@anhth2/spec-driven-dev-plugin 0.10.0 → 0.11.0

package/skills/qc/qa-runner/e2e.md

@@ -0,0 +1,49 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Gen Script — E2E Journey
+
+Skill **tự chứa**: convert journey `.Test.md` → Python pytest + Playwright đầu→cuối xuyên nhiều màn/module. Chỉ cần load file này.
+
+## Khi nào trigger
+- Convert TC E2E journey (output qa-designer e2e/journey) sang script; TC đã Reviewed
+
+## Khi KHÔNG trigger
+- Test 1 màn/field → `functional/*` · 1 điểm tích hợp → `integration.md`
+
+## Quy ước script (bám CLAUDE.md)
+- Tiền đề: `.Test.md` đã Reviewed. KHÔNG hardcode → `Env.*`/`CONFIG`; KHÔNG `time.sleep()` → `expect()`.
+- Dùng chuỗi Page Object xuyên các màn; verify points (V1…Vn) thành các `assert_*()` rõ ràng.
+- **Tiền điều kiện qua fixture** (tài khoản role, dữ liệu lớp/buổi); **cleanup** sau journey; mỗi journey độc lập.
+- Cần verify hệ thống ngoài (DB/CRM) → client/fixture riêng. Bọc mỗi chặng `with step("…")` (`from utils.steps import step`, **KHÔNG Allure**); assertion `expect()`; marker `@pytest.mark.e2e`.
+- Naming: class `TestFeatureE2E`, method `test_E2E_<NN>_<snake>`.
+
+## Phase 1 — Clarify
+Đọc journey `.Test.md` · các màn/PO + hệ thống verify · tài khoản role & data cần dựng · điểm cleanup.
+
+## Phase 2 — Generate
+Mỗi journey → 1 test; bọc từng bước `with step("…")`; cuối journey verify đủ V1…Vn (tạo/mã/định tuyến/đồng bộ/danh sách).
+Journey còn phụ thuộc gap → tạo test `@pytest.mark.skip(reason="GAP-xx")` hoặc xfail.
+
+## Phase 3 — Verify
+`py_compile` + `pytest --collect-only -q` · chạy (môi trường staging + CRM) · cập nhật Status TC.
+**Phân loại FAIL: script-bug vs product-gap** — journey fail vì 1 bước feature chưa wire = gap (giữ FAIL/skip + bằng chứng), không phải lỗi script; sai selector/state mới sửa script.
+
+## Output
+Script `tests/<project>/e2e/test_<feature>.py` + Page Object/client tái dùng. Bàn giao `qa-reviewer`.
+
+## Phase 4 — Report (bắt buộc sau khi chạy test)
+Report = **Playwright Trace viewer + pytest-html** (KHÔNG Allure, KHÔNG dashboard tự viết).
+
+1. Chạy test kèm pytest-html (trace đã bật sẵn ở conftest → mỗi test có `test-results/<nodeid>/trace.zip`):
+   ```bash
+   pytest tests/<project>/e2e/test_<feature>.py --html=reports/<feature>/report.html --self-contained-html
+   ```
+2. Gửi cho người dùng:
+   - HTML report: `reports/<feature>/report.html` (self-contained, mở trực tiếp).
+   - Trace từng test (debug step-by-step): `python3 -m playwright show-trace test-results/<nodeid>/trace.zip`.
+   - Tóm tắt: **TOTAL / PASS / FAIL / SKIP** + duration.
+3. TC Fail → mở trace tương ứng để xem timeline/DOM snapshot/network, phân loại script-bug vs product-gap; ghi mô tả lỗi tiếng Việt dễ hiểu vào Status/khối kết quả của `.Test.md`.

package/skills/qc/qa-runner/exploratory/session.md

@@ -0,0 +1,36 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Session Template & Convert Findings — Exploratory
+
+Skill **tự chứa**, 2 mode:
+- **Mode 1:** sinh session note template trước khi test.
+- **Mode 2:** convert findings (#BUG/#IDEA) → bug report + functional TC mới.
+
+## Khi nào trigger
+- Mode 1: "tạo template cho session" / trước exploratory session
+- Mode 2: "convert findings" / "tạo bug report" / sau session
+
+## Khi KHÔNG trigger
+- Sinh charter → `qa-designer/exploratory/charter.md` · review session note → `qa-reviewer/script/exploratory.md`
+
+---
+
+## Mode 1 — Generate Session Template
+Input: charter + tour + tester + time-box. Tạo file gồm: metadata (date/tester/charter/tour/env/data) ·
+`#SETUP` (bước chuẩn bị) · `#TEST` (5–8 gợi ý theo tour) · `#BUG` template (title, severity, steps,
+expected/actual) · `#QUESTION`, `#IDEA` placeholder · summary cuối session.
+→ `docs/<project>/exploratory/sessions/<YYYY-MM-DD>_<tester>.md`
+
+## Mode 2 — Convert Findings
+Input: session note (#BUG + #IDEA).
+- **Bug report** mỗi #BUG: title, severity, priority, steps to reproduce, expected/actual, hypothesis root cause.
+- **Functional TC mới:** mỗi #BUG đã fix → 1–2 TC regression; mỗi #IDEA → TC nếu đủ rõ (hoặc backlog).
+  Đặt đúng layer; bám format TC (Test Data list, Trace BR, 🚫 Block); trace "Origin: Exploratory session <date>".
+- **Weekly summary** (nếu yêu cầu): overview, top findings, coverage gap, recommendations.
+
+## Output
+Mode 1: file session note. Mode 2: bug reports + file TC trong `docs/<project>/<feature>/` + summary.

