@ngocsangairvds/vsaf 4.1.25 → 4.1.26

package/skills/sdlc/ba-validate-srs/steps-v/step-v-10-report.md

@@ -0,0 +1,308 @@
+---
+# model: haiku
+# findingsDir: '{output_folder}/_validation_temp'
+# contextFile: '{output_folder}/_validation_temp/context.yaml'
+---
+
+# Step V-10: Tổng hợp Report
+
+---
+
+## 0. Load All Findings from YAML Files
+
+**Read all findings YAML files from `{findingsDir}` (do NOT re-read the report prose):**
+- `findings-business-alignment.yaml` (weight 1.2)
+- `findings-completeness.yaml` (weight 1.0)
+- `findings-consistency.yaml` (weight 1.0)
+- `findings-quality.yaml` (weight 1.0)
+- `findings-system-nfr.yaml` (weight 1.0)
+- `findings-delivery-readiness.yaml` (weight 1.0)
+- `findings-risk-analysis.yaml` (weight 0.8)
+- `findings-user-impact.yaml` (weight 0.7)
+
+Extract `dimension_score` and `findings` array from each file. These 8 YAML files (~400 lines total) replace re-reading the entire prose report.
+
+**Score verification (MANDATORY — unconditional):**
+
+Canonical N per dimension: BA=11, COMP=29, CONS=19, QUAL=15, NFR=7, DR=13, RA=13, UI=13
+
+For each findings YAML file:
+1. Read ALL `item_scores` values (must have exactly N entries per dimension above)
+2. Compute `verified_score = round(sum(item_scores.values()) / N × 100, 1)` using canonical N
+3. **ALWAYS use `verified_score` as the authoritative score — unconditionally, regardless of diff size**
+   - If verified ≠ header: log "⚠ {dimension}: header={X}%, calculated={Y}%, using {Y}%"
+   - If verified = header: log "✓ {dimension}: {Y}% (verified)"
+4. Use verified scores for ALL subsequent calculations
+
+Do NOT use `dimension_score` from YAML header under any circumstance — headers are informational only.
+Do NOT apply any post-hoc adjustment or calibration to verified scores.
+
+## 1. Tính Overall Score & Confidence Level
+
+Lấy scores từ các steps trước:
+
+| Dimension | Weight | Score | Status |
+|---|---|---|---|
+| Business Alignment | 1.2 | {biz_alignment_score}% | {OK/WARN/FAIL} |
+| Completeness | 1.0 | {completeness_score}% | {OK/WARN/FAIL} |
+| Consistency | 1.0 | {consistency_score}% | {OK/WARN/FAIL} |
+| Quality | 1.0 | {quality_score}% | {OK/WARN/FAIL} |
+| System & NFR | 1.0 | {system_nfr_score}% | {OK/WARN/FAIL} |
+| Delivery Readiness | 1.0 | {delivery_readiness_score}% | {OK/WARN/FAIL} |
+| Risk Analysis | 0.8 | {risk_score}% | {OK/WARN/FAIL} |
+| User Impact | 0.7 | {user_impact_score}% | {OK/WARN/FAIL} |
+
+```
+overall_score = (biz×1.2 + comp×1.0 + cons×1.0 + qual×1.0 + nfr×1.0 + ready×1.0 + risk×0.8 + user×0.7)
+               / 7.7 × 100%
+```
+
+Overall status: OK ≥ 80%, WARN 60–79%, FAIL < 60%
+
+**Confidence Level:**
+- **High**: ≥ 5 dimensions có Full evidence (score ≥ 0.75)
+- **Medium**: 3–4 dimensions có Full evidence
+- **Low**: ≤ 2 dimensions có Full evidence hoặc có nhiều TBD/Missing
+
+---
+
+## 2. GO / NO-GO Decision
+
+Xác định:
+
+- **GO**: overall_score ≥ 80% AND không có CRITICAL findings
