@trungtamcongnghedigital/vsaf 1.0.0

package/skills/sdlc/validate-srs/steps-v/step-v-01-discovery.md

@@ -0,0 +1,280 @@
+---
+# model: haiku
+# contextOutput: '{output_folder}/_validation_temp/context.yaml'
+# grepPatterns: '../data/grep-patterns-srs.yaml'
+---
+
+# Step V-01: Discovery — Xác định scope và tạo report skeleton
+
+---
+
+## 1. Xác định SRS files cần validate
+
+Hỏi user:
+
+```
+Bạn muốn validate SRS nào?
+
+1. **Single file** — Validate 1 file SRS cụ thể
+2. **Batch** — Validate nhiều files SRS (chọn danh sách)
+3. **Full** — Validate toàn bộ SRS trong folder output
+
+Chọn (1/2/3):
+```
+
+Chờ user chọn.
+
+---
+
+## 2. Scan files
+
+Dựa vào lựa chọn:
+
+### Option 1 — Single file
+- User cung cấp file path hoặc FR number
+- Đọc file, xác nhận file tồn tại và có nội dung
+
+### Option 2 — Batch
+- Liệt kê tất cả SRS files trong `{output_folder}/`
+- User chọn files cần validate
+- Đọc danh sách files đã chọn
+
+### Option 3 — Full
+- Scan toàn bộ `{output_folder}/05-srs*.md`
+- Liệt kê cho user confirm
+
+---
+
+## 3. Load Project Context (auto — không cần user input)
+
+Tự động load project context từ VSAF artifacts:
+
+1. Read `graphify-out/graph.json` (nếu tồn tại) → extract document nodes, feature nodes, service nodes
+2. For each SRS file in scope:
+   - Determine feature name from file path
+   - Locate upstream docs trong `.vsaf/docs/features/{feature-name}/`
+3. Cross-validation sources (auto-load):
+   - PRD: `.vsaf/docs/features/{feature-name}/02-prd.md`
+   - Architecture: `.vsaf/docs/features/{feature-name}/03-adr.md`
+   - Epics: `.vsaf/docs/features/{feature-name}/04-epics.md` (nếu tồn tại)
+   - BRD: tìm trong `.vsaf/docs/` (optional — chỉ khi project có BRD)
+   - System SRS: tìm trong `.vsaf/docs/` (optional)
+
+**Token efficiency rule:** PRD và ADR trong VSAF thường nhỏ (<500 lines). Đọc toàn bộ file thay vì dùng line-range optimization.
+
+---
+
+## 3b. Build SRS Section Index and UC Cache
+
+After loading KG context, build two caches to eliminate redundant reads in parallel validation steps.
+
+### 3b-1. SRS Section Index
+
+Load `../data/grep-patterns-srs.yaml` → get `srs_structure` patterns.
+
+For each SRS file in scope:
+- Run Grep `^## ` on SRS file, output_mode="content", -n=true → extract all Level-2 section headers with line numbers
+- Build `section_index`: list of `{heading, start_line, end_line}` for each `##` header
+- Compute `end_line` for each heading = next heading's `start_line - 1` (or file total lines for last section)
+
+### 3b-2. UC Cache
+
+For each SRS file:
+- Run Grep `^### .*UC-` with -n=true → extract all UC headers and line numbers
+- For each UC found, record:
+  - `id`: UC identifier (e.g., UC-CA-01)
+  - `title`: UC title text  
+  - `start_line`: line of UC header
+  - `end_line`: start_line of next UC (or +80 as estimate)
+  - `has_gherkin`: run Grep `Given.*When.*Then` within UC line range → true/false
+- Build `uc_cache`: list of all UCs across all SRS files in scope
+
+### 3b-3. Extract Glossary Dimension Items
+
+To avoid each parallel agent loading the full glossary:
+- From the glossary already loaded in workflow initialization, extract item IDs per dimension:
+  - completeness_items: [has_overview, has_use_case_summary, has_glossary, ...]
