Prepare Round 03.2 selector review bundle

README.md

@@ -170,10 +170,12 @@ Indexes follow `docs/INDEX_MAINTENANCE_PROTOCOL.md`: every asset change must syn
 
 CCRA review packages are archived by round under `ccra_review_bundle/round-NN_YYYY-MM-DD_topic/`. Do not place new review bundle files directly in `ccra_review_bundle/`; create or update the current round directory instead.
 
+Files intended for GPT / CCRA upload inside a review bundle should include the round version in the filename, for example `_03.2`. Raw changed-file zips should also include the round version while preserving source-relative paths inside the archive.
+
 ## 13. Next Steps
 
-1. Submit the Round 03.1 selector/no-call/regression patch for GPT / CCRA review.
+1. Submit the Round 03.2 selector negative-gate patch for GPT / CCRA review.
 2. Keep QPI and Intellectual Archaeology at `draft / B / pending` until Owner / CCRA review accepts stronger status.
-3. Run a QPI blind-test round with new unlabelled inputs after Round 03.1 passes.
+3. Run a QPI blind-test round with new unlabelled inputs only after Round 03.2 passes.
 4. Defer any third-model expansion until the current QPI selector and case layer are accepted.
 5. Route missing reusable extraction, inspection, or stability-scoring tools to `requirements/skills-vault/` or `requirements/ccpe/` instead of improvising local platform features.

cards/intellectual_archaeology.md

@@ -175,7 +175,7 @@
 - 机制稳定性：中等，七层结构清晰，但层级之间的真实递进关系需要更多案例验证。
 - 边界清晰度：中等，最小充分下潜原则已明确，但执行时容易过度下钻。
 - 来源证据质量：较高，核心定义和七层结构来自原文，停止条件来自综合应用材料。
-- 回归测试表现：pending，已有五条样板用例，尚未扩展到更多领域。
+- 回归测试表现：pending，当前已有 19 条回归用例，已覆盖 positive、boundary、misuse、no_call、selector_gate 和 pipeline；Round 03.2 已纳入明确拒绝思想考古的模型级 hard exclusion 陷阱。
 
 评级理由：纵向下潜机制清晰，适合深度建模，但七层结构和停止条件需要进一步稳定。
 
@@ -185,13 +185,10 @@
 
 `pending`
 
-当前已有五条回归用例：
+当前已有 19 条回归用例，覆盖 positive、boundary、misuse、no_call、selector_gate 和 pipeline。Round 03.2 重点新增：
 
-- `case_ia_positive_deep_modeling_001`
+- `case_ia_negative_explicit_do_not_call_001`
-- `case_ia_positive_resilience_001`
+- `case_ia_negative_only_qpi_001`
-- `case_ia_boundary_minimum_depth_001`
-- `case_ia_misuse_philosophizing_001`
-- `case_ia_misuse_no_validation_001`
 
 ## 示例输入
 

cards/qpi.md

@@ -202,7 +202,7 @@ QPI 的三分结构、核心匮乏物、主体性和动态性有清晰来源支
 - 机制稳定性：中高，扫描匮乏物和匹配处理范式的机制明确。
 - 边界清晰度：中等，混合型问题、多视角分歧和上下文不足场景仍需更多测试。
 - 来源证据质量：较高，主体性和动态性来自 2025 原文，核心匮乏物和误框定规则来自 2026 原文，应用规则来自综合文档。
-- 回归测试表现：pending，当前已有 52 条 QPI 回归用例；其中 Round 03 / 03.1 已纳入 owner-reviewed case layer、selector no-call、multi-perspective 和低上下文 provisional 边界用例。生命周期仍保持 draft。
+- 回归测试表现：pending，当前已有 55 条 QPI 回归用例；其中 Round 03 / 03.1 已纳入 owner-reviewed case layer、selector no-call、multi-perspective 和低上下文 provisional 边界用例，Round 03.2 已纳入普通 how-to / 词义判断 no-call 陷阱。生命周期仍保持 draft。
 
 评级理由：三分结构清晰，适合作为入口路由模型，但需要补充大量边界案例，防止过度升维或降维。
 
@@ -212,7 +212,7 @@ QPI 的三分结构、核心匮乏物、主体性和动态性有清晰来源支
 
 `pending`
 
-当前已有 52 条 QPI 回归用例，覆盖 positive、boundary、misuse、no_call、selector_gate 和 pipeline。Round 03 / 03.1 重点新增：
+当前已有 55 条 QPI 回归用例，覆盖 positive、boundary、misuse、no_call、selector_gate 和 pipeline。Round 03 / 03.1 重点新增：
 
 - owner-reviewed flow / disappointment / organizational year-end-review cases；
 - `case_qpi_international_logistics_no_call_001`
@@ -222,6 +222,12 @@ QPI 的三分结构、核心匮乏物、主体性和动态性有清晰来源支
 - `case_qpi_direct_summary_no_call_001`
 - `case_qpi_analysis_override_should_call_001`
 
+Round 03.2 重点新增：
+
+- `case_qpi_how_to_excel_no_call_001`
+- `case_qpi_how_to_markdown_pdf_no_call_001`
+- `case_qpi_dictionary_judgment_no_call_001`
+
 ## 示例输入
 
 - “我们团队每次都说要长期主义，但一到季度 KPI 就回到短期动作，这到底是什么问题？”
@@ -280,7 +286,7 @@ QPI 的三分结构、核心匮乏物、主体性和动态性有清晰来源支
 
 QPI 应作为问题回答系统的前置路由模型，用于防止系统在问题类型错误的情况下直接给答案。
 
-selector 应优先使用 `trigger_keywords`、`negative_triggers`、`pipeline_position` 和 `selection_priority` 做规则推荐。
+selector 应优先使用 `trigger_keywords`、`negative_triggers`、模型级 hard exclusion、`pipeline_position` 和 `selection_priority` 做规则推荐。
 
 ## 版本信息
 

ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/00_OPEN_THIS_FIRST_CCRA_REVIEW_BRIEF_03.2.md