+- **CONDITIONAL GO**: overall_score 60–79% OR có HIGH findings nhưng không CRITICAL
+- **NO-GO**: overall_score < 60% OR có CRITICAL findings
+
+---
+
+## 3. Viết Executive Summary (section 1)
+
+Thay nội dung section "1. Executive Summary" trong report file:
+
+```markdown
+## 1. Executive Summary
+
+**Decision: {GO / CONDITIONAL GO / NO-GO}**
+**Overall Score: {overall_score}% — {overall_status}**
+**Confidence: {High / Medium / Low}**
+
+| Metric | Value |
+|---|---|
+| Files reviewed | {count} |
+| Total findings | {total} |
+| CRITICAL | {count_critical} |
+| HIGH | {count_high} |
+| MEDIUM | {count_medium} |
+| LOW | {count_low} |
+| Unresolved Questions | {count_questions} |
+
+### Tóm tắt
+{2-3 câu tổng quan về chất lượng SRS, highlight vấn đề lớn nhất}
+
+### Top 3 Risks
+1. {risk 1 — 1 dòng}
+2. {risk 2 — 1 dòng}
+3. {risk 3 — 1 dòng}
+```
+
+---
+
+## 4. Viết Overall Score & Confidence (section 2)
+
+```markdown
+## 2. Overall Score & Confidence
+
+**Score: {overall_score}% ({overall_status}) | Confidence: {level}**
+
+| Dimension | Weight | Score | Status |
+|---|---|---|---|
+| Business Alignment | 1.2 | {X}% | {status} |
+| Completeness | 1.0 | {X}% | {status} |
+| Consistency | 1.0 | {X}% | {status} |
+| Quality | 1.0 | {X}% | {status} |
+| System & NFR | 1.0 | {X}% | {status} |
+| Delivery Readiness | 1.0 | {X}% | {status} |
+| Risk Analysis | 0.8 | {X}% | {status} |
+| User Impact | 0.7 | {X}% | {status} |
+| **Overall** | **7.7** | **{X}%** | **{status}** |
+```
+
+---
+
+## 5. Viết Heatmap (section 3)
+
+```markdown
+## 3. Heatmap — Score by Dimension
+
+| Dimension | Score | Bar |
+|---|---|---|
+| Business Alignment | {X}% | {████░░░░░░} |
+| Completeness | {X}% | {████████░░} |
+| Consistency | {X}% | {██████░░░░} |
+| Quality | {X}% | {███████░░░} |
+| System & NFR | {X}% | {████░░░░░░} |
+| Delivery Readiness | {X}% | {██████░░░░} |
+| Risk Analysis | {X}% | {████████░░} |
+| User Impact | {X}% | {█████░░░░░} |
+
+Legend: █ = 10%, OK ≥ 80%, WARN 60–79%, FAIL < 60%
+```
+
+Use █ for filled blocks and░ for empty, 10 blocks total per row.
+
+---
+
+## 6. Viết Top Missing Areas (section 6)
+
+Phân tích findings và liệt kê top areas bị thiếu nhất:
+
+```markdown
+## 6. Top Missing Areas
+
+| Priority | Area | Dimensions Affected | Severity |
+|---|---|---|---|
+| 1 | {VD: NFR / Performance targets} | System & NFR | CRITICAL |
+| 2 | {VD: API contracts thiếu schema} | Delivery Readiness | HIGH |
+| 3 | {VD: Business KPI chưa có} | Business Alignment | HIGH |
+| ... | | | |
+```
+
+---
+
+## 7. Viết Action Plan (section 8)
+
+```markdown
+## 8. Action Plan
+
+### Fix ngay — Trước Sprint (CRITICAL)
+1. {action} — Ref: finding #{n}
+2. ...
+
+### Fix trước Dev — Sprint 1 (HIGH)
+1. {action} — Ref: finding #{n}
+2. ...
+
+### Fix sau Release — Sprint 2+ (MEDIUM/LOW)
+1. {action}
+2. ...
+
+### Cần PO/Stakeholder quyết định
+1. {question} — Ref: question #{n}
+2. ...
+```
+
+---
+
+## 7b. Regression Analysis
+
+Read `regression` from context.yaml.