+  - consistency_items: [uc_id_pattern, screen_id_pattern, ...]
+  - quality_items: [no_ambiguous_language, acceptance_criteria_format, ...]
+  - (etc. for each dimension)
+
+### 3b-4. Write context.yaml
+
+Create directory `{output_folder}/_validation_temp/` if it does not exist.
+
+Write `{output_folder}/_validation_temp/context.yaml`:
+
+```yaml
+# SRS Validation session context — auto-generated by step-v-01
+# DO NOT EDIT manually
+validation_date: "{date}"
+scope: "{single|batch|full}"
+report_path: "{report_output_path}"
+silent_mode: {false}   # set true if --silent invocation arg
+
+# SRS files in scope
+srs_files:
+  - path: "{srs_file_path}"
+    fr_id: "{FR_ID}"
+    section_index:    # from Grep ^## with computed end_line
+      - heading: "1. Mô tả chung"
+        start_line: 39
+        end_line: 134      # next heading start_line - 1
+      - heading: "2. Đặc tả Use Cases"
+        start_line: 135
+        end_line: 1318
+      # ... all ## headings with computed end_line (last section end_line = total file lines)
+    uc_cache:         # from Grep ^### UC-
+      - id: "UC-CA-01"
+        title: "Đăng ký tài khoản chủ động"
+        start_line: 120
+        end_line: 200
+        has_gherkin: true
+      # ... etc
+
+# Project context (VSAF convention)
+upstream:
+  prd_path: ".vsaf/docs/features/{feature-name}/02-prd.md"
+  adr_path: ".vsaf/docs/features/{feature-name}/03-adr.md"
+  epics_path: ".vsaf/docs/features/{feature-name}/04-epics.md"
+  brd_path: null  # set if project has BRD
+  graph_json: "graphify-out/graph.json"  # set if graphify index exists
+
+# Glossary item IDs per dimension (pre-extracted for agent efficiency)
+dimension_items:
+  business_alignment: []
+  completeness: [has_overview, has_use_case_summary, has_glossary, has_references, has_uc_screens, has_uc_sequence, has_uc_flow_table, has_erd, has_error_codes, has_impact, has_business_rules, has_nfrs, has_dependencies, has_state_diagrams, has_traceability, has_edge_cases, has_acceptance_criteria]
+  consistency: []
+  quality: [no_ambiguous_language, acceptance_criteria_format, no_duplicate_requirements, clear_scope, flow_step_numbering, diagram_syntax_valid]
+  system_nfr: []
+  delivery_readiness: []
+  risk_analysis: []
+  user_impact: []
+```
+
+**Token efficiency note:** context.yaml (~100-150 lines) replaces 8+ separate KG file reads across parallel agents. Each agent reads context.yaml once instead of re-loading project artifacts independently.
+
+**Derived-from traceability rule (CRITICAL):**
+SRS is a **derived document** — every requirement MUST trace back to PRD, BRD, or Architecture. If SRS contains requirements not in any upstream source, flag as:
+- **CRITICAL** if SRS invents new functional requirements not in PRD/BRD
+- **HIGH** if SRS adds business rules not traceable to BRD
+- **MEDIUM** if SRS adds NFRs/constraints beyond Architecture specs
+
+Upstream chain: `BRD → PRD → SRS`, `Architecture → SRS`
+
+---
+
+## 3c. Regression Check — Load Previous Reports
+
+Scan `.vsaf/docs/features/{feature-name}/05-srs-validation-report*.md` for existing reports.
