@longph2102/v-flow 1.5.0

package/skills/implement/tdd-driver/config.json

@@ -0,0 +1,33 @@
+{
+  "_note": "setup_questions là PROMPT. Đáp án resolve persist vào .v-flow/config.yaml (xem .v-flow/templates/vflow-config-template.yaml). Skill đọc .v-flow/config.yaml trước, thiếu mới dùng default dưới đây.",
+  "answers_persisted_to": ".v-flow/config.yaml",
+  "setup_questions": [
+    {
+      "id": "test_framework",
+      "question": "Test framework nào đang dùng?",
+      "options": ["jest", "mocha", "vitest", "flutter_test", "pytest", "other"],
+      "default": "jest"
+    },
+    {
+      "id": "coverage_target",
+      "question": "Coverage target tối thiểu (%)?",
+      "type": "number",
+      "default": 80
+    },
+    {
+      "id": "test_location",
+      "question": "Test files đặt ở đâu?",
+      "options": ["colocated", "test/", "__tests__/", "spec/"],
+      "default": "test/"
+    }
+  ],
+  "hooks": {
+    "pre_implement": "Đọc test-cases.md → map TC-xxx cho TASK hiện tại",
+    "post_implement": "Chạy full test suite → verify không regression"
+  },
+  "memory": {
+    "file": "tdd-history.log",
+    "format": "[date] | [feature] | [task] | [tests_written] | [tests_passed] | [refactor_needed]",
+    "purpose": "Track TDD discipline — Agent đọc để biết patterns trước đó"
+  }
+}

package/skills/implement/tdd-driver/examples/good/tdd-cycle-product-repo.md

@@ -0,0 +1,81 @@
+# Good Example — TDD Cycle cho T-003: Product Repository
+
+## Context
+- Task: Tạo ProductRepository với parameterized SQL
+- REQ: REQ-U01 (parameterized queries), REQ-E05 (soft delete)
+- Test cases: TC-01 (parameterized), TC-09 (soft delete)
+
+## 🔴 RED — Viết test trước
+
+```javascript
+// test/repositories/product.repo.test.js
+const { ProductRepository } = require('../../repositories/product.repo');
+
+describe('ProductRepository', () => {
+  let repo, mockPool;
+
+  beforeEach(() => {
+    mockPool = { query: jest.fn() };
+    repo = new ProductRepository(mockPool);
+  });
+
+  // TC-01: REQ-U01 — parameterized query
+  it('should_use_parameterized_query_when_finding_by_id', async () => {
+    mockPool.query.mockResolvedValue({ rows: [{ id: 1, name: 'Test' }] });
+    
+    await repo.findById(1);
+    
+    // Assert: query uses $1 parameter, NOT string concat
+    const [sql, params] = mockPool.query.mock.calls[0];
+    expect(sql).toContain('$1');
+    expect(sql).not.toMatch(/['"]?\s*\+/); // no string concat
+    expect(params).toEqual([1]);
+  });
+
+  // TC-09: REQ-E05 — soft delete
+  it('should_set_deleted_at_when_soft_deleting', async () => {
+    mockPool.query.mockResolvedValue({ rowCount: 1 });
+    
+    await repo.softDelete(1);
+    
+    const [sql] = mockPool.query.mock.calls[0];
+    expect(sql).toContain('deleted_at');
+    expect(sql).not.toContain('DELETE FROM');
+  });
+});
+```
+
+**Run → FAIL**: `ProductRepository is not defined` ✅ (expected)
+
+## 🟢 GREEN — Code tối thiểu
+
+```javascript
+// repositories/product.repo.js
+class ProductRepository {
+  constructor(pool) { this.pool = pool; }
+
+  async findById(id) {
+    const { rows } = await this.pool.query(
+      'SELECT * FROM products WHERE id = $1 AND deleted_at IS NULL', [id]
+    );
+    return rows[0] || null;
+  }
+
+  async softDelete(id) {
+    const { rowCount } = await this.pool.query(
+      'UPDATE products SET deleted_at = NOW() WHERE id = $1', [id]
+    );
+    return rowCount > 0;
+  }
+}
+module.exports = { ProductRepository };
+```
+
+**Run → PASS** ✅
+
+## 🔵 REFACTOR — Clean theo constitution
+
+- Extracted SQL strings to constants
+- Added JSDoc comments
+- Verified: all tests still pass ✅
+- No-Go check: không đụng `db/pool.js` ✅