+
+If `previous_report` exists:
+1. **Score trend:** `{previous_score}% → {current_score}%` (↑/↓/→)
+2. **Fixed findings:** For each previous CRITICAL/HIGH finding ID, check if it still appears in current findings. If not → "FIXED ✓"
+3. **Recurring findings:** Present in both → "RECURRING ⚠ — not fixed since {previous_date}"
+4. **New findings:** Current findings not in previous → "NEW"
+
+Write regression table in report:
+
+```markdown
+## 7b. Regression Analysis
+
+Previous: {previous_report} — {previous_score}% {previous_status}
+Current: {current_score}% {current_status}
+Trend: {↑X% | ↓X% | →}
+
+| Finding | Previous | Current | Status |
+|---|---|---|---|
+| C-001 (service boundary) | CRITICAL | CRITICAL | RECURRING ⚠ |
+| DR-002 (API specs) | HIGH | HIGH | RECURRING ⚠ |
+| RISK-001 (appeal flow) | — | HIGH | NEW |
+```
+
+If `previous_report` is null: write "First validation run — no regression baseline available."
+
+---
+
+## 8. Viết Appendix: Full Scoring Matrix (section 9)
+
+```markdown
+## 9. Appendix: Full Scoring Matrix
+
+### Per-File Summary
+
+| File | Biz Align | Completeness | Consistency | Quality | Sys NFR | Readiness | Risk | User Impact | Average |
+|---|---|---|---|---|---|---|---|---|---|
+| {file1} | {X}% | {X}% | {X}% | {X}% | {X}% | {X}% | {X}% | {X}% | {X}% |
+
+### All Findings
+
+| # | Severity | Dimension | File | Item ID | Score | Evidence | Suggestion | Source |
+|---|---|---|---|---|---|---|---|---|
+| {complete findings table from all steps} |
+```
+
+---
+
+## 9. Update frontmatter
+
+Update report file frontmatter:
+
+```yaml
+overall_score: {overall_score}%
+overall_status: {OK/WARN/FAIL}
+confidence_level: {High/Medium/Low}
+go_nogo: {GO/CONDITIONAL GO/NO-GO}
+stepsCompleted:
+  - step-v-01-discovery
+  - step-v-02-business-alignment
+  - step-v-03-completeness
+  - step-v-04-consistency
+  - step-v-05-quality
+  - step-v-06-system-nfr
+  - step-v-07-delivery-readiness
+  - step-v-08-risk-analysis
+  - step-v-09-user-impact
+  - step-v-10-report
+```
+
+---
+
+## 10. Final Output
+
+Thông báo cho user:
+
+```
+SRS Validation Report (V2) hoàn tất!
+
+Decision: {GO / CONDITIONAL GO / NO-GO}
+Overall Score: {overall_score}% — {overall_status}
+Confidence: {High / Medium / Low}
+Report: {output_path}
+
+Findings: {total} (CRITICAL: {c}, HIGH: {h}, MEDIUM: {m}, LOW: {l})
+Unresolved Questions: {count}
+
+Dimension scores:
+  Business Alignment: {X}% {status} [w:1.2]
+  Completeness:       {X}% {status} [w:1.0]
+  Consistency:        {X}% {status} [w:1.0]
+  Quality:            {X}% {status} [w:1.0]
+  System & NFR:       {X}% {status} [w:1.0]
+  Delivery Readiness: {X}% {status} [w:1.0]
+  Risk Analysis:      {X}% {status} [w:0.8]
+  User Impact:        {X}% {status} [w:0.7]
+
+Action Required:
+  - {count} CRITICAL issues → Fix trước Sprint
+  - {count} HIGH issues → Fix trước implementation
+  - {count} questions → Cần PO decision
+```
+
+**Workflow hoàn tất. Không có step tiếp theo.**
+
+## [NEW] Cleanup
+
+After report is complete:
+- Delete all `{findingsDir}/findings-*.yaml` temp files
+- Keep `{findingsDir}/context.yaml` for potential re-runs

package/skills/sdlc/ba-validate-srs/workflow.md

@@ -0,0 +1,269 @@
+---
+validateWorkflow: './steps-v/step-v-01-discovery.md'
+glossary: './data/srs-review-glossary.md'
+outputDir: '.vsaf/docs/features/{feature-name}'
+upstreamPrd: '.vsaf/docs/features/{feature-name}/02-prd.md'
+upstreamAdr: '.vsaf/docs/features/{feature-name}/03-adr.md'
+upstreamEpics: '.vsaf/docs/features/{feature-name}/04-epics.md'
+kgGraph: 'graphify-out/graph.json'
+domainChecklistDir: './data/domain-checklists'
+parallel_mode: true
+silent_mode: false
+---
+
+# SRS Validate Workflow
+
+**Goal:** Validate SRS documents against 6 review dimensions with weighted scoring, producing actionable findings report.