+
+If previous reports exist:
+1. Find the most recent report (by date in filename, excluding the report being created now)
+2. Read its frontmatter to get `overall_score`, `overall_status`, `files_reviewed`
+3. Read its "Critical & High Findings" section (section 4) — extract finding IDs
+4. Write to context.yaml:
+
+```yaml
+regression:
+  previous_report: "05-srs-validation-report-2026-04-06.md"
+  previous_score: 72.3
+  previous_status: WARN
+  previous_critical_high_ids:
+    - {id: C-001, dimension: consistency, summary: "service boundary violation"}
+    - {id: DR-002, dimension: delivery, summary: "API specs missing"}
+```
+
+If no previous reports exist:
+```yaml
+regression:
+  previous_report: null
+  note: "First validation run — no regression baseline"
+```
+
+---
+
+## 4. Tạo report skeleton
+
+Tạo file output: `.vsaf/docs/features/{feature-name}/05-srs-validation-report.md`
+
+Viết frontmatter:
+
+```yaml
+---
+type: srs-validation-report
+date: {date}
+validator: BA Validate SRS Skill
+scope: {scope}
+files_reviewed:
+  - {list of files}
+overall_score: TBD
+overall_status: TBD
+stepsCompleted:
+  - step-v-01-discovery
+---
+```
+
+Viết section headers (chưa có nội dung):
+
+```markdown
+## 1. Executive Summary
+
+_(Sẽ được điền ở Step V-10)_
+
+## 2. Overall Score & Confidence
+
+_(Sẽ được điền ở Step V-10)_
+
+## 3. Heatmap — Score by Dimension
+
+_(Sẽ được điền ở Step V-10)_
+
+## 4. Critical & High Findings
+
+_(Sẽ được điền ở Steps V-02 → V-09)_
+
+## 5. Medium & Low Findings
+
+_(Sẽ được điền ở Steps V-02 → V-09)_
+
+## 6. Top Missing Areas
+
+_(Sẽ được điền ở Step V-10)_
+
+## 7. Unresolved Questions
+
+_(Sẽ được điền ở Steps V-02 → V-09)_
+
+## 8. Action Plan
+
+_(Sẽ được điền ở Step V-10)_
+
+## 9. Appendix: Full Scoring Matrix
+
+_(Sẽ được điền ở Step V-10)_
+```
+
+---
+
+## 5. Confirm & Route
+
+Xuất cho user:
+
+```
+Scope: {scope}
+Files: {count} file(s)
+  - {file list}
+Cross-validate: PRD=auto, ADR=auto, Epics=auto (if exists), BRD=auto (if exists)
+Report: .vsaf/docs/features/{feature-name}/05-srs-validation-report.md
+Context: {output_folder}/_validation_temp/context.yaml ✓
+UC Cache: {uc_count} UCs indexed ✓
+Graph: {graph_json status}
+
+[C] Continue — Bắt đầu validate Business Alignment (Step V-02)
+[X] Cancel — Dừng lại
+```
+
+**Chờ user chọn.**
+
+Khi user chọn **C** và `{parallel_mode}` là `true`: Hiển thị "Parallel validation mode: launching Round 1..." rồi follow **PARALLEL EXECUTION ROUNDS → Round 1** trong workflow.md
+Khi user chọn **C** và `{parallel_mode}` là `false`: đọc fully và follow `./step-v-02-business-alignment.md` (sequential fallback)

package/skills/sdlc/validate-srs/steps-v/step-v-02-business-alignment.md

@@ -0,0 +1,351 @@
+---
+# model: sonnet
+# findingsOutput: '{output_folder}/_validation_temp/findings-business-alignment.yaml'
+# contextFile: '{output_folder}/_validation_temp/context.yaml'
+---
+
+# Step V-02: Business Alignment — Validate mục tiêu kinh doanh
+
+Weight: **1.2** (highest — foundation of the entire SRS)
+
+---
+
+## 0. Load Context
+Read `{contextFile}` → get `srs_path`, `kg.prd_key_sections`, `kg.brd_key_sections`. Use KG line ranges for all upstream document reads — do NOT read full PRD/BRD.
