row["sign"] for row in product["mahadasha"] if row["sign"] == "Sagittarius"]
+    assert signs, "0 年段必须留在 mahadasha 数据里"
+    kept = next(row for row in product["mahadasha"] if row["sign"] == "Sagittarius")
+    assert kept["years"] == 0 and kept["zero_duration"] is True
+
+
+def test_fixtures_cover_all_seven_counterexamples() -> None:
+    assert sorted(CASES) == [
+        "a_missing_moon_paksha", "b_low_grade_undecided", "c_ad_blocked_md_known",
+        "d_moon_self_reference", "e_birth_balance", "f_second_cycle", "g_zero_duration",
+    ]
+    for name in CASES:
+        payload = FIXTURES[name]
+        assert payload["reference_date"] == REFERENCE_DATE
+        assert payload["engine"]
+
+
+@pytest.mark.parametrize("case", sorted(CASES))
+def test_no_real_person_data_in_fixtures(case: str) -> None:
+    """公开名人盘只以公开的出生日期出现；不得有姓名、邮箱、会话或用户标识。"""
+    text = (OUT / f"{case}.json").read_text(encoding="utf-8")
+    for marker in ("@", "password", "jwt", "cookie", "case_id"):
+        assert marker not in text.lower(), (case, marker)
@@ -1,80 +0,0 @@
-"""BUG-1809: rectification test sets resealed on the round-7 integration (staging b453505d).
-
-The tag pins sensitivity measures, not accuracy (event scoring is near chance, BUG-1644).
-Older tags must stay byte-for-byte what they were; this file only adds a new tag.
-"""
-import json
-from pathlib import Path
-
-ROOT = Path(__file__).resolve().parents[1]
-RESEARCH = ROOT / "docs/research"
-TAG = "round7_integration_2026_11_09"
-OLD = "round5_pvr_narayana_2026_11_08"
-REASON = "产品负责人 2026-11-09 批准；第七轮整合（R1 默认 p50、R2 局部传播 seed_only、可比领先）"
-
-
-def _load(name: str) -> dict:
-    return json.loads((RESEARCH / name).read_text(encoding="utf-8"))
-
-
-def test_sealed_20_pins_the_new_tag() -> None:
-    # 原值：sealed_holdout_rerun_round5_pvr_narayana_2026_11_08 top-1 0.35、top-3 0.45、平均分钟误差 6.80
-    # 新值：sealed_holdout_rerun_round7_integration_2026_11_09 top-1 0.30、top-3 0.45、平均分钟误差 6.75
-    # 原因：产品负责人 2026-11-09 批准；第七轮整合（R1 默认 p50、R2 局部传播 seed_only、可比领先）
-    report = _load(f"sealed_holdout_rerun_{TAG}.json")
-    assert report["trial_count"] == 20 and report["implementation_hash_matches_at_replay"] is True
-    assert report["metrics"]["top_1_rate"] == 0.3
-    assert report["metrics"]["top_3_rate"] == 0.45
-    assert report["metrics"]["mean_absolute_minute_error"] == 6.75
-    assert report["metrics"]["false_confirmation_rate"] == 0.0
-    assert report["verified_minute_claim_allowed"] is False and report["results_previously_seen"] is True
-    assert report["status"] == "blocked_independent_blind_evidence"
-
-
-def test_reported_offset_900_pins_the_new_tag() -> None:
-    # 原值：reported_offset_round5_pvr_narayana_2026_11_08 top-1 0.0156、交付覆盖率 0.900
-    # 新值：reported_offset_round7_integration_2026_11_09 top-1 0.0100、交付覆盖率 0.883（164/900 次、7 例变）
-    # 原因：产品负责人 2026-11-09 批准；第七轮整合（R1 默认 p50、R2 局部传播 seed_only、可比领先）
-    report = _load(f"reported_offset_{TAG}.json")
-    trials = report["trials"]
-    assert len(trials) == 900 and report["implementation_hash_matches_at_replay"] is True
-    assert report["freeze_record_path"] == f"docs/research/reported_offset_{TAG}.freeze.json"
-    assert round(sum(t["top_1_hit"] for t in trials) / 900, 4) == 0.01
-    assert round(sum(t["delivery_covers_truth"] for t in trials) / 900, 4) == 0.8833
-    old = {(t["case_ordinal"], t["offset_minutes"], t["radius_minutes"]): t for t in _load(f"reported_offset_{OLD}.json")["trials"]}
-    changed = [t for t in trials if old[(t["case_ordinal"], t["offset_minutes"], t["radius_minutes"])] != t]
-    assert len(changed) == 164 and len({t["case_ordinal"] for t in changed}) == 7
-
-
-def test_live_v5_and_comparability_are_recorded_in_the_tag() -> None:
-    # 原值：live_astrologer_v5_round5_pvr_narayana_2026_11_08 最佳 = 真值 8、真值在前三 19
-    # 新值：live_astrologer_v5_round7_integration_2026_11_09 最佳 = 真值 8、真值在前三 18；
-    #       可比性策略 common_evidence_coverage_v1，暂列 / 可比领先计数写进标签
-    # 原因：产品负责人 2026-11-09 批准；第七轮整合（R1 默认 p50、R2 局部传播 seed_only、可比领先）
-    live = _load(f"live_astrologer_v5_{TAG}.json")
-    assert live["tag"] == TAG and live["baseline_tag"] == OLD
-    assert len(live["trials"]) == 231 and live["summary"]["new"]["best_is_truth"] == 8
-    assert live["summary"]["changed"]["any"] == 124
-    assert live["summary"]["pre_integration_staging"]["new_vs_it"]["any"] == 115
-    assert live["narayana_strength_scope"].startswith("seed_only_v0")
-    assert live["comparability_policy"].startswith("common_evidence_coverage_v1")
-    assert live["leader_coverage_status"] == {"comparable": 226, "provisional_unequal_coverage": 5}
-    comparability = _load(f"{TAG}_comparability.json")
-    assert comparability["policy"] == "common_evidence_coverage_v1"
-    assert comparability["sealed_20"]["leader_status"] == {"comparable": 40}
-    assert comparability["reported_offset_900"]["leader_status"] == {
-        "comparable": 863, "provisional_unequal_coverage": 35, "no_single_leader": 2}
-    assert comparability["live_v5_231"]["leader_status"] == live["leader_coverage_status"]
-
-
-def test_older_tags_stay_intact() -> None:
-    sealed = _load(f"sealed_holdout_rerun_{OLD}.json")["metrics"]
-    assert (sealed["top_1_rate"], sealed["top_3_rate"], sealed["mean_absolute_minute_error"]) == (0.35, 0.45, 6.8)
-    offset = _load(f"reported_offset_{OLD}.json")["trials"]
-    assert round(sum(t["top_1_hit"] for t in offset) / 900, 4) == 0.0156
-    assert _load(f"live_astrologer_v5_{OLD}.json")["summary"]["new"]["best_is_truth"] == 8
-    for old_tag in ("float_split_2026_11_08", "astrologer_166d_2026_11_05", OLD):
-        for kind in ("sealed_holdout_rerun", "reported_offset"):
-            assert (RESEARCH / f"{kind}_{old_tag}.json").exists()
-            assert (RESEARCH / f"{kind}_{old_tag}.freeze.json").exists()
-    assert REASON in Path(__file__).read_text(encoding="utf-8")