package/skills/qc/qa-runner/functional/api.md

@@ -0,0 +1,35 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Gen Script — Functional API
+
+Skill **tự chứa**: convert `.Test.md` API → Python pytest (requests/httpx), không qua UI. Chỉ cần load file này.
+
+## Khi nào trigger
+- Convert TC API (output qa-designer functional/api) sang script; TC đã Reviewed
+
+## Khi KHÔNG trigger
+- Test qua UI → `functional/gui-screen.md` · tích hợp đa thành phần → `integration.md`
+
+## Quy ước script (bám CLAUDE.md)
+- Tiền đề: `.Test.md` đã Reviewed. KHÔNG hardcode base URL/token → `Env.*`/`CONFIG`.
+- Client API gói trong helper/fixture (base url, auth header); KHÔNG rải request rời rạc trong test.
+- Assertion: `assert` cho status code + field response (jsonpath); `with step(...)` (`from utils.steps`) cho mỗi call.
+- Marker `@pytest.mark.api` + category; fixture `function`; test độc lập (tạo→cleanup data qua API/teardown).
+- Naming: class `TestFeatureAPI`, method `test_TC<NNN>_<snake>`.
+
+## Phase 1 — Clarify
+Đọc `.Test.md` · base url/auth/role · client/fixture đã có chưa · data setup/cleanup.
+
+## Phase 2 — Generate
+Mỗi TC → 1 test gọi endpoint với request từ Test Data; assert **status code + field body** theo Expected.
+Nhóm happy/validation/auth/not-found/edge. Data từ `test_data/`.
+
+## Phase 3 — Verify
+`py_compile` + `pytest --collect-only -q` · chạy · cập nhật Status TC · in mapping.
+
+## Output
+Script `tests/<project>/functional/api/test_<feature>.py` + API client/fixture nếu mới. Bàn giao `qa-reviewer`.

package/skills/qc/qa-runner/functional/gui-feature.md

@@ -0,0 +1,51 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Gen Script — Functional GUI Feature (đa màn hình)
+
+Skill **tự chứa**: convert `.Test.md` feature span ≥2 màn → Python pytest + Playwright. Chỉ cần load file này.
+
+## Khi nào trigger
+- Convert TC feature đa màn (output qa-designer gui-feature) sang script; TC đã Reviewed
+
+## Khi KHÔNG trigger
+- Gọn 1 màn → `functional/gui-screen.md` · API → `functional/api.md` · đầu-cuối + đồng bộ → `e2e.md`
+
+## Quy ước script (bám CLAUDE.md)
+- Tiền đề: `.Test.md` đã Reviewed. KHÔNG `time.sleep()` → `expect()`/auto-wait; KHÔNG hardcode → `Env.*`/`CONFIG`.
+- **Một Page Object / màn**; điều hướng giữa màn là action trả về PO màn kế (`return NextPage(self.page)`).
+- PO 3 lớp (kế thừa slim `BasePage`, **KHÔNG Allure**) + bọc bước `with step("…")` (`from utils.steps import step`); assertion `expect()`; mọi interaction qua PO; selector constants `UPPER_SNAKE`.
+- Marker + fixture pytest-playwright từ root `tests/conftest.py` (`page`/`logged_in_page`/`logged_in_page_gv`/`login_page`/`dashboard`); test độc lập + cleanup; naming `Test*{...}` / `test_TC<NNN>_<snake>`.
+
+## Phase 1 — Clarify
+Đọc `.Test.md` · liệt kê các màn/PO cần · state truyền giữa màn · fixture dựng tiền điều kiện (data qua nhiều bước).
+**Probe DOM thật trước khi viết selector** (SPA React/Next không `data-testid`): dump class/`aria-label`/role → BEM `feature__el`, carousel dot thường `role="tab"` + class `--active` (không `aria-selected`).
+
+## Phase 2 — Generate
+**PHỦ HẾT 100%**: 1 test cho **MỌI** TC trong file (`grep -cE "^#{2,4} *TC_"` = số test phải sinh), KHÔNG chọn tập đại diện, KHÔNG để TC nào Draft; TC bất khả thi → `pytest.skip`/`xfail` + lý do.
+Bọc mỗi chặng bằng `with step("…")`; dùng chuỗi PO theo điều hướng.
+Phủ TC điều hướng forward/back/giữ-reset state. Data từ `test_data/`. Marker mới (`bva ep e2e`…) đăng ký `pytest.ini`.
+
+## Phase 3 — Verify
+`py_compile` + `pytest --collect-only -q` (**số collect = tổng TC**; thiếu → sinh nốt) · chạy · cập nhật Status TC (verify KHÔNG còn Draft) · in mapping.
+**Gom nhóm role/account** tự áp qua `utils/test_ordering.py` (root conftest); fixture auth mới → `register_auth_fixtures([...])`. ⚠️ Run dài bị **WSL suspend** có thể gây flaky login/timeout → re-run TC đó + merge report.
+**Phân loại FAIL: script-bug (sửa selector/logic, chạy lại) vs product-gap** (feature chưa wire/defect → giữ FAIL + ghi bằng chứng vào khối "Kết quả thực thi" đầu `.Test.md`, không fake-pass).
+
+## Output
+Script + nhiều Page Object (mỗi màn) trong `pages/<project>/...`. Bàn giao `qa-reviewer` (script).
+
+## Phase 4 — Report (bắt buộc sau khi chạy test)
+Report = **Playwright Trace viewer + pytest-html** (KHÔNG Allure, KHÔNG dashboard tự viết).
+
+1. Chạy test kèm pytest-html (trace đã bật sẵn ở conftest → mỗi test có `test-results/<nodeid>/trace.zip`):
+   ```bash
+   pytest tests/<project>/.../test_<feature>.py --html=reports/<feature>/report.html --self-contained-html
+   ```
+2. Gửi cho người dùng:
+   - HTML report: `reports/<feature>/report.html` (self-contained, mở trực tiếp).
+   - Trace từng test (debug step-by-step): `python3 -m playwright show-trace test-results/<nodeid>/trace.zip`.
+   - Tóm tắt: **TOTAL / PASS / FAIL / SKIP** + duration.
+3. TC Fail → mở trace tương ứng để xem timeline/DOM snapshot/network, phân loại script-bug vs product-gap; ghi mô tả lỗi tiếng Việt dễ hiểu vào Status/khối kết quả của `.Test.md`.