package/skills/implement/tdd-driver/gotchas.md

@@ -0,0 +1,34 @@
+# Gotchas — TDD Driver
+
+> Cập nhật liên tục khi Agent gặp edge case mới.
+> Mỗi lỗi lặp lại 2 lần → BẮT BUỘC thêm vào đây.
+
+---
+
+## Anti-Patterns
+
+1. ❌ **Implement trước, test sau** — Test sẽ bias theo code đã viết. "Green bar = proof" là ảo.
+   - **Hậu quả**: Test chỉ confirm code hoạt động, không confirm requirement.
+
+2. ❌ **Skip test vì "đơn giản"** — Bugs hay ẩn ở code đơn giản nhất.
+   - **VD**: Off-by-one, null check, empty array.
+
+3. ❌ **Test quá lớn** — 1 test cover 5 behaviors → khi fail, không biết cái nào sai.
+   - **Fix**: 1 test = 1 assertion = 1 behavior.
+
+4. ❌ **Mock quá nhiều** — Test full mocks → integration issues leak ra production.
+   - **Rule**: Mock external deps only. Business logic test real code.
+
+5. ❌ **Sửa test để pass** — Nếu test fail → fix CODE, KHÔNG fix test.
+   - **Exception**: Test implementation sai (sai mock setup, sai expected value) → fix test.
+
+6. ❌ **Quên concerns** — W-xxx carry-forward từ R.0 phải được resolve trong TDD cycle.
+   - **Check**: Đọc `_session.md` → search `W-` → resolve trong task assigned.
+
+7. ❌ **Test depend test khác** — Test 2 fail vì Test 1 setup shared state.
+   - **Fix**: Mỗi test PHẢI chạy standalone. Fresh state per test.
+
+8. ❌ **Test tên vô nghĩa** — `test1`, `testLogin` → khi fail, không biết test gì.
+   - **Fix**: `should_return_404_when_product_not_found`.
+
+9. ✅ **Nếu test PASS ngay ở RED** — Test SAI, không phải code ĐÚNG. Xem lại.