@@ -0,0 +1,55 @@
+# Round 03.2 CCRA Review Brief
+
+Date: 2026-06-17
+
+Repository: `C:\Users\wangq\Documents\Codex\work-projects\the-mindscape-of-bro-tsong`
+
+Patch title: Selector Negative-Gate Patch
+
+## Scope
+
+Round 03.2 is a tiny follow-up to Round 03.1.
+
+It addresses GPT review finding:
+
+> IA / 思想考古 negative triggers were still score penalties, not model-level hard exclusions.
+
+This patch does not:
+
+- add a third model;
+- upgrade QPI or Intellectual Archaeology lifecycle status;
+- introduce an LLM selector;
+- introduce RAG, database, frontend, backend, user accounts, or platform features;
+- start Round 04 blind testing.
+
+Current model status remains:
+
+```text
+qpi: draft / B / pending
+intellectual_archaeology: draft / B / pending
+```
+
+## What Changed
+
+1. Added model-level hard exclusion for explicit IA refusal.
+2. Removed bare QPI gate signals `如何` and `判断`.
+3. Replaced broad gates with compound problem-definition signals.
+4. Strengthened selector calibration smoke checks:
+   - `no_call` requires `selected_models == []` and `no_call == true`;
+   - `no_call_or_low_priority` rejects both QPI and IA;
+   - `select_qpi_reject_ia` requires QPI selected and IA rejected;
+   - IA selection expectations fail if IA hard exclusion is present.
+5. Added five GPT-requested traps to regression and calibration data.
+
+## Suggested Review Order
+
+1. `00_OPEN_THIS_FIRST_CCRA_REVIEW_BRIEF_03.2.md`
+2. `01_PATCH_MATRIX_03.2.md`
+3. `02_CURRENT_ASSET_PACK_03.2.md`
+4. `03_VALIDATION_AND_COMMAND_LOG_03.2.md`
+5. `04_REVIEW_QUESTIONS_FOR_GPT_03.2.md`
+6. `optional_raw_changed_files_03.2.zip` only if exact changed-file inspection is needed
+
+## Primary Review Question
+
+Does Round 03.2 close the model-level negative gate issue enough to close Round 03 and proceed to Round 04 blind input evaluation?

ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/01_PATCH_MATRIX_03.2.md

@@ -0,0 +1,53 @@
+# Round 03.2 Patch Matrix
+
+## GPT Requirement Mapping
+
+| Requirement | Implementation | Files |
+| --- | --- | --- |
+| IA explicit refusal must be hard exclusion, not score penalty | Added `hard_exclusion_triggers` for IA and forced rejected decision when matched | `selector/selector_rules.json`, `scripts/run_selector_demo.py` |
+| Include refusal phrases such as `不要思想考古`, `先不要思想考古`, `不调用思想考古`, `只做 QPI`, `只做问题定性` | Added IA hard exclusion trigger list | `selector/selector_rules.json` |
+| Preserve QPI explicit-analysis override but do not let it re-enable IA | Hard exclusion applies per model after scoring and is not bypassed by QPI override | `scripts/run_selector_demo.py` |
+| Remove or narrow bare `如何` and `判断` QPI gates | Removed bare signals; added compound problem-definition phrases | `selector/selector_rules.json` |
+| Strict `no_call` smoke behavior | `no_call` now requires no selected models and `no_call=true` | `scripts/run_selector_calibration_smoke.py` |
+| Strict `no_call_or_low_priority` behavior | Fails if QPI or IA is selected | `scripts/run_selector_calibration_smoke.py` |
+| Add `select_qpi_reject_ia` behavior | Requires QPI selected and IA rejected | `scripts/run_selector_calibration_smoke.py`, `selector/selector_calibration_inputs.json` |
+| Fail IA selection expectations if IA hard exclusion exists | Added hard-exclusion check for `select_intellectual_archaeology` expected cases | `scripts/run_selector_calibration_smoke.py` |
+| Add five traps | Added two IA hard-exclusion traps and three generic how-to / dictionary no-call traps | `tests/*.regression.json`, `selector/selector_calibration_inputs.json` |
+
+## Added Regression Cases
+
+### Intellectual Archaeology
+
+- `case_ia_negative_explicit_do_not_call_001`
+- `case_ia_negative_only_qpi_001`
+
+### QPI
+
+- `case_qpi_how_to_excel_no_call_001`
+- `case_qpi_how_to_markdown_pdf_no_call_001`
+- `case_qpi_dictionary_judgment_no_call_001`
+
+## Added Calibration Inputs
+
+- `selector_calibration_ia_negative_explicit_do_not_call_001`
+- `selector_calibration_ia_negative_only_qpi_001`
+- `selector_calibration_qpi_how_to_excel_no_call_001`
+- `selector_calibration_qpi_how_to_markdown_pdf_no_call_001`
+- `selector_calibration_qpi_dictionary_judgment_no_call_001`
+
+## Current Counts
+
+| Asset | Round 03.1 | Round 03.2 |
+| --- | ---: | ---: |
+| QPI regression cases | 52 | 55 |
+| IA regression cases | 17 | 19 |
+| Aggregate regression cases | 69 | 74 |
+| Selector calibration inputs | 85 | 90 |
+
+## Non-Goals Preserved
+
+- No new model.
+- No lifecycle upgrade.
+- No LLM selector.
+- No RAG / database / frontend / backend.
+- No Round 04 blind-test execution.

ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/02_CURRENT_ASSET_PACK_03.2.md

@@ -0,0 +1,73 @@
+# Round 03.2 Current Asset Pack
+
+## Canonical Changed Assets
+
+- `selector/selector_rules.json`
+- `selector/selector_calibration_inputs.json`
+- `tests/qpi.regression.json`
+- `tests/intellectual_archaeology.regression.json`
+- `tests/regression_cases.json`
+- `tests/regression_index.json`
+- `scripts/run_selector_demo.py`
+- `scripts/run_selector_calibration_smoke.py`
+
+## Generated / Derived Assets
+
+- `models/model_index.json`
+- `cards/card_index.md`
+- `reports/index_rebuild_report.md`
+- `reports/validation_report.md`
+- `reports/selector_regression_report_v0.2.md`
+- `reports/selector_calibration_smoke_report.md`
+- `reports/model_card_sync_report_v0.2.md`
+
+## Documentation Updated
+
+- `README.md`
+- `docs/DECISIONS.md`
+- `selector/README.md`
+- `scripts/README.md`
+- `reports/README.md`
+- `cards/qpi.md`
+- `cards/intellectual_archaeology.md`
+- `reports/content_review_report_v0.2.md`
+
+## Selector Behavior Summary
+
+### IA explicit refusal
+
+Inputs such as:
+
+```text
+QPI 已判断：这是 Issue。先不要思想考古，只做问题定性复核。
+不要思想考古，只做 QPI 复核。
+```
+
+now select QPI and reject `intellectual_archaeology`.
+
+### Generic how-to / dictionary no-call
+
+Inputs such as:
+
+```text
+如何在 Excel 里冻结首行？
+如何把 Markdown 转 PDF？
+判断这个词是什么意思。
+```
+
+now produce no selected models and `no_call=true`.
+
+### Still preserved
+
+QPI remains callable for explicit problem-definition or owner-reviewed calibration inputs.
+
+IA remains callable when IA is explicitly requested without hard exclusion and the heavy-depth / QPI-completed gate is satisfied.
+
+## Current Lifecycle State
+
+No lifecycle upgrade was made.
+
+```text
+qpi: draft / B / pending
+intellectual_archaeology: draft / B / pending
+```

ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/03_VALIDATION_AND_COMMAND_LOG_03.2.md

