ai-developer-skill-os 7.0.2 → 7.5.0

package/.qk-ai-skill-os/skills/qk-ui-audit/references/anti-slop-checklist.md

@@ -0,0 +1,136 @@
+# Anti-Slop Checklist — 57 Points
+> Referenced by: `qk-ui-audit/SKILL.md`
+> Purpose: Full inspection checklist for detecting generic AI-generated UI
+
+---
+
+## Category P1: Accessibility (15 pts)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 1 | All images have `alt` text (non-empty, descriptive) | 2 | grep `alt=""` or `alt={` in JSX |
+| 2 | Color contrast ratio ≥ 4.5:1 (text) / 3:1 (large text) | 3 | Check computed HSL values vs background |
+| 3 | All interactive elements have `aria-label` if no visible text | 2 | grep for `<button>` without text content |
+| 4 | Focus states are visible — `outline: none` not used without replacement | 2 | Check CSS for `outline: none` |
+| 5 | Keyboard navigation works (Tab order logical) | 2 | `tabIndex` only for non-interactive elements |
+| 6 | `role` attributes used for non-semantic elements (div acting as button) | 2 | grep `onClick` on non-button elements |
+| 7 | Form inputs have associated `<label>` (htmlFor) | 2 | grep `<input` without surrounding `<label` |
+
+---
+
+## Category P2: Layout & Spacing (15 pts)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 8 | No hardcoded pixel spacing — uses design token scale (4, 8, 16, 24...) | 3 | grep `padding: [0-9]px` or `margin: [0-9]px` |
+| 9 | CSS Grid or Flexbox used — no `position: absolute` for layout | 2 | grep `position: absolute` in layout context |
+| 10 | Consistent spacing unit (not mixing 5px, 7px, 11px random values) | 2 | Check spacing values across components |
+| 11 | No `div` soup — semantic HTML used (article, section, nav, main, aside) | 2 | grep excessive `<div>` nesting |
+| 12 | Component has defined max-width (not stretching to 100vw on desktop) | 2 | Check container max-width |
+| 13 | Whitespace feels intentional — content is not cramped | 2 | Visual check or spacing scale adherence |
+| 14 | No horizontal overflow on mobile (320px viewport) | 2 | Check for `overflow-x: hidden` band-aids |
+
+---
+
+## Category P3: Typography & Colors (15 pts)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 15 | Font families from DESIGN.md only (no Times New Roman, Arial default) | 3 | grep `font-family:` vs DESIGN.md |
+| 16 | Font sizes use token scale — no arbitrary px values | 2 | grep `font-size: [0-9]` raw values |
+| 17 | Colors use DESIGN.md tokens — no hardcoded hex (#ff0000) or rgb() | 3 | grep `#[0-9a-fA-F]{3,6}` in CSS/JSX |
+| 18 | No pure black (#000000) or pure white (#ffffff) — use near-black/near-white | 2 | grep `#000\|#000000\|#fff\|#ffffff` |
+| 19 | Color palette is harmonious — not "rainbow slop" (5+ unrelated colors) | 2 | Count distinct hues in palette |
+| 20 | Heading hierarchy correct — one `<h1>` per page, no skipping levels | 1 | grep heading tags per page |
+| 21 | Line height appropriate (1.4–1.7 for body, 1.1–1.3 for headings) | 2 | Check `line-height` values |
+
+---
+
+## Category P4: Interactions & Animations (15 pts)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 22 | Every button has hover state — color shift OR shadow change | 2 | grep `:hover` for button styles |
+| 23 | Every button has focus-visible state | 2 | grep `:focus-visible` |
+| 24 | Active/pressed state exists for clickable elements | 1 | grep `:active` |
+| 25 | Transition duration from DESIGN.md — not raw `0.3s` everywhere | 2 | grep `transition:` vs token values |
+| 26 | No janky instant state changes — all transitions ≥ 100ms, ≤ 400ms | 2 | grep transition durations |
+| 27 | Animations respect `prefers-reduced-motion` | 2 | grep `prefers-reduced-motion` in CSS |
+| 28 | Links have hover underline OR color change (not just cursor pointer) | 2 | grep `a:hover` styles |
+| 29 | Form validation feedback uses animation (shake, pulse) not just color | 2 | Check error state CSS |