package/skills/metrics/metrics-collector/SKILL.md

@@ -0,0 +1,133 @@
+---
+name: metrics-collector
+version: "1.0"
+agent: metrics
+phase: M.0
+triggers:
+  - post_review
+  - manual_collection
+dependencies:
+  - _shared/constitution-reader
+input:
+  - .flow-state.yaml
+  - _session.md
+  - validation-report.md
+  - code-review-report.md (if exists)
+  - git log / timestamps
+output:
+  - metrics-report.md
+complexity: medium
+---
+
+# Metrics Collector Skill
+
+> **Purpose**: Thu thập và phân tích Agent performance metrics sau mỗi feature cycle.
+> **Khi nào dùng**: Sau khi R.1 (Review) hoàn tất — chạy `/v.metrics`.
+
+## Quick Reference
+
+```
+Input:  .flow-state.yaml + _session.md + validation/review reports
+Output: metrics-report.md
+Agent:  Metrics Agent
+Phase:  M.0 (post R.1)
+```
+
+## What This Skill Does
+
+1. **Collect Timeline Data** — Timestamps từ `.flow-state.yaml` phase_history
+2. **Count Iterations** — Bao nhiêu lần revision tại mỗi phase
+3. **Measure Quality** — First-pass approval rates, Master Check verdict
+4. **Analyze Bottlenecks** — Phase nào mất thời gian nhất
+5. **Generate Report** — Structured metrics-report.md
+
+## Metrics Categories
+
+### Per-Agent Metrics
+
+| Metric | Source | Formula |
+|--------|--------|---------|
+| **First-pass approval rate** | `.flow-state.yaml` phase_history | phases_passed_first_try / total_phases |
+| **Revision count** | `_session.md` execution trace | Count re-entries per phase |
+| **Processing time** | `.flow-state.yaml` timestamps | exited_at - entered_at per phase |
+| **Quality score** | `validation-report.md` | (PASS=100, WITH_CONCERNS=70, FAIL=0) |
+
+### Per-Pipeline Metrics
+
+| Metric | Source | Formula |
+|--------|--------|---------|
+| **Total cycle time** | `.flow-state.yaml` | last_phase.exited_at - first_phase.entered_at |
+| **Master Check pass rate** | `.flow-state.yaml` routing | PASS_count / total_check_count |
+| **Rework rate** | Phase history | phases_with_rework / total_phases |
+| **Concern resolution rate** | `_session.md` | concerns_resolved / concerns_raised |
+
+### Per-Feature Metrics
+
+| Metric | Source | Formula |
+|--------|--------|---------|
+| **Requirements count** | `spec.md` | Count REQ-xxx |
+| **Test coverage** | `test-cases.md` | tests_count / requirements_count |
+| **Spec change count** | `.flow-state.yaml` spec_version | Version number (v1.0 → v1.2 = 2 changes) |
+
+## How to Use
+
+### Step 1: Gather Data Sources
+
+```markdown
+Đọc các file sau từ feature directory:
+1. .flow-state.yaml → phase_history, artifacts, routing
+2. _session.md → execution trace, decisions, open questions
+3. validation-report.md → verdict, concerns
+4. spec.md → requirements count
+5. test-cases.md → test count
+```
+
+### Step 2: Calculate Metrics
+
+```markdown
+Cho mỗi phase trong phase_history:
+- Tính duration = exited_at - entered_at
+- Đếm số lần phase xuất hiện (revision count)
+- Kiểm tra verdict/approval status
+
+Tổng hợp:
+- Total cycle time
+- Bottleneck phase (longest duration)
+- First-pass vs revision phases
+```
+
+### Step 3: Generate Report
+
+```markdown
+## 📊 Metrics Report — [Feature Name]
+
+### Summary
+| Metric | Value |
+|--------|-------|
+| Total cycle time | X days |
+| Phases completed | Y/Z |
+| First-pass rate | XX% |
+| Rework rate | XX% |
+| Requirements | N |
+| Test cases | M |
+
+### Phase Timeline
+| Phase | Duration | Revisions | Verdict |
+|-------|----------|-----------|---------|
+| S.1 | 2h | 0 | approved |
+| B.0a | 1h | 0 | - |
+| R.0 | 30m | 1 | PASS_WITH_CONCERNS |
+| B.1 | 4h | 0 | - |
+| R.1 | 1h | 0 | APPROVED |
+
+### Recommendations
+1. [Based on bottleneck analysis]
+2. [Based on rework patterns]
+```
+
+## Gotchas
+
+1. **Missing timestamps** — Nếu `.flow-state.yaml` không có đủ timestamps, ước tính từ git log
+2. **Manual phases** — Một số phases (e.g., human approval) không có agent metrics
+3. **Cross-feature comparison** — Chỉ so sánh features cùng project_type (greenfield vs legacy)
+4. **First feature bias** — Feature đầu tiên thường chậm hơn do setup overhead

package/skills/metrics/metrics-collector/config.json

@@ -0,0 +1,16 @@
+{
+  "name": "metrics-collector",
+  "version": "1.0",
+  "description": "Collects and analyzes Agent performance metrics after each feature cycle",
+  "agent": "metrics",
+  "phase": ["M.0"],
+  "triggers": ["post_review", "manual"],
+  "dependencies": ["_shared/constitution-reader"],
+  "config": {
+    "metrics_categories": ["per_agent", "per_pipeline", "per_feature"],
+    "comparison_threshold": 5,
+    "report_format": "markdown",
+    "include_recommendations": true,
+    "benchmark_window": "last_5_features"
+  }
+}

package/skills/metrics/metrics-collector/examples/bad/incomplete-report.md

@@ -0,0 +1,48 @@
+# Metrics Collector — Bad Example
+
+> Ví dụ output THIẾU SÓT — không nên làm theo.
+
+## ❌ Bad Output — Quá sơ sài
+
+```markdown
+## Metrics Report
+
+Feature completed in 6 hours.
+All tests passed. 
+No major issues.
+```
+
+## Vấn đề
+
+| # | Lỗi | Giải thích |
+|---|------|-----------|
+| 1 | Không có per-phase breakdown | Không biết phase nào bottleneck |
+| 2 | Không có per-agent analysis | Không biết agent nào cần cải thiện |
+| 3 | Không có benchmark comparison | "6 hours" có tốt hay không? |
+| 4 | Không có trend data | Không biết đang cải thiện hay xuống |
+| 5 | Không có recommendations | Metrics vô dụng nếu không actionable |
+| 6 | Không track concerns | Concerns từ R.0 có resolved không? |
+
+## ❌ Bad Output — Chỉ liệt kê không phân tích
+
+```markdown
+## Metrics
+
+- S.1: 1.5h
+- B.0a: 45m
+- B.0b: 30m
+- R.0: 20m
+- B.1: 2.5h
+- R.1: 40m
+
+Total: 6.5h
+```
+
+## Vấn đề
+
+| # | Lỗi | Giải thích |
+|---|------|-----------|
+| 1 | Raw data không insight | Liệt kê số nhưng không phân tích |
+| 2 | Thiếu context | Không biết revision count, verdict |
+| 3 | Không có percentage | Không thấy proportions |
+| 4 | Không actionable | Không biết cần làm gì tiếp |