package/skills/qc/qa-runner/functional/gui-screen.md

@@ -0,0 +1,55 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Gen Script — Functional GUI Screen (1 màn hình)
+
+Skill **tự chứa**: convert `.Test.md` (1 màn) → Python pytest + Playwright. Chỉ cần load file này.
+
+## Khi nào trigger
+- "generate script cho [Screen]" / convert TC sang Python; đã có `.Test.md` (output qa-designer gui-screen), TC đã Reviewed
+
+## Khi KHÔNG trigger
+- Chưa có TC `.md` → chạy qa-designer trước · TC chưa review → qa-reviewer trước
+- Feature cross-screen → `functional/gui-feature.md` · API → `functional/api.md`
+
+## Quy ước script (bám CLAUDE.md)
+- Tiền đề: có `.Test.md` đã Reviewed. KHÔNG `time.sleep()` → `expect()`/auto-wait; KHÔNG hardcode → `Env.*`/`CONFIG`.
+- Page Object 3 lớp (kế thừa slim `BasePage`, **KHÔNG Allure**): locators `_x()` → actions `verb_noun()`+`with step("…")`+`return self` → assertions `assert_x()`+`expect()`; mọi interaction **qua PO** (không `page.click()` trực tiếp); selector constants `UPPER_SNAKE` đầu class.
+- Bọc bước bằng `with step("…")` (`from utils.steps import step`) — **KHÔNG Allure** · marker category+domain (đăng ký `pytest.ini`) · fixture pytest-playwright từ root `tests/conftest.py` (`page`/`logged_in_page`/`logged_in_page_gv`/`login_page`/`dashboard`) · test độc lập + cleanup.
+- Naming: class `TestFeature{UI,Functional,Negative}`, method `test_TC<NNN>_<snake>`; docstring = TC title; comment TC_ID + link `.md`.
+
+## Phase 1 — Clarify
+Đọc `.Test.md` (confirm Reviewed) · platform (web Playwright/mobile) · Page Object đã có chưa → tạo nếu cần · fixture setup data?
+**Probe DOM thật trước khi viết selector** (SPA không `data-testid`): dump class/`aria-label`/role bằng script Playwright tạm → ghi selector đúng (BEM `feature__el`; element interactive có thể `role="tab/menuitem"` + class `--active`).
+
+## Phase 2 — Generate
+**PHỦ HẾT 100%**: sinh 1 `test_TC<NNN>_<scenario>` cho **MỌI** TC trong file — KHÔNG chọn tập đại diện, KHÔNG bỏ TC nào. Đếm tổng TC đầu file (`grep -cE "^#{2,4} *TC_"`) = số test phải sinh.
+TC không thể tự động hóa (precondition bất khả thi, cần data cố định, feature chưa wire) → vẫn viết 1 test với `pytest.skip("lý do")` / `pytest.mark.xfail` — KHÔNG để Draft.
+Map nhóm GUI→`TestFeatureUI`, Functional→`TestFeatureFunctional`, Negative→`TestFeatureNegative`. Assertion qua `assert_*()` của PO; test data từ `test_data/`, không hardcode. Marker mới (`bva ep`…) đăng ký `pytest.ini`.
+
+## Phase 3 — Verify
+`py_compile` + `pytest --collect-only -q` (**số collect = tổng TC trong file**, nếu thiếu → quay lại Phase 2 sinh nốt) · chạy test · cập nhật **Status** TC (Pass/Fail/Skip) · in mapping TC_ID→function→file.
+**Gom nhóm role/account**: thứ tự chạy đã tự gom cùng (role, account) liền nhau qua `utils/test_ordering.py` (hook ở root conftest) — fixture auth mới thì `register_auth_fixtures([...])`.
+⚠️ Run dài có thể bị **WSL suspend** (máy ngủ) làm vài TC lỗi login/timeout = flaky (không phải gap SP) → re-run đúng các TC đó + merge vào report (xem `report/report.md`).
+**Verify KHÔNG còn Draft**: `grep -c "Status: Draft" <file>.Test.md` = 0 trước khi bàn giao.
+**Mỗi FAIL phân loại script-bug vs product-gap** (probe trực tiếp): sai selector/expectation → sửa script & chạy lại; feature không phản hồi sau timeout → giữ FAIL + ghi bằng chứng (không fake-pass).
+
+## Output
+Script `tests/<project>/.../test_<screen>.py` + Page Object `pages/<project>/.../<Screen>Page.py` (nếu mới).
+Bàn giao `qa-reviewer` (script).
+
+## Phase 4 — Report (bắt buộc sau khi chạy test)
+Report = **Playwright Trace viewer + pytest-html** (KHÔNG Allure, KHÔNG dashboard tự viết).
+
+1. Chạy test kèm pytest-html (trace đã bật sẵn ở conftest → mỗi test có `test-results/<nodeid>/trace.zip`):
+   ```bash
+   pytest tests/<project>/.../test_<screen>.py --html=reports/<feature>/report.html --self-contained-html
+   ```
+2. Gửi cho người dùng:
+   - HTML report: `reports/<feature>/report.html` (self-contained, mở trực tiếp).
+   - Trace từng test (debug step-by-step): `python3 -m playwright show-trace test-results/<nodeid>/trace.zip`.
+   - Tóm tắt: **TOTAL / PASS / FAIL / SKIP** + duration.
+3. TC Fail → mở trace tương ứng để xem timeline/DOM snapshot/network, phân loại script-bug vs product-gap; ghi mô tả lỗi tiếng Việt dễ hiểu vào Status/khối kết quả của `.Test.md`.