+
+---
+
+## Category P5: Anti-Slop Detection (15 pts)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 30 | No Bootstrap/generic UI kit classes (btn-primary, card, badge) | 2 | grep `class="btn\|card\|badge\|modal"` |
+| 31 | No Tailwind utility spam (> 15 classes on single element) | 2 | grep extremely long className strings |
+| 32 | No Lorem Ipsum or placeholder text in production UI | 3 | grep `Lorem ipsum\|placeholder text` |
+| 33 | Color palette is NOT blue-gray default (slop signature) | 2 | Check if primary color is `hsl(210, *)` Tailwind default |
+| 34 | Layout is NOT a simple centered card on white background (slop template) | 2 | Structural check |
+| 35 | Gradient usage is intentional — not `linear-gradient(to bottom, #fff, #f0f0f0)` | 2 | grep gradient definitions |
+| 36 | No `box-shadow: 0 2px 4px rgba(0,0,0,0.1)` everywhere (Bootstrap shadow) | 2 | grep this exact shadow pattern |
+
+---
+
+## Category P6: Performance & Best Practices (10 pts)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 37 | Images have `loading="lazy"` for below-fold content | 2 | grep `<img` without `loading` |
+| 38 | No `!important` overrides (except CSS reset) | 1 | grep `!important` in component CSS |
+| 39 | No inline styles except dynamic values | 2 | grep `style={{` in non-dynamic contexts |
+| 40 | Component file ≤ 300 lines | 1 | Line count check |
+| 41 | No unused CSS classes or dead styles | 1 | Visual code scan |
+| 42 | SVG icons used instead of emoji for UI elements | 1 | grep emoji characters in JSX |
+| 43 | `key` prop used correctly in lists (not index as key for dynamic lists) | 2 | grep `.map(` + check key prop |
+
+---
+
+## Category BONUS: Mobile & Cross-Browser (5 pts extra)
+
+| # | Check | Points | How to Verify |
+|---|-------|--------|---------------|
+| 44 | Touch targets ≥ 44×44px on mobile | 1 | Check button/link dimensions |
+| 45 | No hover-only interactions (mobile has no hover) | 1 | Check if critical info only on hover |
+| 46 | Custom fonts have system-ui fallback | 1 | Check font-family fallback stack |
+| 47 | Input types correct: email, tel, number, date (triggers correct mobile keyboard) | 1 | grep `<input type=` |
+| 48 | No `position: fixed` elements blocking content on small screens | 1 | Check fixed/sticky headers/footers |
+
+---
+
+## Extended Checks (9 additional points for comprehensive audit)
+
+| # | Check | Category |
+|---|-------|---------|
+| 49 | Loading skeleton matches actual content shape (not generic gray bars) | UX |
+| 50 | Empty state has illustration + helpful action button (not just "No data") | UX |
+| 51 | Error state specifies the error — not generic "Something went wrong" | UX |
+| 52 | Success feedback is clear — confirmation message or visual indicator | UX |
+| 53 | Destructive actions (delete) require confirmation dialog | Safety |
+| 54 | Modals/dialogs have accessible `role="dialog"` + focus trap | A11y |
+| 55 | Scrollable areas have visible scroll indicator | UX |
+| 56 | Data tables have sortable columns if > 10 rows | UX |
+| 57 | Forms have autofocus on first input (modal forms) | UX |
+
+---
+
+## Scoring Summary
+
+| Category | Max Points |
+|----------|-----------|
+| P1 Accessibility | 15 |
+| P2 Layout & Spacing | 15 |
+| P3 Typography & Colors | 15 |
+| P4 Interactions & Animations | 15 |
+| P5 Anti-Slop Detection | 15 |
+| P6 Performance | 10 |
+| **Base Total** | **85** |
+| BONUS Mobile | +5 |
+| Extended (49–57) | +9 |
+| **Maximum Possible** | **99** |
+
+**Pass threshold: ≥ 76 / 85 base (90%) — PASS**
+**Fail: < 76 / 85 — requires redesign, not patch**

package/.qk-ai-skill-os/skills/qk-ui-system-builder/SKILL.md

@@ -0,0 +1,225 @@
+---
+name: qk-ui-system-builder
+category: frontend
+version: 7.5.0
+description: "Xây dựng Design System và token library từ DESIGN.md — không tự đặt ra token ngoài contract."