+
+## 1. Problem Statement check
+
+Kiểm tra mỗi SRS file:
+
+- Có section mô tả **vấn đề cần giải quyết** (problem statement)?
+- Problem statement rõ ràng: ai bị ảnh hưởng, ảnh hưởng như thế nào, tại sao cần giải quyết?
+- Phân biệt được problem vs solution?
+
+**Scoring:**
+- Có rõ ràng, cụ thể → 1.0
+- Có nhưng chung chung / generic → 0.5
+- Chỉ có solution description, không có problem → 0.25
+- Không có → 0.0
+
+Item ID: `problem_statement_clear`
+
+---
+
+## 2. Business Goal & KPI check
+
+Kiểm tra:
+
+- Có **business goal** / mục tiêu kinh doanh cụ thể?
+- Có **KPI hoặc success metrics** đo lường được?
+  - Ví dụ: conversion rate, revenue target, NPS, error reduction %, SLA
+- Metrics có SMART (Specific, Measurable, Achievable, Relevant, Time-bound)?
+
+**Scoring:**
+- Goal + KPI + SMART → 1.0
+- Goal + KPI nhưng không SMART → 0.75
+- Có goal nhưng không KPI → 0.5
+- Chỉ mention feature, không có goal/KPI → 0.25
+- Không có → 0.0
+
+Item ID: `business_goal_kpi_defined`
+
+---
+
+## 3. Scope Alignment check
+
+Kiểm tra SRS scope có aligned với business goal:
+
+- Features trong SRS có trực tiếp address business goal không?
+- Có features nào vượt quá / không liên quan business goal? → scope creep risk
+- Có features nào cần thiết nhưng bị thiếu để đạt goal?
+
+**Scoring:**
+- Fully aligned, không gap, không excess → 1.0
+- Mostly aligned, minor mismatch → 0.75
+- Partial alignment, some gaps hoặc excess → 0.5
+- Misaligned — features không rõ liên quan đến goal → 0.25
+- Không thể đánh giá vì không có goal → 0.0
+
+Item ID: `scope_aligned_with_goal`
+
+---
+
+## 4. Out-of-Scope Clarity check
+
+Kiểm tra:
+
+- Có section **ngoài phạm vi** (out-of-scope) rõ ràng?
+- Out-of-scope có liệt kê các items explicitly?
+- Có ranh giới rõ với các FRs liên quan (để tránh assumption)?
+
+**Scoring:**
+- Có explicit out-of-scope list + boundary rõ → 1.0
+- Có out-of-scope nhưng generic → 0.75
+- Implied nhưng không explicit → 0.5
+- Không có → 0.0
+
+Item ID: `out_of_scope_defined`
+
+---
+
+## 5. Stakeholder & Context check
+
+Kiểm tra:
+
+- Ai là stakeholders chính? (PO, Business, Dev, Legal, Partner)
+- Context về regulatory / compliance requirements liên quan?
+- Dependencies với business processes khác?
+
+**Scoring:**
+- Đầy đủ stakeholders + context → 1.0
+- Có stakeholders nhưng thiếu context → 0.75
+- Chỉ có actors technical, thiếu business stakeholders → 0.5
+- Không có → 0.0
+
+Item ID: `stakeholder_context_defined`
+
+---
+
+## 6. Business Rule assumptions check
+
+Kiểm tra mỗi Business Rule (BR) trong SRS:
+
+- BR có ghi nguồn xác nhận rõ ràng: PRD AC, BRD, nghị định, hoặc PO sign-off?
+- **Không chấp nhận**: BR được viết sẵn trong luồng nhưng file ghi "câu hỏi cần confirm", "cần PO xác nhận", "giả định tạm", hoặc tương đương — đây là CONTRADICTION: luồng đã cứng hoá một quyết định chưa được chốt.
+- BR có assumption ngầm (implicit business decision) mà không được ghi nhận?
+  - VD: auto-cancel ngay sau grace period (không có final notice) — quyết định này từ đâu?