package/skills/qc/qa-runner/integration.md

@@ -0,0 +1,47 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Gen Script — Integration
+
+Skill **tự chứa**: convert `.Test.md` tích hợp (API↔DB↔service↔queue↔UI) → Python pytest. Chỉ cần load file này.
+
+## Khi nào trigger
+- Convert TC integration (output qa-designer integration/*) sang script; TC đã Reviewed
+
+## Khi KHÔNG trigger
+- Chỉ 1 endpoint/UI đơn → `functional/*` · hành trình đầu-cuối → `e2e.md`
+
+## Quy ước script (bám CLAUDE.md)
+- Tiền đề: `.Test.md` đã Reviewed. KHÔNG hardcode endpoint/DSN/topic → `Env.*`/`CONFIG`.
+- Mỗi hệ thống có client/helper riêng (API client, DB session, Kafka producer/consumer) gói trong fixture.
+- Verify đúng chặng: response (assert), DB (query qua fixture), event (consume + assert payload).
+- **Cleanup bắt buộc** sau khi tạo data; test độc lập; concurrency dùng thread/async khi cần.
+- Bọc mỗi chặng `with step("…")` (`from utils.steps import step`, **KHÔNG Allure**); assertion `expect()`; marker `@pytest.mark.integration` + domain; naming `TestFeatureIntegration` / `test_TC<NNN>_<snake>`.
+
+## Phase 1 — Clarify
+Đọc `.Test.md` · chuỗi tích hợp & chặng cần verify · client/fixture (API/DB/Kafka) đã có chưa · setup/teardown data.
+
+## Phase 2 — Generate
+Mỗi TC → 1 test theo data flow: gọi action → verify từng chặng (response/DB/event). Nhóm happy/contract-negative/failure-retry/concurrency/đồng bộ.
+
+## Phase 3 — Verify
+`py_compile` + `pytest --collect-only -q` · chạy (cần môi trường staging/CRM/Kafka) · cập nhật Status TC.
+
+## Output
+Script `tests/<project>/integration/test_<feature>.py` + client/fixture (DB/Kafka/API) nếu mới. Bàn giao `qa-reviewer`.
+
+## Phase 4 — Report (bắt buộc sau khi chạy test)
+Report = **Playwright Trace viewer + pytest-html** (KHÔNG Allure, KHÔNG dashboard tự viết).
+
+1. Chạy test kèm pytest-html (trace đã bật sẵn ở conftest → mỗi test có `test-results/<nodeid>/trace.zip`):
+   ```bash
+   pytest tests/<project>/integration/test_<feature>.py --html=reports/<feature>/report.html --self-contained-html
+   ```
+2. Gửi cho người dùng:
+   - HTML report: `reports/<feature>/report.html` (self-contained, mở trực tiếp).
+   - Trace từng test (debug step-by-step): `python3 -m playwright show-trace test-results/<nodeid>/trace.zip`.
+   - Tóm tắt: **TOTAL / PASS / FAIL / SKIP** + duration.
+3. TC Fail → mở trace tương ứng để xem timeline/DOM snapshot/network, phân loại script-bug vs product-gap; ghi mô tả lỗi tiếng Việt dễ hiểu vào Status/khối kết quả của `.Test.md`.

package/skills/qc/qa-runner/non-functional.md

@@ -0,0 +1,49 @@
+---
+version: 1.0
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Gen Script — Non-Functional
+
+Skill **tự chứa**: convert TC non-functional (performance/security/accessibility/compatibility) → script/đo. Chỉ cần load file này.
+
+## Khi nào trigger
+- Convert TC non-functional (output qa-designer non-functional) sang script/kịch bản đo; TC đã Reviewed
+
+## Khi KHÔNG trigger
+- Kiểm thử chức năng → `functional/*` · tích hợp → `integration.md`
+
+## Quy ước script (bám CLAUDE.md)
+- Tiền đề: `.Test.md` đã Reviewed, có **ngưỡng đo cụ thể**. KHÔNG hardcode → `Env.*`/`CONFIG`.
+- Mỗi loại dùng công cụ phù hợp, gói trong fixture/helper:
+  - **Performance:** đo response time / throughput (pytest + timer, hoặc tích hợp k6/locust); assert ngưỡng.
+  - **Security:** test authZ (role không quyền → 403), input injection, PII không lộ, session/timeout, rate limit.
+  - **Accessibility:** axe-core/lighthouse qua Playwright; assert vi phạm = 0 ở mức WCAG mục tiêu.
+  - **Compatibility:** parametrize trình duyệt/thiết bị.
+- Assert theo **ngưỡng pass** trong TC bằng `expect()`/`assert`; bọc bước `with step("…")` (`from utils.steps import step`, **KHÔNG Allure**); marker `@pytest.mark.{performance,security,accessibility}`.
+
+## Phase 1 — Clarify
+Đọc `.Test.md` · loại + ngưỡng + công cụ · môi trường/tải mẫu · data đặc biệt.
+
+## Phase 2 — Generate
+Mỗi TC → 1 test đo + assert ngưỡng; đánh dấu test cần môi trường/tải riêng (`@pytest.mark.slow`).
+
+## Phase 3 — Verify
+`py_compile` + collect · chạy (môi trường phù hợp) · cập nhật Status TC + số đo thực tế.
+
+## Output
+Script `tests/<project>/non_functional/test_<feature>.py` + helper đo/scan. Bàn giao `qa-reviewer`.
+
+## Phase 4 — Report (bắt buộc sau khi chạy test)
+Report = **Playwright Trace viewer + pytest-html** (KHÔNG Allure, KHÔNG dashboard tự viết).
+
+1. Chạy test kèm pytest-html (trace đã bật sẵn ở conftest → mỗi test có `test-results/<nodeid>/trace.zip`):
+   ```bash
+   pytest tests/<project>/non_functional/test_<feature>.py --html=reports/<feature>/report.html --self-contained-html
+   ```
+2. Gửi cho người dùng:
+   - HTML report: `reports/<feature>/report.html` (self-contained, mở trực tiếp) + số đo thực tế.
+   - Trace từng test (debug step-by-step): `python3 -m playwright show-trace test-results/<nodeid>/trace.zip`.
+   - Tóm tắt: **TOTAL / PASS / FAIL / SKIP** + duration.
+3. TC Fail → mở trace tương ứng để xem timeline/DOM snapshot/network, phân loại script-bug vs product-gap; ghi mô tả lỗi tiếng Việt dễ hiểu + số đo vào Status/khối kết quả của `.Test.md`.