package/skills/metrics/metrics-collector/examples/good/full-metrics-report.md

@@ -0,0 +1,101 @@
+# Metrics Collector — Good Example
+
+> Ví dụ output chuẩn của Metrics Agent sau khi feature cycle hoàn tất.
+
+## Input Context
+
+**Feature:** 005-user-notifications
+**Pipeline:** S.1 → B.0a ∥ B.0b → R.0 → B.1 → R.1 → M.0
+**Outcome:** Deployed successfully
+
+## Output — metrics-report.md
+
+```markdown
+## 📊 Metrics Report — 005-user-notifications
+
+**Generated**: 2026-04-25 14:30
+**Pipeline**: Full Flow (with Dual BA)
+**Outcome**: ✅ Deployed
+
+### Summary
+
+| Metric | Value | Benchmark | Status |
+|--------|-------|-----------|--------|
+| Total cycle time | 6.5 hours | 4-8 hours | ✅ Normal |
+| Phases completed | 7/7 | 7/7 | ✅ Complete |
+| First-pass rate | 71% (5/7) | >60% | ✅ Good |
+| Rework rate | 29% (2/7) | <40% | ✅ Acceptable |
+| Requirements | 8 (3U + 2E + 2S + 1O) | — | — |
+| Test cases | 15 | — | 1.88 per REQ |
+| Spec changes | 2 (v1.0 → v1.2) | <3 | ✅ Normal |
+
+### Phase Timeline
+
+| Phase | Duration | Revisions | Verdict | Notes |
+|-------|----------|-----------|---------|-------|
+| S.1 Spec | 1.5h | 1 | approved v1.2 | Dual BA: 2 rounds to converge |
+| B.0a Plan | 45m | 0 | approved | Clean architecture applied |
+| B.0b Test | 30m | 0 | approved | 15 TCs (8 happy, 4 edge, 3 error) |
+| R.0 Check | 20m | 1 | PASS_WITH_CONCERNS | [W-001] Missing pagination spec |
+| B.1 Implement | 2.5h | 0 | — | TDD: all 15 tests pass |
+| R.1 Review | 40m | 0 | APPROVED_WITH_NOTES | 4 issues (0 critical, 1 major, 3 minor) |
+| M.0 Metrics | 15m | 0 | — | This report |
+
+### Per-Agent Analysis
+
+| Agent | First-pass? | Quality | Insight |
+|-------|:-----------:|---------|---------|
+| BA Agent | ❌ (rev 1) | Good | Spec v1.0 thiếu error paths → v1.2 after BA Critic |
+| Plan Agent | ✅ | Excellent | Components mapped 100% to REQs |
+| Test Agent | ✅ | Good | Coverage 1.88 TC/REQ — above target |
+| Master Check | ❌ (rev 1) | Good | Caught [W-001] → resolved in B.1 |
+| Implement Agent | ✅ | Excellent | TDD discipline maintained |
+| Review Agent | ✅ | Good | Found 1 major (missing index) |
+
+### Concern Resolution
+
+| Concern | Source | Resolution | Verified |
+|---------|--------|------------|:--------:|
+| [W-001] Missing pagination | R.0 | Added pagination in B.1 (limit=20, offset) | ✅ R.1 |
+
+### Bottleneck Analysis
+
+```
+Phase Duration Distribution:
+S.1  ████████████████████ 1.5h (23%)  ← Longest — Dual BA overhead
+B.1  ██████████████████████████████████ 2.5h (38%)  ← Expected — most work
+B.0a ██████████ 45m (12%)
+R.1  █████████ 40m (10%)
+B.0b ██████ 30m (8%)
+R.0  ████ 20m (5%)
+M.0  ██ 15m (4%)
+```
+
+**Bottleneck**: S.1 (23%) — Dual BA review added ~45m overhead.
+**Action**: Consider using single BA for features with <5 requirements.
+
+### Recommendations
+
+1. **BA Agent**: Pre-define error paths in initial spec to reduce Dual BA rounds
+2. **Master Check**: Pagination should be in standard checklist for list APIs
+3. **Review Agent**: Database index check should be automated (add to constitution)
+4. **Overall**: Cycle time within normal range — no urgent improvements needed
+
+### Trend (last 3 features)
+
+| Feature | Cycle Time | First-pass | Rework | REQs |
+|---------|-----------|-----------|--------|------|
+| 003-export | 4.0h | 83% | 17% | 5 |
+| 004-dashboard | 7.5h | 57% | 43% | 12 |
+| **005-notifications** | **6.5h** | **71%** | **29%** | **8** |
+| **Average** | **6.0h** | **70%** | **30%** | **8.3** |
+```
+
+## Why This Is Good
+
+1. ✅ **Quantitative** — tất cả metrics đều có số liệu cụ thể
+2. ✅ **Benchmarked** — so sánh với benchmark biết normal hay không
+3. ✅ **Actionable** — recommendations cụ thể cho từng agent
+4. ✅ **Visual** — bar chart cho bottleneck analysis
+5. ✅ **Trending** — so sánh với features trước để thấy xu hướng
+6. ✅ **Concern tracking** — chứng minh concerns đã resolved

