fix: enforce existing mevg invocation across timing workflows

AGENTS.md

@@ -41,7 +41,30 @@
    - 关键功能凶星
    - 对结论置信度的影响
 
-## 3. Honesty Boundary
+## 3. Existing MEVG Invocation Hard Constraint
+
+**强制执行既有 MEVG 规则，不得把它当成可选增强项。**
+
+本节不是新增一套验证系统，而是把 `SKILL.md` 与
+`references/mandatory-verification-gate-protocol.md` 中已经存在的 MEVG
+外部验证门控提升为协作代理硬约束。
+
+执行要求：
+
+1. 对用户提出的 **所有星盘运势类问题**、**所有有关印度占星推运的问题**，包括命盘解读、事业、财富、婚恋、健康、流年、流月、应期、事件预测、出生时间校正辅助和技法可靠性判断，必须执行 MEVG。
+2. MEVG 必须包含：
+   - 全球 / 全网外部资料采集
+   - 真实案例参考
+   - 来源分级
+   - 冲突仲裁
+   - 未验证声明降级
+3. 输出的 Technique Audit Table 必须出现以下两行：
+   - `MEVG / Global Web Evidence`
+   - `Real Case Calibration`
+4. 若无法完成外部资料采集、无法找到真实案例、网络/工具不可用、或来源之间出现重大冲突，必须写成 `blocked` 或降级置信度，不得静默跳过。
+5. 只有 **纯计算 / 纯代码 / 纯项目维护** 可以豁免 MEVG，例如运行测试、检查 Git 状态、修复代码、输出未解释的原始度数或 Dasha 边界。一旦开始解释“这代表什么运势”，豁免立即失效。
+
+## 4. Honesty Boundary
 
 以下情况必须明确写成 `blocked` 或降级置信度：
 
@@ -50,5 +73,6 @@
 - 缺少分盘、Ayanamsa、Node mode 或出生精度
 - 功能性吉凶星层未完成
 - 双重大运或多系统结果发生实质冲突
+- MEVG / Global Web Evidence 或 Real Case Calibration 未完成
 
 禁止把内部一致性伪装成“已经全球顶级精度”。

references/event_judgment_career.md

@@ -40,6 +40,8 @@
 - `Transit / Double Transit`
 - `Shadbala`
 - `Functional Benefic/Malefic`
+- `MEVG / Global Web Evidence`
+- `Real Case Calibration`
 
 高严谨时尽量补：
 
@@ -130,6 +132,18 @@
 3. `main conflicts`
 4. `Technique Audit Table`
 5. `raw evidence`
+6. `MEVG / Global Web Evidence`
+7. `Real Case Calibration`
+
+`MEVG / Global Web Evidence` 必须复用
+`references/mandatory-verification-gate-protocol.md`，至少说明 source tier、
+global web evidence collection、conflict arbitration 和未验证声明如何降级。
+
+`Real Case Calibration` 必须给出真实案例参考、公开 benchmark case，或明确的
+case gap。没有可比案例时不得假装完成，应降低置信度。
+
+pure calculation exemption 只适用于纯计算、纯代码、纯项目维护或不解释运势意义的原始
+数据输出；一旦解释事业运势、职业窗口、项目落地或职位变化，必须执行 MEVG 与真实案例校正。
 
 示例 verdict：
 
@@ -155,3 +169,5 @@
 - 没交代 `AmK / Karakamsha`
 - 未说明 birth time precision
 - 未说明 `Ayanamsa / Node mode`
+- 未完成 `MEVG / Global Web Evidence`
+- 未完成 `Real Case Calibration`

references/event_judgment_marriage.md

@@ -38,6 +38,8 @@
 - `Transit / Double Transit`
 - `Vivah Saham`
 - `Functional Benefic/Malefic`
+- `MEVG / Global Web Evidence`
+- `Real Case Calibration`
 
 若用户要求高严谨，还应尽量加：
 
@@ -133,6 +135,18 @@
 3. `main conflicts`
 4. `Technique Audit Table`
 5. `raw evidence`
+6. `MEVG / Global Web Evidence`
+7. `Real Case Calibration`
+
+`MEVG / Global Web Evidence` 必须复用
+`references/mandatory-verification-gate-protocol.md`，至少说明 source tier、
+global web evidence collection、conflict arbitration 和未验证声明如何降级。
+
+`Real Case Calibration` 必须给出真实案例参考、公开 benchmark case，或明确的
+case gap。没有可比案例时不得假装完成，应降低置信度。
+
+pure calculation exemption 只适用于纯计算、纯代码、纯项目维护或不解释运势意义的原始
+数据输出；一旦解释婚恋运势、关系窗口或婚期，必须执行 MEVG 与真实案例校正。
 