+  - VD: không cho gia hạn khi có claim đang mở — quyết định này ai confirm?
+
+**Scoring:**
+- Tất cả BR có nguồn xác nhận rõ ràng → 1.0
+- Phần lớn có nguồn, 1-2 BR implicit nhưng low-risk → 0.75
+- Nhiều BR implicit, hoặc 1+ BR contradicts stated open question → 0.25
+- Luồng đã cứng hoá quyết định chưa được business confirm → 0.0 (CRITICAL)
+
+Item ID: `br_assumptions_confirmed`
+
+---
+
+## 6b. PRD Business Goal Alignment (cross-validation via KG)
+
+**Token-efficient:** Read PRD entity → find FR section line range → targeted read.
+
+1. From KG: read PRD entity `key_sections` to find the FR section matching this SRS's FR ID
+2. Targeted read: `Read(prd_path, offset=section_start, limit=section_end-section_start)`
+3. Compare:
+   - SRS problem statement aligns with PRD FR description?
+   - SRS business goals/KPIs match PRD acceptance criteria for this FR?
+   - SRS scope matches PRD FR scope?
+
+**Scoring:**
+- Fully aligned with PRD → 1.0
+- Minor misalignment → 0.75
+- Significant gaps between SRS and PRD → 0.5
+- Contradicts PRD → 0.25
+- PRD has no matching FR → 0.0
+
+Item ID: `prd_business_alignment`
+
+## 6c. BRD Business Rule Coverage (cross-validation via KG)
+
+**Token-efficient:** Read BRD entity → find FR section line range → targeted read.
+
+1. From KG: read BRD entity `key_sections` to find the FR group matching this SRS's FR ID
+2. Targeted read of BRD FR section
+3. Compare:
+   - BRD business rules for this FR are reflected in SRS Business Rules section?
+   - BRD use case descriptions covered by SRS UCs?
+   - Any BRD requirement silently dropped in SRS?
+
+**Scoring:**
+- All BRD requirements covered → 1.0
+- Most covered, minor gaps → 0.75
+- Significant BRD requirements missing from SRS → 0.5
+- Major BRD requirements not addressed → 0.25
+- Cannot assess (BRD has no matching section) → 0.0
+
+Item ID: `brd_business_rule_coverage`
+
+## 6d. Scope Invention Check (derived-from rule)
+
+Check if SRS scope EXCEEDS what PRD/BRD authorize:
+
+1. List all SRS use cases and features
+2. For each: can it be derived from PRD FR scope + BRD business requirements?
+3. Flag any SRS content that appears to be invented (not in upstream sources):
+   - New actors not in PRD/BRD → CRITICAL
+   - New business flows not in PRD user journeys → CRITICAL
+   - Expanded scope beyond PRD FR boundaries → HIGH
+
+Item ID: `scope_invention_check`
+
+---
+
+## 6f. Process Alignment — Current vs Future Process
+
+Kiểm tra SRS có mô tả rõ sự chuyển đổi quy trình từ hiện tại sang tương lai:
+
+- **AS-IS process**: SRS hoặc section phụ trợ có ghi nhận quy trình hiện tại (thủ công, tool cũ, pain point) không?
+- **TO-BE process**: Flow mới trong SRS có giải quyết đúng pain point của AS-IS không?
+- **Gap rõ ràng**: Sự khác biệt giữa AS-IS và TO-BE được phản ánh rõ trong scope + business rule không?
+- **Stakeholder impact**: Người dùng hiện tại (ví dụ: nhân viên nghiệp vụ, agent) có quy trình thay đổi lớn nhưng SRS không mô tả transition plan / change impact?
+
+**Không yêu cầu SRS phải có section riêng AS-IS/TO-BE.** Chỉ cần đảm bảo TO-BE flows phản ánh đúng bài toán thực tế, không viết trên giả định sai về trạng thái hiện tại.