+
+**Your Role:** SRS Quality Assurance Specialist and Validation Architect.
+
+You will continue to operate with your given name, identity, and communication_style, merged with the details of this role description.
+
+## REVIEW DIMENSIONS & WEIGHTS
+
+| Dimension | Weight | Description |
+|---|---|---|
+| Business Alignment | 1.2 | Mục tiêu: problem statement, KPI/success metrics, scope alignment, out-of-scope clarity |
+| Completeness | 1.0 | Đầy đủ: Overview, Use Cases, AC, NFRs, ERD, Business Rules, Error Codes, Dependencies, Traceability |
+| Consistency | 1.0 | Nhất quán: entity naming, UC ID prefix, state machine, data model, API patterns, terminology |
+| Quality | 1.0 | Chất lượng: ngôn ngữ mơ hồ, testability, AC format, duplicate requirements, scope clarity |
+| System & NFR | 1.0 | Kiến trúc + NFR: performance, reliability, security, observability, scalability |
+| Delivery Readiness | 1.0 | Sẵn sàng triển khai: API contracts, DB schema, integration specs, rollout strategy, migration plan |
+| Risk Analysis | 0.8 | Rủi ro: flow completeness, external dependencies, integration gaps, scope creep, assumptions |
+| User Impact | 0.7 | Tác động người dùng: persona coverage, journey completeness, error UX, onboarding, accessibility |
+
+## SEVERITY LEVELS
+
+| Severity | Description | Action |
+|---|---|---|
+| CRITICAL | Block implementation — dev không thể implement | Phải sửa TRƯỚC khi bắt đầu Sprint |
+| HIGH | Nên clarify sớm — risk rework hoặc bugs | Fix trước khi bắt đầu implementation |
+| MEDIUM | Nice to fix — improve quality, UX, maintainability | Phân bổ vào Sprint 2 |
+| LOW | Minor — pass hoặc cải thiện nhỏ | Fix khi có thời gian (optional enhancement) |
+
+## SCORING
+
+### Score Levels
+- 1.0 = Good — đầy đủ, rõ ràng, không vấn đề
+- 0.75 = Minor issue — vấn đề nhỏ, dễ fix
+- 0.5 = Partial — có nhưng thiếu sót đáng kể
+- 0.25 = Major gap — có hình thức nhưng substance rất thiếu
+- 0.0 = Missing — không có
+
+### Dimension & Overall
+- Dimension score: Weighted average of all findings in dimension
+- Overall score formula:
+  ```
+  overall = (bizalign×1.2 + comp×1.0 + cons×1.0 + qual×1.0 + sysnfr×1.0 + ready×1.0 + risk×0.8 + user×0.7)
+            / (1.2 + 1.0 + 1.0 + 1.0 + 1.0 + 1.0 + 0.8 + 0.7) × 100%
+  ```
+  Total weight divisor = **7.7**
+
+### Thresholds
+- OK: ≥ 80%
+- WARN: 60–79%
+- FAIL: < 60%
+
+### Confidence Level
+| Level | Description |
+|---|---|
+| High | Full evidence, well-defined — tất cả sections đầy đủ |
+| Medium | Some missing sections — một số gaps nhưng còn usable |
+| Low | Many assumptions — nhiều TBD, thiếu evidence cốt lõi |
+
+## WORKFLOW ARCHITECTURE
+
+This uses **step-file architecture** for disciplined execution:
+
+### Core Principles
+
+- **Micro-file Design**: Each step is a self-contained instruction file
+- **Just-In-Time Loading**: Only the current step file is in memory
+- **Sequential Enforcement**: Steps must be completed in order
+- **State Tracking**: Document progress in output file frontmatter using `stepsCompleted` array
+- **Append-Only Building**: Build report by appending findings as directed
+
+### Step Processing Rules