 示例 verdict：
 
@@ -154,3 +168,5 @@
 - 只看单一文章规则
 - 未说明 birth time precision
 - 未说明 `Ayanamsa / Node mode`
+- 未完成 `MEVG / Global Web Evidence`
+- 未完成 `Real Case Calibration`

references/event_judgment_skeleton.md

@@ -66,6 +66,17 @@
 - `node_mode`
 - `maturity`
 
+所有涉及印度占星推运、运势解读、事件预测、流年流月、健康、搬迁、考试、
+出行、出生时间校正辅助或技法可靠性判断的问题，都必须复用既有
+`references/mandatory-verification-gate-protocol.md`：
+
+- `MEVG / Global Web Evidence`：记录全球 / 全网外部资料采集、source tier、
+  conflict arbitration 和未验证声明的降级。
+- `Real Case Calibration`：记录真实案例参考、公开 benchmark case，或明确
+  case gap。
+- pure calculation exemption 只适用于纯计算、纯代码、纯项目维护或不解释运势意义的
+  原始数据输出；一旦解释推运或运势意义，必须执行 MEVG 与真实案例校正。
+
 ---
 
 ## 3. Adjudication
@@ -124,6 +135,8 @@
 3. `conflicts`
 4. `Technique Audit Table`
 5. `raw evidence`
+6. `MEVG / Global Web Evidence`
+7. `Real Case Calibration`
 
 最小 JSON 形态：
 
@@ -151,3 +164,5 @@
 - 题目域分盘未展开（relationship -> D9/UL；career -> D10/A10；wealth -> D2/D11）
 - 外部 oracle 未闭环却假装全局封顶
 - 缺少 `Technique Audit Table`
+- 缺少 `MEVG / Global Web Evidence`
+- 缺少 `Real Case Calibration`

references/event_judgment_wealth.md

@@ -58,6 +58,8 @@
 - `Shadbala`
 - `Ashtakavarga`
 - `Functional Benefic/Malefic`
+- `MEVG / Global Web Evidence`
+- `Real Case Calibration`
 
 高严谨时尽量补：
 
@@ -172,6 +174,18 @@
 3. `main conflicts`
 4. `Technique Audit Table`
 5. `raw evidence`
+6. `MEVG / Global Web Evidence`
+7. `Real Case Calibration`
+
+`MEVG / Global Web Evidence` 必须复用
+`references/mandatory-verification-gate-protocol.md`，至少说明 source tier、
+global web evidence collection、conflict arbitration 和未验证声明如何降级。
+
+`Real Case Calibration` 必须给出真实案例参考、公开 benchmark case，或明确的
+case gap。没有可比案例时不得假装完成，应降低置信度。
+
+pure calculation exemption 只适用于纯计算、纯代码、纯项目维护或不解释运势意义的原始
+数据输出；一旦解释财富运势、收入窗口、到账、资产或变现，必须执行 MEVG 与真实案例校正。
 
 示例 verdict：
 
@@ -197,3 +211,5 @@
 - 把纸面估值误当成流动性到账
 - 未说明 birth time precision
 - 未说明 `Ayanamsa / Node mode`
+- 未完成 `MEVG / Global Web Evidence`
+- 未完成 `Real Case Calibration`

references/real-reading-quality-checklist.md

@@ -48,14 +48,22 @@
 23. 是否把外部 oracle 未闭环的前线（Dasha boundary / Shadbala absolute / Tajika annual）说清楚？
 24. 是否避免把内部一致性伪装成传统软件级真值冻结？
 