@@ -0,0 +1,61 @@
+# Round 03.2 Validation and Command Log
+
+Date: 2026-06-17
+
+## Red Evidence
+
+After adding the five GPT-requested traps and before implementation, selector checks failed as expected.
+
+`python scripts\run_selector_regression.py`
+
+- `case_ia_negative_explicit_do_not_call_001`: IA was selected despite explicit refusal.
+- `case_ia_negative_only_qpi_001`: QPI was not selected.
+- `case_qpi_how_to_excel_no_call_001`: QPI was selected.
+- `case_qpi_how_to_markdown_pdf_no_call_001`: QPI was selected.
+- `case_qpi_dictionary_judgment_no_call_001`: QPI was selected.
+
+`python scripts\run_selector_calibration_smoke.py`
+
+- `selector_calibration_ia_negative_explicit_do_not_call_001`: expected IA rejection.
+- `selector_calibration_ia_negative_only_qpi_001`: expected QPI selection.
+- three generic how-to / dictionary traps expected no QPI selection.
+
+## Green Verification
+
+The following commands passed after the Round 03.2 patch:
+
+```powershell
+python scripts\rebuild_indexes.py --check
+python -m unittest discover -s tests -p "test*.py" -v
+python scripts\validate_model_library.py
+python scripts\check_card_contract.py
+python scripts\run_selector_demo.py
+python scripts\run_selector_regression.py
+python scripts\run_selector_calibration_smoke.py
+python scripts\check_model_card_sync.py
+```
+
+Results:
+
+- Index check: PASS
+- Unit tests: PASS, 17 tests
+- Model library validation: PASS
+- Card contract: PASS
+- Selector demo: PASS
+- Selector regression: PASS, 18 selector-relevant cases
+- Selector calibration smoke: PASS, 90 inputs, 0 failures, 0 warnings
+- Model/card sync: PASS
+
+## Generated Reports
+
+- `reports/index_rebuild_report.md`
+- `reports/validation_report.md`
+- `reports/selector_regression_report_v0.2.md`
+- `reports/selector_calibration_smoke_report.md`
+- `reports/model_card_sync_report_v0.2.md`
+
+## Important Interpretation
+
+Validation pass does not mean model content is stable or owner-approved.
+
+It only means the Round 03.2 selector negative-gate patch is structurally consistent and passes current local checks.

ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/04_REVIEW_QUESTIONS_FOR_GPT_03.2.md

@@ -0,0 +1,29 @@
+# Round 03.2 Review Questions for GPT
+
+Please review Round 03.2 only.
+
+Do not expand into Round 04 blind testing yet.
+
+## Questions
+
+1. Does model-level hard exclusion now correctly handle explicit IA refusal?
+2. Are the current hard exclusion phrases sufficient for Round 03 closure?
+3. Is it acceptable that hard exclusion rejects only IA while still allowing QPI when the user says `只做 QPI` or `只做问题定性`?
+4. Is removing bare `如何` and `判断` while preserving compound problem-definition phrases enough to reduce QPI over-selection?
+5. Are the strengthened calibration smoke assertions strict enough to become a long-term selector gate?
+6. Can Round 03 now be closed with the following conclusion?
+
+```text
+QPI and Intellectual Archaeology remain draft / B / pending.
+QPI is accepted as a draft-callable routing asset at report level.
+Selector v0.2 passes no-call, calibration smoke, model-level exclusion, and QPI-before-IA gates.
+Proceed to Round 04 blind input evaluation.
+```
+
+## Suggested Answer Format
+
+Please answer with:
+
+- `pass`, `revise`, or `block`;
+- specific failing cases if `revise` or `block`;
+- whether Round 04 blind input evaluation may start after this patch.

ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/BUNDLE_FILE_MANIFEST_03.2.md

@@ -0,0 +1,52 @@
+# Round 03.2 Bundle File Manifest
+
+## Review Files
+
+- `00_OPEN_THIS_FIRST_CCRA_REVIEW_BRIEF_03.2.md`
+- `01_PATCH_MATRIX_03.2.md`
+- `02_CURRENT_ASSET_PACK_03.2.md`
+- `03_VALIDATION_AND_COMMAND_LOG_03.2.md`
+- `04_REVIEW_QUESTIONS_FOR_GPT_03.2.md`
+- `BUNDLE_FILE_MANIFEST_03.2.md`
+- `optional_raw_changed_files_03.2.zip`
+
+## Optional Raw Changed Files
+
+The zip is optional and should be used only if exact file inspection is needed.
+
+It preserves source-relative paths.
+
+It excludes:
+
+- `knowledge_assets/`
+- `__pycache__/`
+- `*.pyc`
+- previous review bundle zip files
+
+## Source-Relative Files Included in Raw Zip
+
+- `README.md`
+- `cards/intellectual_archaeology.md`
+- `cards/qpi.md`
+- `docs/DECISIONS.md`
+- `docs/FILE_TAXONOMY.md`
+- `reports/README.md`
+- `reports/content_review_report_v0.2.md`
+- `reports/Codex新会话交接文档_Round03.2评审准备_2026-06-17.md`
+- `reports/index_rebuild_report.md`
+- `reports/model_card_sync_report_v0.2.md`
+- `reports/selector_calibration_smoke_report.md`
+- `reports/selector_regression_report_v0.2.md`
+- `reports/validation_report.md`
+- `scripts/README.md`
+- `scripts/run_selector_calibration_smoke.py`
+- `scripts/run_selector_demo.py`
+- `selector/README.md`
+- `selector/selector_calibration_inputs.json`
+- `selector/selector_rules.json`
+- `tests/intellectual_archaeology.regression.json`
+- `tests/qpi.regression.json`
+- `tests/regression_cases.json`
+- `tests/regression_index.json`
+- `models/model_index.json`
+- `cards/card_index.md`

docs/DECISIONS.md

@@ -156,5 +156,28 @@ This keeps second-round and third-round materials comparable without overwriting
 
 Round 03.1 added two bundle hygiene rules:
 