+
+1. **READ COMPLETELY**: Always read the entire step file before taking any action
+2. **FOLLOW SEQUENCE**: Execute all numbered sections in order
+3. **WAIT FOR INPUT**: If a menu is presented, halt and wait for user selection
+4. **CHECK CONTINUATION**: Only proceed to next step when user selects 'C' (Continue)
+5. **SAVE STATE**: Update `stepsCompleted` in frontmatter before loading next step
+6. **LOAD NEXT**: When directed, read fully and follow the next step file
+
+### Critical Rules (NO EXCEPTIONS)
+
+- NEVER load multiple step files simultaneously (sequential mode) — in **parallel mode**, spawn multiple Agents in a SINGLE response instead
+- ALWAYS read entire step file before execution
+- NEVER skip steps or optimize the sequence
+- ALWAYS update frontmatter of output files when writing final output for a step
+- ALWAYS follow the exact instructions in the step file
+- ALWAYS halt at menus and wait for user input
+- NEVER create mental todo lists from future steps
+- 🔀 **PARALLEL MODE**: When `{parallel_mode}` is true, after Step V-01 completes, follow **PARALLEL EXECUTION ROUNDS** below — do NOT load step-v-02 sequentially
+
+## STEP SEQUENCE (V2)
+
+| Step | File | Dimension |
+|---|---|---|
+| V-01 | step-v-01-discovery.md | Discovery & scope |
+| V-02 | step-v-02-business-alignment.md | Business Alignment (weight 1.2) |
+| V-03 | step-v-03-completeness.md | Completeness (weight 1.0) |
+| V-04 | step-v-04-consistency.md | Consistency (weight 1.0) |
+| V-05 | step-v-05-quality.md | Quality (weight 1.0) |
+| V-06 | step-v-06-system-nfr.md | System & NFR (weight 1.0) |
+| V-07 | step-v-07-delivery-readiness.md | Delivery Readiness (weight 1.0) |
+| V-08 | step-v-08-risk-analysis.md | Risk Analysis (weight 0.8) |
+| V-09 | step-v-09-user-impact.md | User Impact (weight 0.7) |
+| V-10 | step-v-10-report.md | Report & Scoring |
+
+## INITIALIZATION SEQUENCE
+
+### 1. Configuration Loading
+
+Resolve config from VSAF context:
+
+- `project_name` → from `CONTEXT.md` or `.vsaf/config.yaml`
+- `feature_name` → from current feature directory name
+- `output_folder` → `.vsaf/docs/features/{feature-name}/`
+- `planning_artifacts` → `.vsaf/docs/features/{feature-name}/`
+- `communication_language` → `vi` (default)
+- `document_output_language` → `vi` (default)
+- `date` → system-generated current datetime
+- `domain` → from `CONTEXT.md` field `domain` (optional, for domain-specific checklists)
+
+YOU MUST ALWAYS SPEAK OUTPUT in Vietnamese.
+YOU MUST ALWAYS WRITE all artifact and document content in Vietnamese (except IDs, technical terms, code references).
+
+### 2. Load Review Glossary
+
+Read and internalize the review glossary from {glossary}. This defines all review dimensions, item ID patterns, question categories, and scoring methodology.
+
+### 3. Route to Validate Workflow
+
+"**Validate Mode: Validating SRS document(s) against 8 review dimensions (V2).**"
+
+Then read fully and follow: `{validateWorkflow}` (steps-v/step-v-01-discovery.md)
+
+---
+
+## PARALLEL EXECUTION ROUNDS
+
+**Triggered when:** `parallel_mode: true` AND Step V-01 (discovery) completes with user selecting 'C'.