-## H. 输出层
+## H. MEVG 与真实案例层
 
-25. 是否输出 `Technique Audit Table` 或等价审计表？
-26. 是否给出 `used / partial / blocked / not used` 级别说明？
-27. 是否给出置信度，而不是只给断语？
-28. 是否把 must-not-overclaim 边界说清楚？
+25. 是否执行 `MEVG / Global Web Evidence`，即全球 / 全网外部资料采集？
+26. 是否给出 `Real Case Calibration`，即真实案例参考、公开 benchmark case，或明确 case gap？
+27. 是否标出 source tier，而不是把论坛、博客、经典文本、开源项目和黑盒 oracle 混成同一权重？
+28. 若来源冲突，是否做了 conflict arbitration，并说明哪条结论被降级？
+29. 是否正确使用 pure calculation exemption：只在纯计算、纯代码、纯项目维护或未解释运势意义的原始数据输出时豁免？
+
+## I. 输出层
+
+30. 是否输出 `Technique Audit Table` 或等价审计表？
+31. 是否给出 `used / partial / blocked / not used` 级别说明？
+32. 是否给出置信度，而不是只给断语？
+33. 是否把 must-not-overclaim 边界说清楚？
 
 ## 一句话用法
 
 只要这张表里有 3-4 项没过，就不要把输出包装成“高严谨结论”。  
-只要 `参数层 + 承诺层 + 分盘层 + timing 层 + 外部真值边界` 没一起过，就不能把 skill 说成“已经无疏漏封顶”。
+只要 `参数层 + 承诺层 + 分盘层 + timing 层 + 外部真值边界 + MEVG 与真实案例层` 没一起过，就不能把 skill 说成“已经无疏漏封顶”。

references/strict-workflow-router.md

@@ -31,6 +31,33 @@ If any of the above cannot be completed because of missing birth precision, exte
 
 ---
 
+## 0.6 Existing MEVG Invocation Gate
+
+This router reuses the existing MEVG source of truth in
+`references/mandatory-verification-gate-protocol.md`; it does not create a
+second verification system.
+
+For **all chart interpretation readings**, including natal interpretation,
+career, wealth, relationship, health, annual/monthly timing, event prediction,
+rectification support, and technique reliability questions, the reading must
+complete or explicitly block:
+
+1. `MEVG / Global Web Evidence`: global web evidence collection, source tiering,
+   and conflict arbitration.
+2. `Real Case Calibration`: real case reference, benchmark case reference, or a
+   clear case-gap statement with confidence downgrade.
+
+The only exemption is: pure calculation / code / project-maintenance tasks are exempt
+when they do not interpret chart meaning. Examples: running tests, checking Git
+status, fixing code, or returning raw degrees / Dasha boundaries without any
+fortune interpretation. Once the answer explains what the chart means, the
+exemption no longer applies.
+
+Every Level 2+ chart reading must include `MEVG / Global Web Evidence` and
+`Real Case Calibration` in the Technique Audit Table.
+
+---
+
 ## 1. Question router
 
 | User intent | Route | Required depth |
@@ -58,7 +85,8 @@ For every Level 2+ reading, complete or explicitly mark unavailable:
 - Transit: Saturn/Jupiter/Rahu-Ketu, plus Moon trigger for month/day-level timing.
 - Functional benefic/malefic status by Lagna.
 - Shadbala and Ashtakavarga when a claim depends on planetary strength or transit strength.
-- MEVG external verification for non-obvious interpretive claims.
+- MEVG external verification for all interpretive chart-reading claims.
+- Real case calibration or a clearly marked case-evidence gap.
 - Confidence label: A/B/C/insufficient.
 
 ---
@@ -207,6 +235,8 @@ Every Level 2+ output must include this table near the end.
 | Argala | Used / not used | ... | ... |
 | KP / Sub-lord | Used / not required / unavailable | ... | ... |
 | Historical verification | Used / not provided | ... | ... |
+| MEVG / Global Web Evidence | Used / blocked / exempt for pure calculation | Source tier, searches, conflict arbitration | Mandatory for all chart interpretation readings; blocked or downgraded if omitted |
+| Real Case Calibration | Used / blocked / case gap | Real case, benchmark case, or documented lack of matching cases | Cap confidence when no comparable case evidence is available |
 
 Never omit unavailable techniques silently. Mark them as unavailable or not integrated.
 When Functional Benefic/Malefic is used, explicitly surface the functional neutrals and yogakarakas instead of collapsing them into a generic summary.