-- `optional_raw_changed_files.zip` must preserve source-relative paths. Do not create flat archives that can collide on names such as `README.md`.
+- `optional_raw_changed_files_<round-version>.zip` must preserve source-relative paths. Do not create flat archives that can collide on names such as `README.md`.
 - `knowledge_assets/` is excluded from review bundles by default because the project owner manually syncs stable long-term knowledge assets into GPT knowledge storage. Include it only when the current review explicitly targets the knowledge asset itself.
+
+Round 03.2 added one ChatGPT upload hygiene rule:
+
+- Every file intended for GPT / CCRA upload inside a review bundle must include the round version in the filename, such as `_03.2` or `03.2`, because uploading same-named files from multiple rounds in one Web ChatGPT conversation can cause file-reading ambiguity. This applies especially to raw changed-file zip archives, e.g. `optional_raw_changed_files_03.2.zip`.
+- The version suffix belongs to the bundle file names. Zip entries should still preserve source-relative repository paths so reviewers can inspect actual changed files without flattening path context.
+
+## Decision 014: Treat explicit model refusal as model-level hard exclusion
+
+Status: Accepted
+
+Reason:
+
+Round 03.1 fixed QPI default over-selection, but GPT review found that explicit Intellectual Archaeology refusal phrases such as `先不要思想考古` could still lose to positive IA signals and QPI-completed bonuses.
+
+A user instruction that names a model and refuses it should not be represented as a normal score penalty. It must hard-exclude that model unless a later owner-approved rule defines an explicit reversal phrase.
+
+The current Round 03.2 implementation keeps this scoped:
+
+- It rejects only the explicitly refused model.
+- It does not create a global no-call.
+- It does not add an LLM selector, RAG, database, service, frontend, backend, or third model.
+- It preserves QPI explicit-analysis override, but QPI selection does not re-enable IA when IA is explicitly refused.
+- Generic bare QPI gate words such as `如何` and `判断` are removed in favor of compound problem-definition phrases.

docs/FILE_TAXONOMY.md

@@ -97,7 +97,8 @@ Rules:
 
 - Keep review bundles under dated round directories.
 - Do not place new review files directly under `ccra_review_bundle/`.
-- `optional_raw_changed_files.zip` must preserve source-relative paths.
+- `optional_raw_changed_files_<round-version>.zip` must preserve source-relative paths.
+- GPT / CCRA upload files inside review bundles must include the round version in the filename, for example `_03.2`, to avoid same-name ambiguity when multiple rounds are uploaded in one Web ChatGPT conversation.
 - Do not flatten zip contents, because duplicate filenames such as `README.md` can collide.
 - `knowledge_assets/` is excluded from review zips by default because the Owner manually syncs stable knowledge assets into GPT knowledge storage.
 

models/model_index.json

@@ -11,7 +11,7 @@
       "card_file": "cards/intellectual_archaeology.md",
       "source_article_count": 2,
       "source_evidence_count": 5,
-      "regression_case_count": 17,
+      "regression_case_count": 19,
       "stability_level": "B",
       "regression_status": "pending",
       "status": "draft",
@@ -26,7 +26,7 @@
       "card_file": "cards/qpi.md",
       "source_article_count": 3,
       "source_evidence_count": 10,
-      "regression_case_count": 52,
+      "regression_case_count": 55,
       "stability_level": "B",
       "regression_status": "pending",
       "status": "draft",

reports/Codex新会话交接文档_Round03.2评审准备_2026-06-17.md

@@ -0,0 +1,125 @@
+# Codex 新会话交接文档：Round 03.2 评审准备
+
+Date: 2026-06-17
+
+Repository: `C:\Users\wangq\Documents\Codex\work-projects\the-mindscape-of-bro-tsong`
+
+Current phase: `model_library_mvp`
+
+## 1. 本轮目标
+
+根据 GPT 对 Round 03.1 的 `REVISE` 评审意见，完成一个极小的 Round 03.2 selector negative-gate patch，并准备 03.2 评审资料。
+
+## 2. 已完成
+
+- 将 IA / 思想考古的明确拒绝语从普通 penalty 升级为 model-level hard exclusion。
+- 保留 QPI 显式分析 / 复核入口，但不允许 QPI 选择重新启用被明确拒绝的 IA。
+- 移除 QPI gate 中裸 `如何` 和裸 `判断`。
+- 增加复合 problem-definition 信号，避免普通 how-to / 词义判断误召回 QPI。
+- 加严 `scripts/run_selector_calibration_smoke.py`：
+  - `no_call` 必须 `selected_models == []` 且 `no_call == true`；
+  - `no_call_or_low_priority` 不允许 QPI 或 IA 被选中；
+  - `select_qpi_reject_ia` 必须选中 QPI 且拒绝 IA；
+  - `select_intellectual_archaeology` 如果命中 IA hard exclusion 则失败。
+- 增加 5 条 GPT 指定陷阱：
+  - `case_ia_negative_explicit_do_not_call_001`
+  - `case_ia_negative_only_qpi_001`
+  - `case_qpi_how_to_excel_no_call_001`
+  - `case_qpi_how_to_markdown_pdf_no_call_001`
+  - `case_qpi_dictionary_judgment_no_call_001`
+- 同步 per-model regression、aggregate regression、regression index、model/card index 和相关报告。
+- 生成 Round 03.2 评审包：
+
+```text
+ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/
+```
+- Round 03.2 评审包内上传给 GPT / CCRA 的文件名均带 `03.2` 版本号；raw zip 命名为 `optional_raw_changed_files_03.2.zip`，zip 内保持 source-relative paths。
+
+## 3. 当前计数
+
+- QPI regression cases: 55
+- Intellectual Archaeology regression cases: 19
+- Aggregate regression cases: 74
+- Selector calibration inputs: 90
+
+模型状态没有升级：
+
+```text
+qpi: draft / B / pending
+intellectual_archaeology: draft / B / pending
+```
+
+## 4. 主要变更文件
+
+- `selector/selector_rules.json`
+- `scripts/run_selector_demo.py`
+- `scripts/run_selector_calibration_smoke.py`
+- `selector/selector_calibration_inputs.json`
+- `tests/qpi.regression.json`
+- `tests/intellectual_archaeology.regression.json`
+- `tests/regression_cases.json`
+- `tests/regression_index.json`
+- `models/model_index.json`
+- `cards/card_index.md`
+- `cards/qpi.md`
+- `cards/intellectual_archaeology.md`
+- `docs/DECISIONS.md`
+- `README.md`
+- `selector/README.md`
+- `scripts/README.md`
+- `reports/README.md`
+- `reports/content_review_report_v0.2.md`
+- generated reports under `reports/`
+- Round 03.2 review bundle under `ccra_review_bundle/round-03.2_2026-06-17_selector-negative-gate-patch/`
+
+## 5. 验证结果
+
+以下命令已通过：
+
+```powershell
+python scripts\rebuild_indexes.py --check
+python -m unittest discover -s tests -p "test*.py" -v
+python scripts\validate_model_library.py
+python scripts\check_card_contract.py
+python scripts\run_selector_demo.py
+python scripts\run_selector_regression.py
+python scripts\run_selector_calibration_smoke.py
+python scripts\check_model_card_sync.py
+```
+
+结果：
+
+- Index check: PASS
+- Unit tests: PASS, 17 tests
+- Model library validation: PASS
+- Card contract: PASS
+- Selector demo: PASS
+- Selector regression: PASS
+- Selector calibration smoke: PASS, 90 inputs, 0 failures, 0 warnings
+- Model/card sync: PASS
+
+## 6. 假设
+
+- GPT 的 `Round 03.2 Selector Negative-Gate Patch` 指令是本轮唯一修复范围。
+- `只做 QPI` 和 `只做问题定性` 在当前上下文中表示拒绝 IA，同时允许 QPI。
+- `draft-callable` 只能作为评审报告语言，不写入模型 JSON status。
+
+## 7. 仍不代表已经完成的事
+
+- 没有进入 Round 04 blind test。
+- 没有证明 QPI 或 IA 内容已稳定。
+- 没有升级 `status`、`stability_level` 或 `regression_status`。
+- 没有新增第三模型、LLM selector、RAG、数据库、前后端或完整问答系统。
+
+## 8. 需要 GPT / Owner 判断
+
+- Round 03.2 是否 `pass / revise / block`。
+- Round 03 是否可以正式关闭。
+- 是否允许进入 Round 04 blind input evaluation。
+- 当前 IA hard exclusion 词表是否足够，是否需要定义“用户反转前文拒绝”的显式规则。
+
+## 9. 建议下一步
+
+1. 将 Round 03.2 review bundle 提交给 GPT。
+2. 如果 GPT pass，关闭 Round 03 系列。
+3. 仅在 GPT / Owner 同意后启动 Round 04 blind input evaluation。