+
+**Dấu hiệu SAI:**
+- SRS giả định flow đã hoàn toàn digital nhưng BRD ghi vẫn còn bước manual offline
+- SRS bỏ qua bước hiện tại đang làm thủ công mà không nêu lý do
+- TO-BE flow không address pain point nào được nêu trong BRD problem statement
+
+**Scoring:**
+- SRS phản ánh đúng TO-BE, có trace đến AS-IS pain points → 1.0
+- TO-BE đúng nhưng không ghi rõ gap / change impact với AS-IS → 0.75
+- TO-BE có chỗ giả định sai về quy trình hiện tại (minor) → 0.5
+- TO-BE mâu thuẫn với AS-IS thực tế mô tả trong BRD → 0.25
+- Không thể đánh giá (BRD không có mô tả quy trình hiện tại) → skip
+
+Item ID: `process_alignment_current_future`
+
+---
+
+## 6e. Upstream Artifact Alignment (BRD/PRD vs SRS — general diff)
+
+**Zero extra reads:** Reuses the BRD and PRD sections already loaded by 6b and 6c.
+
+This check catches ANY structured artifact mismatch between upstream docs and SRS — not limited to state machines. It extracts all structured artifacts from both sides and runs a general diff.
+
+#### Artifact Extraction Checklist
+
+From the BRD and PRD sections already in context (loaded by 6b/6c), extract:
+
+| # | Artifact Type | Extraction Method | SRS Counterpart |
+|---|--------------|-------------------|-----------------|
+| A1 | Entity status/lifecycle enum values | `entity_enum_values` pattern on BRD entity table | SRS Section 8 state diagram + entity tables |
+| A2 | Entity field names (if in BRD/PRD only) | `entity_field_row` pattern on BRD entity table | SRS flow table "Bảng/Thực thể" column (no ERD in SRS) |
+| A3 | State transitions + ordering | `plantuml_state` + `state_transition` patterns on BRD state diagram | SRS Section 8 PlantUML |
+| A4 | Actor/role names | `actor_name_in_table` pattern on BRD requirement tables | SRS UC actor columns |
+| A5 | Phase assignments (MVP/1.5/2) | `phase_assignment` pattern on BRD/PRD FR table | SRS scope section |
+| A6 | Status names embedded in PRD ACs | `state_chain` pattern on PRD AC section | SRS canonical state names |
+| A7 | Business rule references | BR-xxx pattern on BRD requirements | SRS Section 5 BR table |
+| A8 | Vietnamese state flow text | `state_flow_text` pattern on BRD business reqs | SRS state diagram (translated) |
+
+**Not all artifact types will be present for every FR.** Extract what's available; skip what's not.
+
+#### Diff Algorithm
+
+For each artifact type found in BOTH upstream and SRS:
+
+1. **Build two lists:** `upstream_values` and `srs_values`
+2. **Compute diff:**
+   - **Additions** (in SRS but not upstream): may be legitimate decomposition or invention
+   - **Removals** (in upstream but not SRS): may be a silent drop or intentional scope reduction
+   - **Renames** (same position/semantics, different name): naming inconsistency
+   - **Reorderings** (same items, different sequence): transition order conflict
+   - **Type mismatches** (same field name, different data type): technical inconsistency
+
+3. **Severity assignment** (LLM judgment per diff item):
+   - Removal of upstream-defined item with no SRS note → **CRITICAL** (silent drop)
+   - Addition in SRS with no upstream source → **HIGH** (invention — cross-ref with 6d)
+   - Name mismatch for same concept → **HIGH** (causes Dev/QA confusion)
+   - Transition order reversal → **HIGH** (changes business logic)
+   - Data type mismatch → **HIGH** (runtime bugs)