package/skills/metrics/metrics-collector/gotchas.md

@@ -0,0 +1,26 @@
+# Metrics Collector — Gotchas & Common Mistakes
+
+## ⚠️ Gotcha 1: Vanity Metrics
+
+**Sai**: Focus vào "total features completed" — không phản ánh quality.  
+**Đúng**: Đo **first-pass approval rate** và **rework rate** — phản ánh thực sự Agent performance.
+
+## ⚠️ Gotcha 2: Missing Baseline
+
+**Sai**: Report metrics mà không có benchmark.  
+**Đúng**: So sánh với 5 features gần nhất. Feature đầu tiên là baseline, không phải target.
+
+## ⚠️ Gotcha 3: Ignoring Context
+
+**Sai**: So sánh cycle time giữa simple CRUD feature và complex integration feature.  
+**Đúng**: Normalize theo requirements count. Report requirements/hour thay vì absolute time.
+
+## ⚠️ Gotcha 4: Skipping Recommendations
+
+**Sai**: Chỉ report numbers, không đề xuất action.  
+**Đúng**: Mỗi metrics report PHẢI có ≥2 actionable recommendations dựa trên bottleneck analysis.
+
+## ⚠️ Gotcha 5: Over-optimization
+
+**Sai**: Mỗi feature đều thay đổi agent-config.md dựa trên metrics.  
+**Đúng**: Chỉ update agent config sau 5-10 features. Trends matter, not individual data points.