+platforms: [antigravity, claude-code, cursor, windsurf, kilo-code]
+execution_mode: deterministic
+cost: medium
+latency: medium
+risk: low
+side_effects: edit_files
+produces: [code, tokens]
+consumes: [design-md]
+token_budget:
+  max_files_read: 2
+  max_lines_per_read: 100
+  max_shell_commands: 0
+  stop_early: true
+exit_codes: [SUCCESS, BLOCKED, FAILED, PARTIAL]
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+# qk-ui-system-builder — Design System Constructor
+
+> **Language rule:** Code, identifiers, file names ? English. Explanations, summaries ? Vietnamese.
+
+---
+
+## Preconditions
+- [ ] `DESIGN.md` exists with color, spacing, typography tokens
+
+```
+On missing precondition:
+  EXIT: BLOCKED
+  Message: "DESIGN.md missing. Run qk-project-bootstrap to create one."
+```
+
+---
+
+## Scope
+- ✅ Extract tokens from DESIGN.md ONLY — never invent
+- ✅ Generate CSS custom properties (variables) or framework theme
+- ✅ Ensure all components are accessible (A11y)
+
+## Non-Goals
+- ❌ Invent tokens not in DESIGN.md (e.g., standard Tailwind defaults)
+- ❌ Create duplicate tokens for same visual value
+- ❌ Generate component library without token foundation
+
+---
+
+## Token Naming Convention
+
+```css
+/* Colors — HSL always */
+--color-[role]-[shade]: hsl(H, S%, L%);
+/* Example: */
+--color-primary-500: hsl(220, 80%, 55%);
+--color-surface-100: hsl(220, 15%, 98%);
+
+/* Spacing — scale-based */
+--space-[n]: [n * base]px;
+/* Example: */
+--space-1: 4px; --space-2: 8px; --space-4: 16px;
+
+/* Typography */
+--font-[family]: '[Name]', fallback;
+--font-size-[scale]: [value]rem;
+--font-weight-[name]: [value];
+
+/* Border Radius */
+--radius-[size]: [value]px;
+
+/* Shadows */
+--shadow-[level]: [value];
+
+/* Animation */
+--duration-[speed]: [value]ms;
+--ease-[type]: [cubic-bezier or keyword];
+
+/* Z-index */
+--z-[layer]: [value];  /* dropdown=100, modal=200, toast=300 */
+```
+
+---
+
+## Dark/Light Mode Pattern
+```css
+:root {
+  --color-background: hsl(0, 0%, 100%);
+  --color-text: hsl(220, 15%, 10%);
+}
+
+[data-theme="dark"] {
+  --color-background: hsl(220, 15%, 10%);
+  --color-text: hsl(220, 10%, 90%);
+}
+```
+
+---
+
+## Workflow
+
+### Phase 1 — Extract from DESIGN.md
+1. Read DESIGN.md → list all defined tokens
+2. Categorize: Colors / Spacing / Typography / Radius / Shadow / Animation
+
+### Phase 2 — Generate Token File
+1. Create `src/styles/tokens.css` (or equivalent)
+2. Apply naming convention above
+3. Add dark mode variants if DESIGN.md specifies
+
+### Phase 3 — Verify No Invented Tokens
+1. Cross-check: every generated token has source in DESIGN.md
+2. Flag any token that was "inferred" vs explicitly defined
+
+**Decision:**
+```
+IF all tokens sourced from DESIGN.md → EXIT: SUCCESS
+IF some tokens inferred (MEDIUM confidence) → EXIT: PARTIAL, list them
+```
+
+---
+
+
+---
+
+## Priority Order
+
+[Placeholder for Priority Order]
+
+
+---
+
+## Output Format
+
+[Placeholder for Output Format]
+
+
+---
+## Exit Codes
+| Code | Meaning | When |
+|------|---------|------|
+| SUCCESS | All tokens extracted from DESIGN.md, naming convention applied | All generated |
+| PARTIAL | Some tokens inferred (not explicitly in DESIGN.md) | Inferred values used |