+   - Legitimate decomposition (1 upstream → N SRS) documented in SRS → **LOW**
+   - Phase assignment difference → **MEDIUM** (planning confusion)
+   - Actor name variant for same role → **MEDIUM**
+
+#### What NOT to flag
+
+- SRS naturally has more detail than BRD/PRD — more fields, more states, more transitions is expected
+- Only flag when SRS CONTRADICTS, DROPS, or RENAMES without documentation
+- PRD is user-facing; SRS has internal states — SRS having states not in PRD is normal
+
+**Scoring:**
+- All upstream artifacts traceable in SRS, no contradictions → 1.0
+- Minor: 1-2 naming variants, documented decompositions → 0.75
+- Significant: undocumented additions/renames, or type mismatches → 0.5
+- Major: silently dropped upstream items or transition order conflicts → 0.25
+- Fundamental contradictions (SRS says X, BRD says not-X) → 0.0
+
+Item ID: `upstream_artifact_alignment`
+
+---
+
+## 7. Ghi findings
+
+```markdown
+| # | Severity | Dimension | File | Item ID | Score | Evidence | Suggestion | Business Impact | Source |
+```
+
+Each finding MUST include `business_impact` with three sub-fields:
+- **WHO**: Primary stakeholder affected (Dev, QA, PO, Ops, Customer)
+- **WHAT**: Impact type (Blocked, Rework risk, UX degradation, Compliance risk)
+- **WHEN**: When impact materializes (Before Sprint, During Sprint, After Release)
+
+Scoring scale: 1.0 / 0.75 / 0.5 / 0.25 / 0.0
+
+---
+
+## 8. Tính Business Alignment dimension score
+
+```
+business_alignment_score = round(sum(item_scores.values()) / 11 × 100, 1)
+```
+
+Canonical item IDs (N = 11, fixed):
+`problem_statement_clear`, `business_goal_kpi_defined`, `scope_aligned_with_goal`, `out_of_scope_defined`, `stakeholder_context_defined`, `br_assumptions_confirmed`, `prd_business_alignment`, `brd_business_rule_coverage`, `scope_invention_check`, `process_alignment_current_future`, `upstream_artifact_alignment`
+
+⚠️ MANDATORY: N = 11 (fixed). N/A items → score = 1.0 (do not skip from denominator). No post-hoc calibration. dimension_score in YAML = formula result only.
+
+Weight: **1.2**
+
+---
+
+## 9. Append findings vào report
+
+Append CRITICAL/HIGH → section "4. Critical & High Findings".
+Append MEDIUM/LOW → section "5. Medium & Low Findings".
+
+Update frontmatter: thêm `step-v-02-business-alignment` vào `stepsCompleted`.
+
+---
+
+## [NEW] Write Structured Findings
+Write `{findingsOutput}`:
+```yaml
+dimension: business_alignment
+weight: 1.2
+dimension_score: 0   # = round(sum(item_scores.values()) / 11 × 100, 1)
+status: OK
+findings: []
+summary: {total: 0, critical: 0, high: 0, medium: 0, low: 0}
+item_scores:
+  problem_statement_clear: 0.0
+  business_goal_kpi_defined: 0.0
+  scope_aligned_with_goal: 0.0
+  out_of_scope_defined: 0.0
+  stakeholder_context_defined: 0.0
+  br_assumptions_confirmed: 0.0
+  prd_business_alignment: 0.0
+  brd_business_rule_coverage: 0.0
+  scope_invention_check: 0.0
+  process_alignment_current_future: 0.0
+  upstream_artifact_alignment: 0.0
+```
+
+## 10. Menu
+
+```
+Business Alignment Score: {X}% ({status}) [weight 1.2]
+- CRITICAL: {count}
+- HIGH: {count}
+- MEDIUM: {count}
+- LOW: {count}
+
+[C] Continue — Sang Completeness (Step V-03)
+[D] Detail — Xem chi tiết findings
+[X] Cancel
+```
+
+**Chờ user chọn.**
+
+Khi user chọn **C**: đọc fully và follow `./step-v-03-completeness.md`