@buivietphi/skill-mobile-mt 2.0.1 → 2.2.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (18) hide show
  1. package/AGENTS.md +96 -40
  2. package/README.md +77 -40
  3. package/SKILL.md +762 -54
  4. package/package.json +1 -1
  5. package/shared/bug-detection.md +411 -27
  6. package/shared/code-generation-templates.md +656 -0
  7. package/shared/code-review.md +899 -37
  8. package/shared/complex-ui-patterns.md +526 -0
  9. package/shared/data-flow-patterns.md +422 -0
  10. package/shared/debugging-intelligence.md +787 -0
  11. package/shared/error-handling.md +394 -0
  12. package/shared/i18n-localization.md +426 -0
  13. package/shared/intent-analysis.md +473 -0
  14. package/shared/navigation-patterns.md +375 -0
  15. package/shared/prompt-engineering.md +176 -20
  16. package/shared/spec-to-code.md +293 -0
  17. package/shared/storage-patterns.md +312 -0
  18. package/shared/testing-patterns.md +428 -0
package/AGENTS.md CHANGED
@@ -27,45 +27,56 @@
27
27
 
28
28
  ```
29
29
  skill-mobile-mt/
30
- ├── SKILL.md ← Entry point. Always load first.
30
+ ├── SKILL.md ← Entry point. Always load first. (~25,150 tokens)
31
31
  ├── AGENTS.md ← This file. Multi-agent config.
32
32
  │
33
33
  ├── react-native/
34
- │ └── react-native.md ← RN + Expo patterns (5,108 tokens)
34
+ │ └── react-native.md ← RN + Expo patterns (~5,840 tokens)
35
35
  │
36
36
  ├── flutter/
37
- │ └── flutter.md ← Flutter + Dart 3.x patterns (2,100 tokens)
37
+ │ └── flutter.md ← Flutter + Dart 3.x patterns (~2,350 tokens)
38
38
  │
39
39
  ├── ios/
40
- │ └── ios-native.md ← Swift + UIKit/SwiftUI patterns (1,452 tokens)
40
+ │ └── ios-native.md ← Swift + UIKit/SwiftUI patterns (~1,660 tokens)
41
41
  │
42
42
  ├── android/
43
- │ └── android-native.md ← Kotlin + Compose + Java legacy patterns (4,400 tokens)
43
+ │ └── android-native.md ← Kotlin + Compose + Java legacy patterns (~4,370 tokens)
44
44
  │
45
45
  └── shared/
46
46
  │
47
47
  ├── ── CORE (always load) ──────────────────────────────────
48
- ├── code-review.md ← Senior review checklist (865 tokens)
49
- ├── bug-detection.md ← Auto bug scanner (499 tokens)
50
- ├── prompt-engineering.md ← Auto-think + XML templates + advanced patterns (5,200 tokens)
48
+ ├── code-review.md ← 12-category PR-level review + grounded review + 25 production crash patterns (all platforms) (~11,600 tokens)
49
+ ├── bug-detection.md ← Intelligent bug scanner + git-aware Phase 0 + error classification + stack trace parser (~7,290 tokens)
50
+ ├── prompt-engineering.md ← Auto-think + XML templates + advanced patterns (~7,440 tokens)
51
51
  │
52
52
  ├── ── ON-DEMAND (load by task) ────────────────────────────
53
- ├── error-recovery.md ← 16 build/runtime error fixes (2,435 tokens)
54
- ├── document-analysis.md ← Parse images/PDFs/DOCX → code (1,200 tokens)
55
- ├── anti-patterns.md ← PII, cardinality, payload detection (2,800 tokens)
56
- ├── performance-prediction.md ← Frame budget, FPS prediction (1,500 tokens)
57
- ├── platform-excellence.md ← iOS 18+ vs Android 15+ UX + HIG (2,200 tokens)
58
- ├── version-management.md ← SDK compat matrix + release testing (3,500 tokens)
59
- ├── observability.md ← Sessions as 4th pillar (3,000 tokens)
60
- ├── architecture-intelligence.md ← Patterns from 30+ production repos (4,500 tokens)
61
- ├── common-pitfalls.md ← Known issue patterns (1,160 tokens)
62
- ├── release-checklist.md ← App Store/Play Store checklist (587 tokens)
53
+ ├── error-recovery.md ← 16 build/runtime error fixes (~2,780 tokens)
54
+ ├── document-analysis.md ← Parse images/PDFs/DOCX → code (~1,520 tokens)
55
+ ├── anti-patterns.md ← PII, cardinality, payload detection (~2,910 tokens)
56
+ ├── performance-prediction.md ← Frame budget, FPS prediction (~1,310 tokens)
57
+ ├── platform-excellence.md ← iOS 18+ vs Android 15+ UX + HIG (~1,890 tokens)
58
+ ├── version-management.md ← SDK compat matrix + release testing (~3,660 tokens)
59
+ ├── observability.md ← Sessions as 4th pillar (~5,750 tokens)
60
+ ├── architecture-intelligence.md ← Patterns from 30+ production repos (~3,860 tokens)
61
+ ├── common-pitfalls.md ← Known issue patterns (~1,330 tokens)
62
+ ├── release-checklist.md ← App Store/Play Store checklist (~670 tokens)
63
63
  │
64
- ├── offline-first.md ← Local-first + sync patterns (2,566 tokens)
65
- ├── testing-strategy.md ← Detox + Maestro + XCUITest + Espresso E2E (2,200 tokens)
66
- ├── ci-cd.md ← GitHub Actions CI templates (2,500 tokens)
67
- ├── ai-dlc-workflow.md ← AI-DLC structured workflow for complex features (2,500 tokens)
68
- ├── ui-ux-mobile.md ← Design system, screen templates, touch, navigation, a11y (5,500 tokens)
64
+ ├── offline-first.md ← Local-first + sync patterns (~2,930 tokens)
65
+ ├── testing-strategy.md ← Detox + Maestro + XCUITest + Espresso E2E (~2,500 tokens)
66
+ ├── ci-cd.md ← GitHub Actions CI templates (~2,830 tokens)
67
+ ├── ai-dlc-workflow.md ← AI-DLC structured workflow for complex features (~1,950 tokens)
68
+ ├── ui-ux-mobile.md ← Design system, screen templates, touch, navigation, a11y (~6,910 tokens)
69
+ ├── storage-patterns.md ← MMKV / SecureStore / SQLite / WatermelonDB / Keychain (~2,760 tokens)
70
+ ├── i18n-localization.md ← i18next / slang / .xcstrings / strings.xml / RTL / date format (~3,140 tokens)
71
+ ├── debugging-intelligence.md ← 45+ error patterns (RN+Flutter+iOS+Android) + git-aware debugging + search strategies (~8,730 tokens)
72
+ ├── intent-analysis.md ← Task extraction, scope clarification, intent understanding, spec analysis (~6,250 tokens)
73
+ ├── code-generation-templates.md ← Zustand/Redux/Riverpod, API client, forms, type generation (~5,370 tokens)
74
+ ├── spec-to-code.md ← Spec → dependency graph → file plan → implementation (~2,640 tokens)
75
+ ├── navigation-patterns.md ← Auth flow, deep links, modals, tabs, push, permissions (~2,900 tokens)
76
+ ├── complex-ui-patterns.md ← Carousel, gestures, keyboard, responsive, dark mode, a11y (~3,810 tokens)
77
+ ├── data-flow-patterns.md ← Pagination, optimistic updates, cache, WebSocket, offline queue (~3,250 tokens)
78
+ ├── error-handling.md ← Error hierarchy, retry, error boundary, user messages (~2,960 tokens)
79
+ ├── testing-patterns.md ← Component tests, hook tests, factories, snapshots (~3,410 tokens)
69
80
  │
70
81
  ├── ── TEMPLATES (copy to your project) ────────────────────
71
82
  ├── claude-md-template.md ← CLAUDE.md for Claude Code (copy to project root)
@@ -73,8 +84,13 @@ skill-mobile-mt/
73
84
  ```
74
85
 