+| BLOCKED | DESIGN.md missing or has no tokens | Cannot start |
+| FAILED | Token conflicts detected (duplicate values, naming collision) | Conflict |
+
+---
+
+## Confidence Model
+| Level | Condition | Action |
+|-------|-----------|--------|
+| HIGH | Token explicitly exists in DESIGN.md | Extract and map |
+| MEDIUM | Token value inferred or estimated based on related tokens | Note assumption |
+| LOW | Core scale missing entirely (e.g. no spacing definition) | EXIT: BLOCKED |
+
+---
+
+## Severity
+| Level | Definition | Example |
+|-------|-----------|---------|
+| CRITICAL | Hardcoded values injected without being tokens | Missing the point of the system |
+| HIGH | Token naming convention violated | `primary-color` instead of `--color-primary` |
+| MEDIUM | Dark mode not supported | Hardcoded white background |
+| LOW | Unused token generated | Extraneous variables |
+
+---
+
+## Evidence Format
+```
+[SEVERITY] src/styles/tokens.css:LINE
+Issue:      [specific violation or gap]
+Confidence: HIGH
+Fix:        [specific change]
+```
+
+---
+
+## Retry Policy
+```
+Token generation fails
+  └─ Check if naming collision or syntax error
+       ├─ Syntax error → fix CSS/SCSS format → retry
+       └─ Collision → deduplicate → retry
+            └─ Do NOT auto-retry more than 1 time
+```
+
+---
+
+## Escalation Rules
+```
+BLOCKED: DESIGN.md missing or empty
+Missing:
+  - Valid DESIGN.md with token specifications
+Questions:
+  1. Bạn có file design reference nào khác không?
+Recommended Assumptions:
+  - Generate default Tailwind-like scale if requested
+```
+
+---
+
+## Handoff Contract
+### Consumes
+```json
+{
+  "from": "user or project-bootstrap",
+  "required_fields": ["design_md_path"],
+  "optional_fields": ["css_framework", "output_format"]
+}
+```
+### Produces
+```json
+{
+  "to": "user or design-to-code",
+  "output_fields": ["token_file_path", "inferred_tokens", "exit_code"]
+}
+```
+
+---
+
+

package/.qk-ai-skill-os/skills/qk-validation-gate/SKILL.md

@@ -0,0 +1,361 @@
+---
+name: qk-validation-gate
+category: qa
+version: 7.5.0
+description: "Cổng kiểm tra chất lượng bắt buộc với ngưỡng pass/fail cụ thể — chặn đứng mọi mã nguồn lỗi."
+platforms: [antigravity, claude-code, cursor, windsurf, kilo-code]
+execution_mode: deterministic
+
+cost: medium
+latency: medium
+risk: low
+side_effects: run_commands
+produces: [report]
+consumes: [source-code]
+
+token_budget:
+  max_files_read: 3
+  max_lines_per_read: 100
+  max_shell_commands: 2
+  stop_early: false  # Must run all checks — cannot skip for token budget
+
+exit_codes: [SUCCESS, BLOCKED, FAILED, PARTIAL]
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+# qk-validation-gate — Quality Gate
+
+> **Language rule:** Code, identifiers, file names ? English. Explanations, summaries ? Vietnamese.
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Preconditions
+- [ ] Project has `package.json` with lint/test scripts (or equivalent for other runtimes)
+- [ ] Target scope is specified (specific files, or whole project)
+
+```
+On missing precondition:
+  EXIT: BLOCKED
+  Message: "Không tìm thấy script lint/test. Vui lòng chỉ định lệnh kiểm tra."