references/transit-actionable-output-guide.md

@@ -57,6 +57,18 @@
 
 当动态预测（被发现/合作/破圈/关键事件型）时，**必须先执行案例检索**：
 
+本节是 `references/mandatory-verification-gate-protocol.md` 的推运执行入口之一。
+所有 Transit / Dasha / 年运 / 月运 / 具体事件型推运结论，都必须在最终
+Technique Audit Table 中出现：
+
+- `MEVG / Global Web Evidence`：记录全网外部资料采集、source tier 与
+  conflict arbitration。
+- `Real Case Calibration`：记录真实案例参考、公开 benchmark case，或明确
+  case gap。
+
+pure calculation exemption 只适用于纯计算、纯代码、纯项目维护或不解释运势意义的
+原始数据输出；一旦解释 Transit 或 Dasha 对运势的意义，必须执行 MEVG 与真实案例校正。
+
 ### Step 1：检索同类行星配置真实案例
 
 ```bash

tests/test_real_reading_quality_checklist.py

@@ -25,6 +25,12 @@ def test_real_reading_quality_checklist_exists_and_covers_structural_gaps() -> N
         "Vimshottari + Narayana",
         "production_tuning_allowed",
         "Technique Audit Table",
+        "MEVG / Global Web Evidence",
+        "Real Case Calibration",
+        "全球 / 全网外部资料采集",
+        "真实案例参考",
+        "source tier",
+        "conflict arbitration",
     ]:
         assert token in text
 

tests/test_skill_hard_constraints.py

@@ -40,3 +40,70 @@ def test_agents_declares_functional_benefic_malefic_hard_constraint() -> None:
     assert "Vimshottari + Narayana Dasha" in agents
     assert "Technique Audit Table" in agents
     assert "blocked" in agents
+
+
+def test_mevg_global_web_and_real_case_evidence_is_agent_hard_constraint() -> None:
+    agents = (ROOT / "AGENTS.md").read_text(encoding="utf-8")
+
+    assert "Existing MEVG Invocation Hard Constraint" in agents
+    assert "所有星盘运势类问题" in agents
+    assert "所有有关印度占星推运的问题" in agents
+    assert "references/mandatory-verification-gate-protocol.md" in agents
+    assert "MEVG" in agents
+    assert "全球 / 全网外部资料采集" in agents
+    assert "真实案例参考" in agents
+    assert "MEVG / Global Web Evidence" in agents
+    assert "Real Case Calibration" in agents
+    assert "纯计算 / 纯代码 / 纯项目维护" in agents
+
+
+def test_strict_router_makes_mevg_default_for_all_chart_readings() -> None:
+    router = (ROOT / "references" / "strict-workflow-router.md").read_text(encoding="utf-8")
+
+    assert "Existing MEVG Invocation Gate" in router
+    assert "all chart interpretation readings" in router
+    assert "annual/monthly timing" in router
+    assert "event prediction" in router
+    assert "rectification support" in router
+    assert "MEVG / Global Web Evidence" in router
+    assert "Real Case Calibration" in router
+    assert "pure calculation / code / project-maintenance tasks are exempt" in router
+    assert "MEVG external verification for all interpretive chart-reading claims" in router
+
+
+def test_domain_event_adjudicators_require_mevg_and_real_case_contract() -> None:
+    for filename in [
+        "event_judgment_marriage.md",
+        "event_judgment_career.md",
+        "event_judgment_wealth.md",
+    ]:
+        text = (ROOT / "references" / filename).read_text(encoding="utf-8")
+
+        assert "MEVG / Global Web Evidence" in text
+        assert "Real Case Calibration" in text
+        assert "source tier" in text
+        assert "conflict arbitration" in text
+        assert "pure calculation exemption" in text
+
+
+def test_general_timing_workflow_requires_mevg_and_real_case_contract() -> None:
+    skeleton = (ROOT / "references" / "event_judgment_skeleton.md").read_text(encoding="utf-8")
+    assert "marriage / career / wealth / health / generic event verification" in skeleton
+    assert "MEVG / Global Web Evidence" in skeleton
+    assert "Real Case Calibration" in skeleton
+    assert "source tier" in skeleton
+    assert "conflict arbitration" in skeleton
+    assert "pure calculation exemption" in skeleton
+
+    workflow = (ROOT / "references" / "ai-reading-workflow-prompt.md").read_text(encoding="utf-8")
+    assert "references/mandatory-verification-gate-protocol.md" in workflow
+    assert "Step 3.11" in workflow
+    assert "Step 4.10" in workflow
+    assert "Step 5.5" in workflow
+    assert "案例检索与对比" in workflow
+
+    transit = (ROOT / "references" / "transit-actionable-output-guide.md").read_text(encoding="utf-8")
+    assert "案例检索三步法" in transit
+    assert "任何涉及具体事件类型的 Transit 预测" in transit
+    assert "MEVG / Global Web Evidence" in transit
+    assert "Real Case Calibration" in transit