package/skills/qc/qa-runner/report/report.md

@@ -0,0 +1,37 @@
+---
+version: 1.1
+updated: 2026-06-11
+ported_from: ai-automation-qc-base
+---
+
+# Report — Playwright Trace + pytest-html (stack chuẩn)
+
+Skill **tự chứa**: sinh report sau khi chạy test. Stack chuẩn dự án dùng **Playwright
+Trace viewer** + **pytest-html** (KHÔNG Allure, KHÔNG dashboard tự viết). Chỉ cần load file này.
+
+## Khi nào trigger
+- Sau khi chạy test cần xuất report / xem lại bằng chứng (trace, screenshot).
+
+## Cơ chế (đã dựng sẵn ở tests/conftest.py)
+- **Trace viewer**: fixture `context` (root conftest) gọi `context.tracing.start(screenshots=True, snapshots=True, sources=True)`; teardown lưu `test-results/<nodeid>/trace.zip`.
+  - Xem: `python3 -m playwright show-trace test-results/<...>/trace.zip` → timeline + DOM snapshot + network + console (UI giống Playwright report bản JS).
+- **pytest-html**: 1 file HTML self-contained. Hook `pytest_runtest_makereport` (root conftest) đính screenshot full-page khi FAIL/SKIP.
+  - Mặc định `reports/report.html` (pytest.ini). Theo feature: `--html=reports/<feature>/report.html --self-contained-html`.
+
+## Quy trình
+1. Chạy test với report theo feature:
+   ```
+   python3 -m pytest tests/<project>/<file>.py \
+       --html=reports/<feature>/report.html --self-contained-html
+   ```
+   (Trace tự bật trong conftest → test-results/<nodeid>/trace.zip cho mọi test.)
+2. Báo path:
+   - HTML: `reports/<feature>/report.html` (on WSL, the Windows-accessible path is `\\wsl.localhost\<distro>\<repo-path>\reports\<feature>\report.html`).
+   - Trace TC fail: `test-results/<nodeid>/trace.zip` + lệnh `playwright show-trace`.
+   - Tóm tắt TOTAL / PASS / FAIL / SKIP + duration.
+
+## Nguyên tắc
+- KHÔNG dùng Allure, KHÔNG sinh dashboard HTML tự viết (đã bỏ utils/html_report.py & cộng sự).
+- Evidence trung thực: trace + screenshot phản ánh đúng lần run; FAIL/gap giữ nguyên.
+- `reports/` và `test-results/` đã gitignore (artifact nặng) — không commit.
+- Interpreter `python3`; browser = the system/installed Chromium configured via `browser_type_launch_args` in the root conftest (see the qc-playwright module) — do not hard-code a binary path.

package/skills/setup-ai-first/SKILL.md

@@ -140,6 +140,13 @@ Suggest the logical next command based on workflow phase:
 | /generate-design-spec   | Designer review → Figma links confirmed → PO + Designer sign-off → `/generate-bdd {prd-file}` |
 | /generate-bdd           | `/review-context {feature-file}` to verify coverage |
 | /review-context (BDD)   | `/generate-tech-docs {UC-ID}` if APPROVED; regenerate if NEEDS_FIX |
+| /qc-analyze             | `/qc-plan {UC-ID}` (resolve 🔴 blocker gaps first) |
+| /qc-plan                | `/qc-design-test {UC-ID}`                     |
+| /qc-design-test         | `/qc-review {UC-ID}` (test-case review)       |
+| /qc-review (test-case)  | `/qc-run-test {UC-ID}` if APPROVED; fix TCs if NEEDS_FIX |
+| /qc-run-test            | `/qc-report {UC-ID}` then `/qc-review {UC-ID}` (script review) |
+| /qc-review (script)     | `/qc-report {UC-ID}` then create PR if APPROVED |
+| /qc-report              | `/validate-traces {UC-ID}` to refresh Living Docs (qc_status) |
 | /generate-tech-docs     | `/review-tech-docs {tech-design-file}`        |
 | /review-tech-docs       | `/generate-code {feature-file}` if APPROVED; fix doc if NEEDS_FIX |
 | /generate-code          | First gen → `/review-code {UC-ID}`; re-gen → `/dev-gen-test {UC-ID}` |

package/skills/spec/SKILL.md

@@ -211,6 +211,13 @@ Suggest the logical next command based on workflow phase:
 | /generate-design-spec   | Designer review → Figma links confirmed → PO + Designer sign-off → `/generate-bdd {prd-file}` |
 | /generate-bdd           | `/review-context {feature-file}` to verify coverage |
 | /review-context (BDD)   | `/generate-tech-docs {UC-ID}` if APPROVED; regenerate if NEEDS_FIX |
+| /qc-analyze             | `/qc-plan {UC-ID}` (resolve 🔴 blocker gaps first) |
+| /qc-plan                | `/qc-design-test {UC-ID}`                     |
+| /qc-design-test         | `/qc-review {UC-ID}` (test-case review)       |
+| /qc-review (test-case)  | `/qc-run-test {UC-ID}` if APPROVED; fix TCs if NEEDS_FIX |
+| /qc-run-test            | `/qc-report {UC-ID}` then `/qc-review {UC-ID}` (script review) |
+| /qc-review (script)     | `/qc-report {UC-ID}` then create PR if APPROVED |
+| /qc-report              | `/validate-traces {UC-ID}` to refresh Living Docs (qc_status) |
 | /generate-tech-docs     | `/review-tech-docs {tech-design-file}`        |
 | /review-tech-docs       | `/generate-code {feature-file}` if APPROVED; fix doc if NEEDS_FIX |
 | /generate-code          | First gen → `/review-code {UC-ID}`; re-gen → `/dev-gen-test {UC-ID}` |
@@ -348,6 +355,13 @@ Suggest the logical next command based on workflow phase:
 | /generate-design-spec   | Designer review → Figma links confirmed → PO + Designer sign-off → `/generate-bdd {prd-file}` |
 | /generate-bdd           | `/review-context {feature-file}` to verify coverage |
 | /review-context (BDD)   | `/generate-tech-docs {UC-ID}` if APPROVED; regenerate if NEEDS_FIX |
+| /qc-analyze             | `/qc-plan {UC-ID}` (resolve 🔴 blocker gaps first) |
+| /qc-plan                | `/qc-design-test {UC-ID}`                     |
+| /qc-design-test         | `/qc-review {UC-ID}` (test-case review)       |
+| /qc-review (test-case)  | `/qc-run-test {UC-ID}` if APPROVED; fix TCs if NEEDS_FIX |
+| /qc-run-test            | `/qc-report {UC-ID}` then `/qc-review {UC-ID}` (script review) |
+| /qc-review (script)     | `/qc-report {UC-ID}` then create PR if APPROVED |
+| /qc-report              | `/validate-traces {UC-ID}` to refresh Living Docs (qc_status) |
 | /generate-tech-docs     | `/review-tech-docs {tech-design-file}`        |
 | /review-tech-docs       | `/generate-code {feature-file}` if APPROVED; fix doc if NEEDS_FIX |
 | /generate-code          | First gen → `/review-code {UC-ID}`; re-gen → `/dev-gen-test {UC-ID}` |