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Scope
+- ✅ Run lint, type-check, tests, security audit in sequence
+- ✅ Enforce pass/fail with concrete thresholds
+- ✅ Report results with structured Evidence Format
+
+## Non-Goals
+- ❌ Fix failing tests or lint errors (fixes go to qk-bug-resolution)
+- ❌ Skip any gate check unless user explicitly allows
+- ❌ Override failing thresholds with "ignore" flags
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Pass/Fail Thresholds (Non-Negotiable)
+
+```yaml
+lint:
+  errors: 0           # Zero tolerance for lint errors
+  warnings: allowed   # Warnings do not block
+
+typescript:
+  errors: 0           # Zero type errors
+
+tests:
+  coverage_minimum: 80%       # Overall coverage
+  unit_test_pass_rate: 100%   # All unit tests must pass
+  integration_test_pass_rate: 100%
+
+security:
+  critical_vulnerabilities: 0
+  high_vulnerabilities: 0
+  medium_vulnerabilities: 5  # Max allowed (review required)
+
+build:
+  must_succeed: true
+  max_warnings: 5
+  bundle_size_limit: 500KB    # Gzipped (if applicable)
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Priority Order
+
+| Priority | Gate | Fail = Block? | Skip Allowed? |
+|----------|------|---------------|---------------|
+| P1 | Security audit (`npm audit`) | YES | Never |
+| P2 | TypeScript / type check | YES | Never |
+| P3 | Lint (errors only) | YES | Never |
+| P4 | Unit tests | YES | Never |
+| P5 | Integration tests | YES | Only if no integration tests exist |
+| P6 | Coverage check | YES if < 80% | Never |
+| P7 | Build | YES | Never |
+| P8 | Bundle size | WARN only | Budget < 80% |
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Workflow
+
+### Phase 1 — Security Gate (P1)
+
+**Steps:**
+1. Run: `npm audit --audit-level=high` (or equivalent)
+2. Parse output for CRITICAL and HIGH
+
+**Decision:**
+```
+IF 0 critical + 0 high
+  → PASS P1 → go to Phase 2
+
+ELSE IF critical or high found
+  → EXIT: FAILED immediately
+  → Report: exact package name, CVE, severity
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+### Phase 2 — Type & Lint Gate (P2+P3)
+
+**Steps:**
+1. Run: `npm run typecheck` (or `tsc --noEmit`)
+2. Run: `npm run lint`
+3. Parse error counts
+
+**Decision:**
+```
+IF 0 type errors AND 0 lint errors
+  → PASS P2+P3 → go to Phase 3
+
+ELSE IF type errors > 0
+  → EXIT: FAILED — list each error with file:line
+
+ELSE IF lint errors > 0 (warnings OK)
+  → EXIT: FAILED — list errors
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+### Phase 3 — Test Gate (P4+P5+P6)
+
+**Steps:**
+1. Run: `npm test -- --coverage` (or equivalent)
+2. Parse: pass/fail counts, coverage percentage
+
+**Decision:**
+```
+IF all tests pass AND coverage ≥ 80%
+  → PASS P4+P5+P6 → go to Phase 4
+
+ELSE IF any test fails
+  → EXIT: FAILED — list failing test names
+
+ELSE IF coverage < 80%
+  → EXIT: FAILED — show coverage report (file-level)
+```
+
+**Retry Policy:**
+```
+Test fail
+  └─ Check if it's a flaky test (run once more)
+       ├─ PASS on retry → note "flaky test detected", continue
+       └─ FAIL on retry → EXIT: FAILED — not flaky, real failure
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+### Phase 4 — Build Gate (P7)
+
+**Steps:**
+1. Run: `npm run build`
+2. Check exit code + warnings count
+
+**Decision:**
+```
+IF build succeeds AND warnings ≤ 5
+  → EXIT: SUCCESS
+
+ELSE IF build fails
+  → EXIT: FAILED
+
+ELSE IF warnings > 5
+  → EXIT: PARTIAL — list warnings, continue to release if user confirms