package/skills/operations/incident-runbook/SKILL.md

@@ -0,0 +1,167 @@
+---
+name: incident-runbook
+version: "1.0"
+agent: operations
+phase: O.0
+triggers:
+  - production_incident
+  - hotfix_flow
+  - monitoring_alert
+dependencies:
+  - _shared/context-reader
+input:
+  - incident report / alert
+  - monitoring data (logs, metrics)
+  - context.md
+  - spec.md (related feature)
+output:
+  - operations-report.md
+  - incident ticket (INC-XXX.md)
+complexity: high
+---
+
+# Incident Runbook Skill
+
+> **Purpose**: Structured incident response procedure cho production issues.
+> **Khi nào dùng**: Khi nhận alert/report từ production — chạy `/v.hotfix`.
+
+## Quick Reference
+
+```
+Input:  Incident report + monitoring data + context.md
+Output: operations-report.md + INC-XXX.md
+Agent:  Operations Agent
+Phase:  O.0
+```
+
+## What This Skill Does
+
+1. **Triage** — Phân loại severity (P1/P2/P3/P4) và assign response SLA
+2. **Investigate** — Root Cause Analysis (RCA) có cấu trúc
+3. **Contain** — Immediate mitigation steps
+4. **Fix** — Route to appropriate pipeline (hotfix/quickfix/standard)
+5. **Learn** — Post-mortem và lessons learned
+
+## Severity Classification
+
+| Level | Impact | Response SLA | Pipeline |
+|-------|--------|-------------|----------|
+| **P1** | Service down, data loss | < 15 min | `/v.hotfix` |
+| **P2** | Major feature broken | < 1 hour | `/v.hotfix` |
+| **P3** | Minor feature degraded | < 4 hours | `/v.quickfix` |
+| **P4** | Cosmetic, performance | Next sprint | Standard flow |
+
+## Incident Response Steps
+
+### Step 1: Triage (< 5 min)
+
+```markdown
+## Incident Triage
+
+**Alert Source**: [monitoring / user report / automated test]
+**Timestamp**: [when detected]
+**Impact Assessment**:
+  - Users affected: [all / segment / single]
+  - Feature affected: [feature name]
+  - Data integrity: [safe / at risk / compromised]
+
+**Severity**: P[1-4]
+**Response SLA**: [time]
+```
+
+### Step 2: Investigate (RCA)
+
+```markdown
+## Root Cause Analysis
+
+### What happened?
+[Description của incident]
+
+### Timeline
+| Time | Event |
+|------|-------|
+| T-0 | Description of the trigger event |
+| T+5m | First alert |
+| T+10m | Investigation started |
+
+### Root Cause
+[Direct cause → Contributing factors → Systemic issues]
+
+### Blast Radius
+- Files affected: [list]
+- Modules affected: [list]
+- No-Go Zones touched: [yes/no — CRITICAL if yes]
+```
+
+### Step 3: Contain
+
+```markdown
+## Immediate Mitigation
+
+### Actions Taken
+1. [Rollback / Feature flag / Rate limit / etc.]
+2. [Communication to users]
+3. [Monitoring enhanced]
+
+### Verification
+- [How to verify the containment is effective]
+- [Metrics to watch]
+```
+
+### Step 4: Fix (Route to Pipeline)
+
+```markdown
+## Fix Strategy
+
+### Route Decision
+- P1/P2 → `/v.hotfix` — skip spec, direct implement → review → deploy
+- P3 → `/v.quickfix` — minimal spec → implement → review
+- P4 → Standard flow with backlog ticket
+
+### Fix Details
+- Branch: hotfix/INC-XXX
+- Files to change: [list]
+- Test strategy: [unit test for fix + regression]
+- Deploy strategy: [immediate / canary / next release]
+```
+
+### Step 5: Learn (Post-mortem)
+
+```markdown
+## Post-Mortem
+
+### What went well
+1. [Detection was fast]
+2. [Communication was clear]
+
+### What went wrong
+1. [Missing test coverage for edge case]
+2. [No monitoring for this metric]
+
+### Action Items
+| # | Action | Owner | Due |
+|---|--------|-------|-----|
+| 1 | Add test for edge case X | Dev | Sprint N |
+| 2 | Add monitoring alert for Y | SRE | Sprint N |
+| 3 | Update context.md with new no-go zone | Tech Lead | Sprint N |
+
+### Knowledge Base Update
+→ File: .v-flow/knowledge/lessons-learned/INC-XXX.md
+```
+
+## Integration with V-Flow Pipeline
+
+```
+Alert → O.0 (Triage + RCA) → Route Decision:
+  ├── P1/P2 → B.1 (Hotfix) → R.1 (Review) → Deploy → O.0 (Verify)
+  ├── P3   → S.1 (Quick Spec) → B.1 → R.1 → Deploy
+  └── P4   → Backlog → Standard Pipeline
+```
+
+## Gotchas
+
+1. **Đừng debug production trên production** — Reproduce locally trước khi fix
+2. **No-Go Zones** — Nếu root cause nằm trong no-go zone, PHẢI escalate tới Tech Lead
+3. **Hotfix ≠ Skip quality** — Hotfix vẫn cần test và review, chỉ skip spec
+4. **Communication** — Cập nhật stakeholders tại mỗi bước, không chỉ khi fix xong
+5. **Lessons learned** — Post-mortem PHẢI được ghi vào knowledge base, không chỉ chat/email

package/skills/operations/incident-runbook/config.json

@@ -0,0 +1,21 @@
+{
+  "name": "incident-runbook",
+  "version": "1.0",
+  "description": "Structured incident response procedure for production issues",
+  "agent": "operations",
+  "phase": ["O.0"],
+  "triggers": ["production_incident", "monitoring_alert", "hotfix_flow"],
+  "dependencies": ["_shared/context-reader"],
+  "config": {
+    "severity_levels": {
+      "P1": { "sla_minutes": 15, "route": "hotfix" },
+      "P2": { "sla_minutes": 60, "route": "hotfix" },
+      "P3": { "sla_minutes": 240, "route": "quickfix" },
+      "P4": { "sla_minutes": null, "route": "standard" }
+    },
+    "require_postmortem": true,
+    "incident_dir": ".v-flow/incidents",
+    "lessons_learned_dir": ".v-flow/knowledge/lessons-learned",
+    "report_format": "markdown"
+  }
+}