package/skills/test/SKILL.md

@@ -254,6 +254,13 @@ Suggest the logical next command based on workflow phase:
 | /generate-design-spec   | Designer review → Figma links confirmed → PO + Designer sign-off → `/generate-bdd {prd-file}` |
 | /generate-bdd           | `/review-context {feature-file}` to verify coverage |
 | /review-context (BDD)   | `/generate-tech-docs {UC-ID}` if APPROVED; regenerate if NEEDS_FIX |
+| /qc-analyze             | `/qc-plan {UC-ID}` (resolve 🔴 blocker gaps first) |
+| /qc-plan                | `/qc-design-test {UC-ID}`                     |
+| /qc-design-test         | `/qc-review {UC-ID}` (test-case review)       |
+| /qc-review (test-case)  | `/qc-run-test {UC-ID}` if APPROVED; fix TCs if NEEDS_FIX |
+| /qc-run-test            | `/qc-report {UC-ID}` then `/qc-review {UC-ID}` (script review) |
+| /qc-review (script)     | `/qc-report {UC-ID}` then create PR if APPROVED |
+| /qc-report              | `/validate-traces {UC-ID}` to refresh Living Docs (qc_status) |
 | /generate-tech-docs     | `/review-tech-docs {tech-design-file}`        |
 | /review-tech-docs       | `/generate-code {feature-file}` if APPROVED; fix doc if NEEDS_FIX |
 | /generate-code          | First gen → `/review-code {UC-ID}`; re-gen → `/dev-gen-test {UC-ID}` |
@@ -420,19 +427,41 @@ If the file does not exist → skip silently.
 
 ---
 
-## Step 3 — [CRITICAL] Load CLAUDE.md
+## Step 3 — [CRITICAL] Load CLAUDE.md (layered: root + service overlay)
 
 *This is the highest-priority context — it defines HOW to write code and documents for this project.*
 
-Read `CLAUDE.md`. Extract and store:
+CLAUDE.md is loaded in **two layers** so umbrella-wide rules and service-specific
+architecture/coding standards compose correctly. The agent always sits at the umbrella
+root, but the implementation code lives in a service submodule with its OWN stack,
+architecture, and conventions — so the service's CLAUDE.md must win for code generation.
+
+**Layer 1 — [BASE] Root CLAUDE.md (umbrella-wide).**
+Read `CLAUDE.md` at the repo root. Treat its contents as the **shared baseline** for the
+whole umbrella — git conventions, data-protection posture, cross-cutting rules, and (in
+single-service mode) the project's only architecture + coding standards.
+
+**Layer 2 — [OVERLAY] Service CLAUDE.md (umbrella mode only).**
+*Run only if `service_root` was set in Step 1.6 (i.e. a real service was routed to).*
+Read `{service_root}/CLAUDE.md`. This file defines the architecture + coding standards of
+the **actual stack being implemented** (e.g. `user-service` = java-spring, `web` = nextjs).
+Overlay it on top of Layer 1: **on any conflict, the service value WINS** for architecture,
+coding standards, and error handling. Layer-1 values that the service does not redefine
+(e.g. git conventions, banned patterns shared org-wide) remain in effect.
+
+From the **merged** result, extract and store:
 
 - **§1 Project Overview** → project name, language, framework, build/test commands, domains
-- **§2 Architecture** → layer order (e.g., Controller → Facade → Service → Repository), architectural rules
-- **§3 Coding Standards** → naming conventions (classes, methods), response wrapper type, forbidden patterns
-- **§5 Error Handling** → exception types, HTTP status code mapping, not-found exception class name
-- **§7 Git Conventions** → branch naming pattern, commit message format
+- **§2 Architecture** → layer order (e.g., Controller → Facade → Service → Repository), architectural rules — *service overlay wins*
+- **§3 Coding Standards** → naming conventions (classes, methods), response wrapper type, forbidden patterns — *service overlay wins*
+- **§5 Error Handling** → exception types, HTTP status code mapping, not-found exception class name — *service overlay wins*
+- **§7 Git Conventions** → branch naming pattern, commit message format — *root baseline unless service redefines*
 
-If `CLAUDE.md` does not exist → note it as missing and continue with project-context.yaml data only.
+**Resolution rules:**
+- If both layers exist → merge as above; record `claude_md_source = root + {service_root}`.
+- If only the service overlay exists (no root CLAUDE.md) → use the service file alone; `claude_md_source = {service_root}`.
+- If `service_root` is set but `{service_root}/CLAUDE.md` is **missing** → fall back to root CLAUDE.md and flag ⚠️ in the Step 7 recap (the service has no architecture/coding-standards definition — code generation will use umbrella defaults, which may be the wrong stack).
+- If neither exists → note CLAUDE.md as missing and continue with project-context.yaml data only.
 