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Confidence Model
+
+| Level | Condition | Action |
+|-------|-----------|--------|
+| HIGH | Command exit code 0/non-0, numeric output | Report directly |
+| MEDIUM | Inferred from partial output (truncated logs) | Note potential missing info |
+| LOW | Script not found or output format unknown | EXIT: BLOCKED — clarify commands |
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Severity
+
+| Level | Definition | Example |
+|-------|-----------|---------|
+| CRITICAL | Security vulnerability in production dependency | CVE in auth library |
+| HIGH | Type error or failing test | `Cannot read property of undefined` |
+| MEDIUM | Coverage below threshold | 67% coverage (threshold 80%) |
+| LOW | Build warning | Unused export |
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Evidence Format
+
+```
+[SEVERITY] Gate: [SECURITY|TYPE|LINT|TEST|COVERAGE|BUILD]
+Command:    [command that was run]
+Result:     [PASS | FAIL — exact output snippet]
+Threshold:  [what the threshold is]
+Actual:     [what was measured]
+Fix:        [where to look for the fix]
+```
+
+**Example:**
+```
+[HIGH] Gate: TEST
+Command:    npm test -- --coverage
+Result:     FAIL — 3 tests failed
+Threshold:  100% pass rate
+Actual:     47/50 tests passed
+Fix:        See failing tests: auth.test.ts:L45, user.test.ts:L23, order.test.ts:L89
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Escalation Rules
+
+```
+BLOCKED: Cannot run validation
+Missing:
+  - npm scripts (lint, test, build) in package.json
+  - OR test framework setup
+Questions:
+  1. Dùng lệnh gì để chạy lint? (e.g., eslint, biome, oxlint)
+  2. Dùng lệnh gì để chạy test? (e.g., vitest, jest, playwright)
+  3. Coverage threshold dự án quy định là bao nhiêu?
+Recommended Assumptions:
+  - Standard: npm run lint + npm test + npm run build
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Handoff Contract
+
+### Consumes
+```json
+{
+  "from": "user or qk-orchestrator",
+  "required_fields": ["project_root"],
+  "optional_fields": ["custom_commands", "custom_thresholds", "scope_files"]
+}
+```
+
+### Produces
+```json
+{
+  "to": "qk-production-release (if all pass)",
+  "output_fields": ["gate_results", "coverage_percent", "violations_list", "exit_code"]
+}
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+## Output Format
+
+```
+🔒 Validation Gate Report
+─────────────────────────────────────────────────
+Project:    [path]
+
+Gate Results:
+  P1 Security:    [✅ PASS | ❌ FAIL — N critical, N high]
+  P2 TypeScript:  [✅ PASS | ❌ FAIL — N errors]
+  P3 Lint:        [✅ PASS | ❌ FAIL — N errors (N warnings)]
+  P4 Unit Tests:  [✅ PASS | ❌ FAIL — N/N passed]
+  P5 Integration: [✅ PASS | ❌ FAIL | ⏭️ SKIPPED]
+  P6 Coverage:    [✅ PASS (N%) | ❌ FAIL (N% < 80%)]
+  P7 Build:       [✅ PASS | ❌ FAIL]
+
+Failures (action required):
+  [SEVERITY] Gate — detail — Fix hint
+
+Exit Code:   [SUCCESS | PARTIAL | BLOCKED | FAILED]
+```
+
+skill_version: 7.5.0
+runtime_version: 1
+schema_version: 2
+---
+
+
+---
+
+## Retry Policy
+
+[Placeholder for Retry Policy]
+
+
+---
+## Exit Codes
+
+| Code | Meaning | When |
+|------|---------|------|
+| SUCCESS | All gates passed, thresholds met | Safe to proceed to release |
+| PARTIAL | Build pass, but warnings > 5 OR integration skipped | User confirmation required |
+| BLOCKED | Cannot run — missing scripts or config | Setup required |
+| FAILED | Any gate failed at threshold | Do NOT release — fix first |
+
+---
+
+