reports/README.md

@@ -10,6 +10,7 @@ Current reports:
 - `baseline_before_content_stabilization_v0.2.md`
 - `evidence_coverage_report_v0.2.md`
 - `selector_regression_report_v0.2.md`
+- `selector_calibration_smoke_report.md`
 - `model_card_sync_report_v0.2.md`
 - `content_review_report_v0.2.md`
 - `model_review_status.json`

reports/content_review_report_v0.2.md

@@ -2,7 +2,7 @@
 
 Date: 2026-06-16
 
-Status note: this report is a pre-case-promotion review snapshot. It remains useful for the v0.2 evidence / contract / selector baseline, but its regression counts do not reflect the Round 03 / 03.1 QPI owner-reviewed case promotion. Current QPI regression counts are tracked in `tests/qpi.regression.json`, `models/model_index.json`, and the Round 03.1 review bundle.
+Status note: this report is a pre-case-promotion review snapshot. It remains useful for the v0.2 evidence / contract / selector baseline, but its regression counts do not reflect the Round 03 / 03.1 QPI owner-reviewed case promotion or the Round 03.2 selector negative-gate patch. Current regression counts are tracked in `tests/*.regression.json`, `models/model_index.json`, and the latest review bundle.
 
 ## 1. 本轮修改摘要
 
@@ -111,7 +111,7 @@ Coverage:
 - 已覆盖误用、防误召回、no-call 和 pipeline gate。
 - 仍需人工审查用例真实性和遗漏边界。
 
-Round 03 / 03.1 update: QPI has since expanded to 52 regression cases after owner-reviewed case promotion and selector/no-call repair. This v0.2 table is retained as a historical baseline, not the current count.
+Round 03 / 03.1 / 03.2 update: QPI has since expanded to 55 regression cases after owner-reviewed case promotion, selector/no-call repair, and broad-gate no-call traps. Intellectual Archaeology has expanded to 19 regression cases after Round 03.2 IA hard-exclusion traps. This v0.2 table is retained as a historical baseline, not the current count.
 
 ## 6. Selector Regression 结论
 
@@ -124,10 +124,13 @@ Selector v0.2 仍为规则 selector。
 - QPI-before-IA gate
 - score breakdown
 - rejected model reasons
+- model-level hard exclusion for explicit model refusal
+- stricter calibration smoke checks for no-call, no-call-or-low-priority, and select-qpi-reject-ia
 
 报告位于：
 
 - `reports/selector_regression_report_v0.2.md`
+- `reports/selector_calibration_smoke_report.md`
 
 结论：
 

reports/selector_calibration_smoke_report.md

@@ -4,8 +4,9 @@ Status: `PASS`
 
 Command: `python scripts/run_selector_calibration_smoke.py`
 
-Calibration inputs checked: 85
+Calibration inputs checked: 90
 Failures: 0
+Warnings: 0
 
 ## Cases
 
@@ -93,4 +94,9 @@ Failures: 0
 - `selector_calibration_qpi_research_safety_accountability_001`: PASS; expected=select_qpi; selected=['qpi']; no_call=False
 - `selector_calibration_qpi_research_penalty_integrity_001`: PASS; expected=select_qpi; selected=['qpi']; no_call=False
 - `selector_calibration_direct_summary_no_call_001`: PASS; expected=no_call; selected=[]; no_call=True
+- `selector_calibration_ia_negative_explicit_do_not_call_001`: PASS; expected=select_qpi_reject_ia; selected=['qpi']; no_call=False
+- `selector_calibration_ia_negative_only_qpi_001`: PASS; expected=select_qpi_reject_ia; selected=['qpi']; no_call=False
+- `selector_calibration_qpi_how_to_excel_no_call_001`: PASS; expected=no_call; selected=[]; no_call=True
+- `selector_calibration_qpi_how_to_markdown_pdf_no_call_001`: PASS; expected=no_call; selected=[]; no_call=True
+- `selector_calibration_qpi_dictionary_judgment_no_call_001`: PASS; expected=no_call; selected=[]; no_call=True
 - `selector_calibration_analysis_override_should_call_001`: PASS; expected=select_qpi; selected=['qpi']; no_call=False

reports/selector_regression_report_v0.2.md

@@ -2,7 +2,7 @@
 
 Status: `PASS`
 
-Cases checked: 13
+Cases checked: 18
 Failures: 0
 
 ## Cases
@@ -13,10 +13,15 @@ Failures: 0
 - `case_ia_selector_gate_qpi_first_001`: PASS; selected=['qpi']; no_call=False
 - `case_ia_selector_gate_no_heavy_signal_001`: PASS; selected=[]; no_call=True
 - `case_ia_pipeline_after_qpi_001`: PASS; selected=['intellectual_archaeology', 'qpi']; no_call=False
+- `case_ia_negative_explicit_do_not_call_001`: PASS; selected=['qpi']; no_call=False
+- `case_ia_negative_only_qpi_001`: PASS; selected=['qpi']; no_call=False
 - `case_qpi_no_call_direct_edit_001`: PASS; selected=[]; no_call=True
 - `case_qpi_selector_gate_fact_001`: PASS; selected=[]; no_call=True
 - `case_qpi_pipeline_before_ia_001`: PASS; selected=['qpi']; no_call=False
 - `case_qpi_disappointment_mismatch_diagnostics_001`: PASS; selected=['qpi']; no_call=False
 - `case_qpi_international_logistics_no_call_001`: PASS; selected=[]; no_call=True
 - `case_qpi_direct_summary_no_call_001`: PASS; selected=[]; no_call=True
+- `case_qpi_how_to_excel_no_call_001`: PASS; selected=[]; no_call=True
+- `case_qpi_how_to_markdown_pdf_no_call_001`: PASS; selected=[]; no_call=True
+- `case_qpi_dictionary_judgment_no_call_001`: PASS; selected=[]; no_call=True
 - `case_qpi_analysis_override_should_call_001`: PASS; selected=['qpi']; no_call=False

scripts/README.md

@@ -10,6 +10,7 @@ Current script:
 - `check_model_card_sync.py`
 - `run_selector_regression.py`
 - `run_selector_demo.py`
+- `run_selector_calibration_smoke.py`
 