 ---
 
@@ -532,7 +561,8 @@ Output exactly this block:
 [CTX LOADED]
 Stack     : {language} / {framework} / {database}
 Platform  : {active_module} ({platform_type})
-Layers    : {layer order from CLAUDE.md §2, e.g., Controller → Facade → Service → Repository}
+Layers    : {layer order from merged CLAUDE.md §2, e.g., Controller → Facade → Service → Repository}
+CLAUDE.md : {root + {service_root} | {service_root} only | root only | ⚠️ service overlay MISSING — using root | missing}
 Ticket    : {ticket_prefix}-
 Dict      : {loaded — N canonical terms, M banned terms | missing}
 Entities  : {loaded — EntityA, EntityB, EntityC | missing}
@@ -655,6 +685,13 @@ Suggest the logical next command based on workflow phase:
 | /generate-design-spec   | Designer review → Figma links confirmed → PO + Designer sign-off → `/generate-bdd {prd-file}` |
 | /generate-bdd           | `/review-context {feature-file}` to verify coverage |
 | /review-context (BDD)   | `/generate-tech-docs {UC-ID}` if APPROVED; regenerate if NEEDS_FIX |
+| /qc-analyze             | `/qc-plan {UC-ID}` (resolve 🔴 blocker gaps first) |
+| /qc-plan                | `/qc-design-test {UC-ID}`                     |
+| /qc-design-test         | `/qc-review {UC-ID}` (test-case review)       |
+| /qc-review (test-case)  | `/qc-run-test {UC-ID}` if APPROVED; fix TCs if NEEDS_FIX |
+| /qc-run-test            | `/qc-report {UC-ID}` then `/qc-review {UC-ID}` (script review) |
+| /qc-review (script)     | `/qc-report {UC-ID}` then create PR if APPROVED |
+| /qc-report              | `/validate-traces {UC-ID}` to refresh Living Docs (qc_status) |
 | /generate-tech-docs     | `/review-tech-docs {tech-design-file}`        |
 | /review-tech-docs       | `/generate-code {feature-file}` if APPROVED; fix doc if NEEDS_FIX |
 | /generate-code          | First gen → `/review-code {UC-ID}`; re-gen → `/dev-gen-test {UC-ID}` |
@@ -824,19 +861,41 @@ If the file does not exist → skip silently.
 
 ---
 
-## Step 3 — [CRITICAL] Load CLAUDE.md
+## Step 3 — [CRITICAL] Load CLAUDE.md (layered: root + service overlay)
 
 *This is the highest-priority context — it defines HOW to write code and documents for this project.*
 
-Read `CLAUDE.md`. Extract and store:
+CLAUDE.md is loaded in **two layers** so umbrella-wide rules and service-specific
+architecture/coding standards compose correctly. The agent always sits at the umbrella
+root, but the implementation code lives in a service submodule with its OWN stack,
+architecture, and conventions — so the service's CLAUDE.md must win for code generation.
+
+**Layer 1 — [BASE] Root CLAUDE.md (umbrella-wide).**
+Read `CLAUDE.md` at the repo root. Treat its contents as the **shared baseline** for the
+whole umbrella — git conventions, data-protection posture, cross-cutting rules, and (in
+single-service mode) the project's only architecture + coding standards.
+
+**Layer 2 — [OVERLAY] Service CLAUDE.md (umbrella mode only).**
+*Run only if `service_root` was set in Step 1.6 (i.e. a real service was routed to).*
+Read `{service_root}/CLAUDE.md`. This file defines the architecture + coding standards of
+the **actual stack being implemented** (e.g. `user-service` = java-spring, `web` = nextjs).
+Overlay it on top of Layer 1: **on any conflict, the service value WINS** for architecture,
+coding standards, and error handling. Layer-1 values that the service does not redefine
+(e.g. git conventions, banned patterns shared org-wide) remain in effect.
+
+From the **merged** result, extract and store:
 
 - **§1 Project Overview** → project name, language, framework, build/test commands, domains
-- **§2 Architecture** → layer order (e.g., Controller → Facade → Service → Repository), architectural rules
-- **§3 Coding Standards** → naming conventions (classes, methods), response wrapper type, forbidden patterns
-- **§5 Error Handling** → exception types, HTTP status code mapping, not-found exception class name
-- **§7 Git Conventions** → branch naming pattern, commit message format
+- **§2 Architecture** → layer order (e.g., Controller → Facade → Service → Repository), architectural rules — *service overlay wins*
+- **§3 Coding Standards** → naming conventions (classes, methods), response wrapper type, forbidden patterns — *service overlay wins*
+- **§5 Error Handling** → exception types, HTTP status code mapping, not-found exception class name — *service overlay wins*
+- **§7 Git Conventions** → branch naming pattern, commit message format — *root baseline unless service redefines*
 
-If `CLAUDE.md` does not exist → note it as missing and continue with project-context.yaml data only.
+**Resolution rules:**
+- If both layers exist → merge as above; record `claude_md_source = root + {service_root}`.
+- If only the service overlay exists (no root CLAUDE.md) → use the service file alone; `claude_md_source = {service_root}`.
+- If `service_root` is set but `{service_root}/CLAUDE.md` is **missing** → fall back to root CLAUDE.md and flag ⚠️ in the Step 7 recap (the service has no architecture/coding-standards definition — code generation will use umbrella defaults, which may be the wrong stack).
+- If neither exists → note CLAUDE.md as missing and continue with project-context.yaml data only.
 