+
+**Do NOT load step-v-02 sequentially. Execute rounds below instead.**
+
+---
+
+### Round 0 — Complete (Step V-01 already ran)
+
+Step V-01 has produced:
+- `{output_folder}/_validation_temp/context.yaml` — SRS metadata, section index, UC cache, glossary dimension items, KG pre-loaded data
+- Validation report initialized at path in context.yaml
+
+---
+
+### Round 1 — Parallel Dimension Validation (8 agents simultaneously)
+
+⚠️ **CRITICAL: Spawn ALL 8 agents in a SINGLE response using the Agent tool. Do not send them one at a time.**
+
+Announce to user (unless `silent_mode: true`):
+> "Launching parallel SRS validation across 8 dimension agents. This replaces sequential steps V-02 through V-09..."
+
+**Agent 1 — model: sonnet** (Business Alignment — weight 1.2)
+```
+You are a business alignment validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-02-business-alignment.md
+Write output to: {output_folder}/_validation_temp/findings-business-alignment.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**Agent 2 — model: opus** (Completeness — weight 1.0 — two-pass invention detection)
+```
+You are a completeness validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-03-completeness.md
+Write output to: {output_folder}/_validation_temp/findings-completeness.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+> ⚠️ **DO NOT downgrade Agent 2 to sonnet.** Completeness requires Opus for two-pass invention detection (SRS requirements not traceable to PRD/BRD), cross-file semantic analysis, and subtle naming/scope mismatches (e.g., FR17 label vs PRD FR17 content).
+
+**Agent 3 — model: sonnet** (Consistency — weight 1.0)
+```
+You are a consistency validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-04-consistency.md
+Write output to: {output_folder}/_validation_temp/findings-consistency.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**Agent 4 — model: haiku** (Quality — weight 1.0 — Grep-only mechanical checks)
+```
+You are a quality validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-05-quality.md
+Write output to: {output_folder}/_validation_temp/findings-quality.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**Agent 5 — model: sonnet** (System & NFR — weight 1.0)
+```
+You are a system and NFR validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-06-system-nfr.md
+Write output to: {output_folder}/_validation_temp/findings-system-nfr.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**Agent 6 — model: sonnet** (Delivery Readiness — weight 1.0)
+```
+You are a delivery readiness validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-07-delivery-readiness.md
+Write output to: {output_folder}/_validation_temp/findings-delivery-readiness.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**Agent 7 — model: sonnet** (Risk Analysis — weight 0.8)
+```
+You are a risk analysis validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-08-risk-analysis.md
+Write output to: {output_folder}/_validation_temp/findings-risk-analysis.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**Agent 8 — model: haiku** (User Impact — weight 0.7 — lightweight persona checks)
+```
+You are a user impact validation sub-agent for ba-validate-srs.
+Read file: {output_folder}/_validation_temp/context.yaml
+Then read fully and follow ALL instructions in: ./steps-v/step-v-09-user-impact.md
+Write output to: {output_folder}/_validation_temp/findings-user-impact.yaml
+Do NOT produce any user-facing messages. Write findings file only.
+```
+
+**After ALL 8 agents complete:** Announce (unless silent_mode):
+> "Round 1 complete. All 8 dimensions validated. Running final report synthesis..."
+
+---
+
+### Round 2 — Sequential Report Synthesis
+
+Execute sequentially:
+
+**Step V-10 — model: haiku** (Final report + scoring + cleanup):
+Read fully and follow: `./steps-v/step-v-10-report.md`
+
+This step reads all 8 findings YAML files, computes weighted scores, writes the final validation report, and deletes intermediate findings-*.yaml files.
+
+---
+
+### Fallback: Sequential Mode (`parallel_mode: false`)
+
+When `parallel_mode: false`, load step files in original sequential order after V-01:
+`step-v-02 → step-v-03 → step-v-04 → step-v-05 → step-v-06 → step-v-07 → step-v-08 → step-v-09 → step-v-10`
+
+Each step loads the next via its own `nextStepFile` frontmatter variable.

package/skills/sdlc/ba-write-srs/SKILL.md

@@ -0,0 +1,8 @@
+---
+name: ba-write-srs
+description: "Write SRS feature-level theo template v4 chuẩn 12 sections. Triggers: 'write srs', 'viết srs', 'tạo srs'."
+version: 1.0.0
+author: "@ngocsangairvds/vsaf"
+---
+
+Follow the instructions in ./workflow.md.