 Run:
 
@@ -44,7 +45,10 @@ Run selector regression and model/card sync checks:
 
 ```powershell
 python scripts\run_selector_regression.py
+python scripts\run_selector_calibration_smoke.py
 python scripts\check_model_card_sync.py
 ```
 
+`run_selector_calibration_smoke.py` checks selector calibration inputs, including strict `no_call`, `no_call_or_low_priority`, `select_qpi_reject_ia`, and IA heavy-depth gate expectations.
+
 Prefer Python standard library before adding dependencies.

scripts/run_selector_calibration_smoke.py

@@ -2,7 +2,7 @@ import json
 import sys
 from pathlib import Path
 
-from run_selector_demo import load_selector_rules, recommend
+from run_selector_demo import load_selector_rules, model_hard_exclusion_hits, recommend
 
 
 def read_json(path):
@@ -24,9 +24,24 @@ def evaluate_input(root, selector_rules, item):
     selected_ids = selected_model_ids(result)
     expected = item.get("expected_selector_behavior")
     errors = []
+    warnings = []
 
-    if expected in {"no_call", "no_call_or_low_priority"} and "qpi" in selected_ids:
+    model_rules = {
-        errors.append("expected no QPI selection")
+        model_rule["model_id"]: model_rule
+        for model_rule in selector_rules.get("models", [])
+    }
+
+    if expected == "no_call":
+        if selected_ids:
+            errors.append(f"expected no selected models, got {selected_ids}")
+        if result.get("no_call") is not True:
+            errors.append("expected no_call=true")
+
+    if expected == "no_call_or_low_priority":
+        if "qpi" in selected_ids:
+            errors.append("expected no QPI selection")
+        if "intellectual_archaeology" in selected_ids:
+            errors.append("expected no IA selection")
 
     if expected in {"select_qpi", "select_qpi_low_confidence", "select_qpi_reject_ia"}:
         if "qpi" not in selected_ids:
@@ -40,6 +55,15 @@ def evaluate_input(root, selector_rules, item):
             errors.append("expected IA selection")
         if not has_ia_gate(item["input"], selector_rules):
             errors.append("IA selected expectation lacks heavy-depth or QPI-completed gate")
+        ia_hard_exclusion_hits = model_hard_exclusion_hits(
+            item["input"],
+            model_rules.get("intellectual_archaeology", {}),
+        )
+        if ia_hard_exclusion_hits:
+            errors.append("IA selection expected despite IA hard exclusion signal")
+
+    if expected == "select_qpi" and set(selected_ids) - {"qpi"}:
+        warnings.append("unexpected extra selected models: " + ", ".join(sorted(set(selected_ids) - {"qpi"})))
 
     return {
         "case_id": item["case_id"],
@@ -47,12 +71,14 @@ def evaluate_input(root, selector_rules, item):
         "selected_models": selected_ids,
         "no_call": result.get("no_call"),
         "errors": errors,
+        "warnings": warnings,
     }
 
 
 def write_report(root, results):
     report_path = root / "reports" / "selector_calibration_smoke_report.md"
     failures = [result for result in results if result["errors"]]
+    warnings = [result for result in results if result["warnings"]]
     lines = [
         "# Selector Calibration Smoke Report",
         "",
@@ -62,6 +88,7 @@ def write_report(root, results):
         "",
         f"Calibration inputs checked: {len(results)}",
         f"Failures: {len(failures)}",
+        f"Warnings: {len(warnings)}",
         "",
         "## Cases",
         "",
@@ -74,6 +101,8 @@ def write_report(root, results):
         )
         for error in result["errors"]:
             lines.append(f"  - {error}")
+        for warning in result["warnings"]:
+            lines.append(f"  - WARNING: {warning}")
     report_path.write_text("\n".join(lines) + "\n", encoding="utf-8")
     return report_path
 

scripts/run_selector_demo.py

@@ -33,6 +33,10 @@ def hard_no_call_hits(user_input, global_rules):
     return hit_any(user_input, signals)
 
 
+def model_hard_exclusion_hits(user_input, model_rule):
+    return hit_any(user_input, model_rule.get("hard_exclusion_triggers", []))
+
+
 def score_model(model, model_rule, global_rules, user_input, task_type="", pipeline_position=""):
     model_id = model.get("model_id")
     weights = global_rules.get("weights", {})
@@ -146,10 +150,17 @@ def recommend(root, user_input, task_type="", pipeline_position="", threshold=No
         return hard_no_call_result(root, user_input, no_call_hits)
 
     model_rules = rules_by_model(selector_rules)
-    scored = [
+    scored = []
-        score_model(model, model_rules.get(model.get("model_id"), {}), global_rules, user_input, task_type, pipeline_position)
+    for model in load_models(root):
-        for model in load_models(root)
+        model_id = model.get("model_id")
-    ]
+        model_rule = model_rules.get(model_id, {})
+        item = score_model(model, model_rule, global_rules, user_input, task_type, pipeline_position)
+        exclusion_hits = model_hard_exclusion_hits(user_input, model_rule)
+        if exclusion_hits:
+            item["score"] = 0.0
+            item["hard_excluded"] = True
+            item["penalties"].append("model hard exclusion signal: " + "、".join(sorted(set(exclusion_hits))[:4]))
+        scored.append(item)
     no_call = all(item["score"] < threshold for item in scored)
     selected = []
     rejected = []
@@ -160,7 +171,7 @@ def recommend(root, user_input, task_type="", pipeline_position="", threshold=No
             "score": item["score"],
             "reasons": item["reasons"],
             "penalties": item["penalties"],
-            "decision": "selected" if item["score"] >= threshold and not no_call else "rejected"
+            "decision": "selected" if item["score"] >= threshold and not no_call and not item.get("hard_excluded") else "rejected"
         }
         if output_item["decision"] == "selected":
             selected.append(output_item)

selector/README.md

@@ -9,6 +9,7 @@ It should use simple matching rules:
 - Trigger keywords
 - Input types
 - Negative triggers
+- Model-level hard exclusion triggers
 - Pipeline position
 - Selection priority
 - Negative trigger first
@@ -18,6 +19,12 @@ It should use simple matching rules:
 - Direct-execution no-call signals
 - Calibration smoke checks
 