 ---
 
@@ -936,7 +995,8 @@ Output exactly this block:
 [CTX LOADED]
 Stack     : {language} / {framework} / {database}
 Platform  : {active_module} ({platform_type})
-Layers    : {layer order from CLAUDE.md §2, e.g., Controller → Facade → Service → Repository}
+Layers    : {layer order from merged CLAUDE.md §2, e.g., Controller → Facade → Service → Repository}
+CLAUDE.md : {root + {service_root} | {service_root} only | root only | ⚠️ service overlay MISSING — using root | missing}
 Ticket    : {ticket_prefix}-
 Dict      : {loaded — N canonical terms, M banned terms | missing}
 Entities  : {loaded — EntityA, EntityB, EntityC | missing}
@@ -1074,6 +1134,13 @@ Suggest the logical next command based on workflow phase:
 | /generate-design-spec   | Designer review → Figma links confirmed → PO + Designer sign-off → `/generate-bdd {prd-file}` |
 | /generate-bdd           | `/review-context {feature-file}` to verify coverage |
 | /review-context (BDD)   | `/generate-tech-docs {UC-ID}` if APPROVED; regenerate if NEEDS_FIX |
+| /qc-analyze             | `/qc-plan {UC-ID}` (resolve 🔴 blocker gaps first) |
+| /qc-plan                | `/qc-design-test {UC-ID}`                     |
+| /qc-design-test         | `/qc-review {UC-ID}` (test-case review)       |
+| /qc-review (test-case)  | `/qc-run-test {UC-ID}` if APPROVED; fix TCs if NEEDS_FIX |
+| /qc-run-test            | `/qc-report {UC-ID}` then `/qc-review {UC-ID}` (script review) |
+| /qc-review (script)     | `/qc-report {UC-ID}` then create PR if APPROVED |
+| /qc-report              | `/validate-traces {UC-ID}` to refresh Living Docs (qc_status) |
 | /generate-tech-docs     | `/review-tech-docs {tech-design-file}`        |
 | /review-tech-docs       | `/generate-code {feature-file}` if APPROVED; fix doc if NEEDS_FIX |
 | /generate-code          | First gen → `/review-code {UC-ID}`; re-gen → `/dev-gen-test {UC-ID}` |

package/steps/context-loader.md

@@ -132,19 +132,41 @@ If the file does not exist → skip silently.
 
 ---
 
-## Step 3 — [CRITICAL] Load CLAUDE.md
+## Step 3 — [CRITICAL] Load CLAUDE.md (layered: root + service overlay)
 
 *This is the highest-priority context — it defines HOW to write code and documents for this project.*
 
-Read `CLAUDE.md`. Extract and store:
+CLAUDE.md is loaded in **two layers** so umbrella-wide rules and service-specific
+architecture/coding standards compose correctly. The agent always sits at the umbrella
+root, but the implementation code lives in a service submodule with its OWN stack,
+architecture, and conventions — so the service's CLAUDE.md must win for code generation.
+
+**Layer 1 — [BASE] Root CLAUDE.md (umbrella-wide).**
+Read `CLAUDE.md` at the repo root. Treat its contents as the **shared baseline** for the
+whole umbrella — git conventions, data-protection posture, cross-cutting rules, and (in
+single-service mode) the project's only architecture + coding standards.
+
+**Layer 2 — [OVERLAY] Service CLAUDE.md (umbrella mode only).**
+*Run only if `service_root` was set in Step 1.6 (i.e. a real service was routed to).*
+Read `{service_root}/CLAUDE.md`. This file defines the architecture + coding standards of
+the **actual stack being implemented** (e.g. `user-service` = java-spring, `web` = nextjs).
+Overlay it on top of Layer 1: **on any conflict, the service value WINS** for architecture,
+coding standards, and error handling. Layer-1 values that the service does not redefine
+(e.g. git conventions, banned patterns shared org-wide) remain in effect.
+
+From the **merged** result, extract and store:
 
 - **§1 Project Overview** → project name, language, framework, build/test commands, domains
-- **§2 Architecture** → layer order (e.g., Controller → Facade → Service → Repository), architectural rules
-- **§3 Coding Standards** → naming conventions (classes, methods), response wrapper type, forbidden patterns
-- **§5 Error Handling** → exception types, HTTP status code mapping, not-found exception class name
-- **§7 Git Conventions** → branch naming pattern, commit message format
+- **§2 Architecture** → layer order (e.g., Controller → Facade → Service → Repository), architectural rules — *service overlay wins*
+- **§3 Coding Standards** → naming conventions (classes, methods), response wrapper type, forbidden patterns — *service overlay wins*
+- **§5 Error Handling** → exception types, HTTP status code mapping, not-found exception class name — *service overlay wins*
+- **§7 Git Conventions** → branch naming pattern, commit message format — *root baseline unless service redefines*
 
-If `CLAUDE.md` does not exist → note it as missing and continue with project-context.yaml data only.
+**Resolution rules:**
+- If both layers exist → merge as above; record `claude_md_source = root + {service_root}`.
+- If only the service overlay exists (no root CLAUDE.md) → use the service file alone; `claude_md_source = {service_root}`.
+- If `service_root` is set but `{service_root}/CLAUDE.md` is **missing** → fall back to root CLAUDE.md and flag ⚠️ in the Step 7 recap (the service has no architecture/coding-standards definition — code generation will use umbrella defaults, which may be the wrong stack).
+- If neither exists → note CLAUDE.md as missing and continue with project-context.yaml data only.
 
 ---
 
@@ -244,7 +266,8 @@ Output exactly this block:
 [CTX LOADED]
 Stack     : {language} / {framework} / {database}
 Platform  : {active_module} ({platform_type})
-Layers    : {layer order from CLAUDE.md §2, e.g., Controller → Facade → Service → Repository}
+Layers    : {layer order from merged CLAUDE.md §2, e.g., Controller → Facade → Service → Repository}
+CLAUDE.md : {root + {service_root} | {service_root} only | root only | ⚠️ service overlay MISSING — using root | missing}
 Ticket    : {ticket_prefix}-
 Dict      : {loaded — N canonical terms, M banned terms | missing}
 Entities  : {loaded — EntityA, EntityB, EntityC | missing}