+Round 03.2 selector rule:
+
+- Explicit refusal of a specific model, such as `不要思想考古` or `只做 QPI`, is a model-level hard exclusion for that model.
+- A hard exclusion rejects only the named model; it does not globally no-call the request.
+- Generic bare QPI gate words such as `如何` and `判断` should not be used alone. Use compound problem-definition phrases instead.
+
 Current files:
 
 - `selector_rules.json`

selector/selector_calibration_inputs.json

@@ -591,6 +591,41 @@
       "expected_selector_behavior": "no_call",
       "expected_notes": "Round 03.1 direct summary request must not select QPI by default score."
     },
+    {
+      "case_id": "selector_calibration_ia_negative_explicit_do_not_call_001",
+      "category": "round_03_2_model_level_negative_gate",
+      "input": "QPI 已判断：这是 Issue。先不要思想考古，只做问题定性复核。",
+      "expected_selector_behavior": "select_qpi_reject_ia",
+      "expected_notes": "Explicit IA refusal must be a model-level hard exclusion, not a score penalty."
+    },
+    {
+      "case_id": "selector_calibration_ia_negative_only_qpi_001",
+      "category": "round_03_2_model_level_negative_gate",
+      "input": "不要思想考古，只做 QPI 复核。",
+      "expected_selector_behavior": "select_qpi_reject_ia",
+      "expected_notes": "Only-QPI instruction should reject IA even when IA terms appear."
+    },
+    {
+      "case_id": "selector_calibration_qpi_how_to_excel_no_call_001",
+      "category": "round_03_2_broad_qpi_gate_no_call",
+      "input": "如何在 Excel 里冻结首行？",
+      "expected_selector_behavior": "no_call",
+      "expected_notes": "Generic tool how-to should not be routed to QPI by bare 如何."
+    },
+    {
+      "case_id": "selector_calibration_qpi_how_to_markdown_pdf_no_call_001",
+      "category": "round_03_2_broad_qpi_gate_no_call",
+      "input": "如何把 Markdown 转 PDF？",
+      "expected_selector_behavior": "no_call",
+      "expected_notes": "Generic conversion how-to should not be routed to QPI by bare 如何."
+    },
+    {
+      "case_id": "selector_calibration_qpi_dictionary_judgment_no_call_001",
+      "category": "round_03_2_broad_qpi_gate_no_call",
+      "input": "判断这个词是什么意思。",
+      "expected_selector_behavior": "no_call",
+      "expected_notes": "Dictionary meaning request should not be routed to QPI by bare 判断."
+    },
     {
       "case_id": "selector_calibration_analysis_override_should_call_001",
       "category": "round_03_1_analysis_override",

selector/selector_rules.json

@@ -79,8 +79,15 @@
       "这是问题",
       "执行路径问题",
       "组织共识问题",
-      "如何",
+      "如何判断这是什么问题",
-      "判断",
+      "如何定义这个问题",
+      "如何提高流量",
+      "如何区分 Q/P/I",
+      "判断主导稀缺",
+      "判断这是问题还是课题",
+      "判断应该调用哪个模型",
+      "只做 QPI",
+      "只做问题定性",
       "真实调用",
       "上下文",
       "审计边界",
@@ -92,6 +99,8 @@
       "行动杠杆",
       "依赖风险",
       "误路由",
+      "真实独立 agent 调用记录",
+      "正式输出",
       "过度升维",
       "负激励",
       "成本转嫁",
@@ -202,6 +211,14 @@
         "先不要思想考古",
         "不要思想考古"
       ],
+      "hard_exclusion_triggers": [
+        "不要思想考古",
+        "先不要思想考古",
+        "不要调用思想考古",
+        "不调用思想考古",
+        "只做 QPI",
+        "只做问题定性"
+      ],
       "pipeline_position": "modeling_depth_analysis",
       "selection_priority": 7,
       "routing_notes": "通常在 QPI 判断为中重型难题或课题后调用；遇到轻量事实检索应避免调用。"

tests/intellectual_archaeology.regression.json

@@ -17,6 +17,8 @@
     {"case_id": "case_ia_misuse_layer_jump_001", "model_id": "intellectual_archaeology", "case_type": "misuse", "input": "我们还没弄清应用层事实，就想直接讨论哲学基岩。", "expected_behavior": "应阻止层级跳跃，从应用层事实和证据缺口开始。", "failure_signal": "跳过中间层级。", "expected_output_elements": ["layers_to_analyze", "application"], "should_call_model": true, "expected_primary_model": "intellectual_archaeology", "negative_expected_models": [], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "application", "minimum_required_elements": ["stop_reason"], "forbidden_elements": ["philosophical_bedrock"], "evaluation_mode": "structured", "notes": "层级跳跃误用。"},
     {"case_id": "case_ia_selector_gate_qpi_first_001", "model_id": "intellectual_archaeology", "case_type": "selector_gate", "input": "这个问题到底是缺数据、缺路径还是缺共识？", "expected_behavior": "selector 应先选 QPI，不应直接选思想考古。", "failure_signal": "IA 抢在 QPI 前。", "expected_output_elements": ["qpi first"], "should_call_model": false, "expected_primary_model": "qpi", "negative_expected_models": ["intellectual_archaeology"], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "no_call", "minimum_required_elements": ["rejected_models"], "forbidden_elements": ["selected IA"], "evaluation_mode": "structured", "notes": "QPI-before-IA gate。"},
     {"case_id": "case_ia_selector_gate_no_heavy_signal_001", "model_id": "intellectual_archaeology", "case_type": "selector_gate", "input": "底层这个词怎么翻译？只要一个英文。", "expected_behavior": "虽然有底层一词，也不应召回思想考古。", "failure_signal": "关键词误召回。", "expected_output_elements": ["rejected_models"], "should_call_model": false, "expected_primary_model": "none", "negative_expected_models": ["intellectual_archaeology"], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "no_call", "minimum_required_elements": ["negative trigger"], "forbidden_elements": ["七层"], "evaluation_mode": "structured", "notes": "关键词误召回防护。"},
-    {"case_id": "case_ia_pipeline_after_qpi_001", "model_id": "intellectual_archaeology", "case_type": "pipeline", "input": "QPI 已判断为中重型课题，且需要抽取底层机制和验证路径。", "expected_behavior": "IA 可作为 QPI 后续模型被调用。", "failure_signal": "在 QPI 放行后仍拒绝。", "expected_output_elements": ["QPI 后续", "should_call=true"], "should_call_model": true, "expected_primary_model": "intellectual_archaeology", "negative_expected_models": [], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "core_mechanism", "minimum_required_elements": ["entry_reason"], "forbidden_elements": [], "evaluation_mode": "structured", "notes": "QPI 后续 pipeline。"}
+    {"case_id": "case_ia_pipeline_after_qpi_001", "model_id": "intellectual_archaeology", "case_type": "pipeline", "input": "QPI 已判断为中重型课题，且需要抽取底层机制和验证路径。", "expected_behavior": "IA 可作为 QPI 后续模型被调用。", "failure_signal": "在 QPI 放行后仍拒绝。", "expected_output_elements": ["QPI 后续", "should_call=true"], "should_call_model": true, "expected_primary_model": "intellectual_archaeology", "negative_expected_models": [], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "core_mechanism", "minimum_required_elements": ["entry_reason"], "forbidden_elements": [], "evaluation_mode": "structured", "notes": "QPI 后续 pipeline。"},
+    {"case_id": "case_ia_negative_explicit_do_not_call_001", "model_id": "intellectual_archaeology", "case_type": "selector_gate", "input": "QPI 已判断：这是 Issue。先不要思想考古，只做问题定性复核。", "expected_behavior": "selector 应选择 QPI，并硬拒绝思想考古。", "failure_signal": "因为“思想考古”和“QPI 已判断”加分而仍选中 IA。", "expected_output_elements": ["selected=qpi", "rejected=intellectual_archaeology"], "should_call_model": false, "expected_primary_model": "qpi", "negative_expected_models": ["intellectual_archaeology"], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "no_call", "minimum_required_elements": ["model_level_hard_exclusion"], "forbidden_elements": ["selected IA"], "evaluation_mode": "structured", "notes": "Round 03.2 hard exclusion trap: explicit IA refusal must override score."},
+    {"case_id": "case_ia_negative_only_qpi_001", "model_id": "intellectual_archaeology", "case_type": "selector_gate", "input": "不要思想考古，只做 QPI 复核。", "expected_behavior": "selector 应选择 QPI，并硬拒绝思想考古。", "failure_signal": "IA 负触发只被当作 penalty，仍可能被选中。", "expected_output_elements": ["selected=qpi", "rejected=intellectual_archaeology"], "should_call_model": false, "expected_primary_model": "qpi", "negative_expected_models": ["intellectual_archaeology"], "expected_classification": "not_applicable", "expected_dominant_scarcity": "not_applicable", "expected_max_depth": "no_call", "minimum_required_elements": ["model_level_hard_exclusion"], "forbidden_elements": ["selected IA"], "evaluation_mode": "structured", "notes": "Round 03.2 hard exclusion trap: only-QPI instruction rejects IA."}
   ]
 }

tests/qpi.regression.json

@@ -1550,6 +1550,90 @@
       "evaluation_mode": "structured",
       "notes": "Round 03.1 direct execution no-call."
     },
+    {
+      "case_id": "case_qpi_how_to_excel_no_call_001",
+      "model_id": "qpi",
+      "case_type": "no_call",
+      "input": "如何在 Excel 里冻结首行？",
+      "expected_behavior": "应 no_call：普通工具 how-to 不应因裸“如何”触发 QPI。",
+      "failure_signal": "selector 选中 QPI。",
+      "expected_output_elements": [
+        "classification=no_call"
+      ],
+      "should_call_model": false,
+      "expected_primary_model": "none",
+      "negative_expected_models": [
+        "qpi",
+        "intellectual_archaeology"
+      ],
+      "expected_classification": "no_call",
+      "expected_dominant_scarcity": "not_applicable",
+      "expected_max_depth": "not_applicable",
+      "minimum_required_elements": [
+        "no_call"
+      ],
+      "forbidden_elements": [
+        "problem_definition"
+      ],
+      "evaluation_mode": "structured",
+      "notes": "Round 03.2 no-call trap: narrow bare qpi_gate_signals."
+    },
+    {
+      "case_id": "case_qpi_how_to_markdown_pdf_no_call_001",
+      "model_id": "qpi",
+      "case_type": "no_call",
+      "input": "如何把 Markdown 转 PDF？",
+      "expected_behavior": "应 no_call：普通格式转换 how-to 不应调用 QPI。",
+      "failure_signal": "selector 选中 QPI。",
+      "expected_output_elements": [
+        "classification=no_call"
+      ],
+      "should_call_model": false,
+      "expected_primary_model": "none",
+      "negative_expected_models": [
+        "qpi",
+        "intellectual_archaeology"
+      ],
+      "expected_classification": "no_call",
+      "expected_dominant_scarcity": "not_applicable",
+      "expected_max_depth": "not_applicable",
+      "minimum_required_elements": [
+        "no_call"
+      ],
+      "forbidden_elements": [
+        "problem_definition"
+      ],
+      "evaluation_mode": "structured",
+      "notes": "Round 03.2 no-call trap: generic how-to is not QPI."
+    },
+    {
+      "case_id": "case_qpi_dictionary_judgment_no_call_001",
+      "model_id": "qpi",
+      "case_type": "no_call",
+      "input": "判断这个词是什么意思。",
+      "expected_behavior": "应 no_call：普通词义判断不应因裸“判断”触发 QPI。",
+      "failure_signal": "selector 选中 QPI。",
+      "expected_output_elements": [
+        "classification=no_call"
+      ],
+      "should_call_model": false,
+      "expected_primary_model": "none",
+      "negative_expected_models": [
+        "qpi",
+        "intellectual_archaeology"
+      ],
+      "expected_classification": "no_call",
+      "expected_dominant_scarcity": "not_applicable",
+      "expected_max_depth": "not_applicable",
+      "minimum_required_elements": [
+        "no_call"
+      ],
+      "forbidden_elements": [
+        "problem_definition"
+      ],
+      "evaluation_mode": "structured",
+      "notes": "Round 03.2 no-call trap: dictionary judgment is not QPI."
+    },
     {
       "case_id": "case_qpi_analysis_override_should_call_001",
       "model_id": "qpi",

tests/regression_index.json

@@ -1,33 +1,33 @@
 {
   "index_version": "0.2",
-  "last_updated": "2026-06-16",
+  "last_updated": "2026-06-17",
   "source_of_truth": "per_model_files",
   "aggregate_file": "tests/regression_cases.json",
   "models": [
     {
       "model_id": "qpi",
       "regression_file": "tests/qpi.regression.json",
-      "case_count": 17,
+      "case_count": 55,
       "case_type_counts": {
-        "positive": 9,
+        "boundary": 13,
-        "boundary": 3,
+        "misuse": 16,
-        "misuse": 2,
+        "no_call": 6,
-        "no_call": 1,
+        "pipeline": 1,
-        "selector_gate": 1,
+        "positive": 16,
-        "pipeline": 1
+        "selector_gate": 3
       }
     },
     {
       "model_id": "intellectual_archaeology",
       "regression_file": "tests/intellectual_archaeology.regression.json",
-      "case_count": 17,
+      "case_count": 19,
       "case_type_counts": {
-        "positive": 5,
         "boundary": 3,
         "misuse": 3,
         "no_call": 3,
-        "selector_gate": 2,
+        "pipeline": 1,
-        "pipeline": 1
+        "positive": 5,
+        "selector_gate": 4
       }
     }
   ]