75
86
  **Token totals:**
76
- - Smart load (1 platform + core shared): **~42,000 tokens** (32.8% of 128K)
77
- - Full load (all files): **~79,000 tokens** (61.7% of 128K)
87
+ - Smart load (SKILL.md + 1 platform + core shared): **~53,350 – 57,350 tokens** (42% – 45% of 128K)
88
+ - Full load (all files): **~166,380 tokens** (exceeds 128K — use smart load for 128K models)
89
+ - Full load fits within **200K context** (83% of 200K)
90
+ - On-demand files (NOT loaded unless triggered): debugging-intelligence, storage-patterns, i18n-localization, observability, architecture-intelligence, etc.
91
+ - code-review.md ~11,600 tokens — 12-category PR-level review + 25 production crash patterns (all 4 platforms)
92
+ - bug-detection.md ~7,290 tokens — git-aware Phase 0, error classification, stack trace parsing (all 4 platforms)
93
+ - debugging-intelligence.md ~8,730 tokens — 45+ error patterns across RN, Flutter, iOS Swift, Android Kotlin
78
94
 
79
95
  ---
80
96
 
@@ -86,13 +102,13 @@ skill-mobile-mt/
86
102
 
87
103
  **Loads automatically:**
88
104
  ```
89
- SKILL.md (~15,800 tokens)
90
- + 1 platform file (~1,580–5,730 tokens depending on platform)
91
- + shared/code-review.md (~1,500 tokens)
92
- + shared/bug-detection.md (~800 tokens)
93
- + shared/prompt-engineering.md (~5,200 tokens)
105
+ SKILL.md (~25,150 tokens)
106
+ + 1 platform file (~1,660–5,840 tokens depending on platform)
107
+ + shared/code-review.md (~11,600 tokens)
108
+ + shared/bug-detection.md (~7,290 tokens)
109
+ + shared/prompt-engineering.md (~7,440 tokens)
94
110
  ─────────────────────────────────────────────────
95
- ≈ 42,000 tokens total (estimated)
111
+ ≈ 53,350 – 57,350 tokens total (42%–45% of 128K)
96
112
  ```
97
113
 
98
114
  **Use case:** Regular coding, new features, code review. Covers 90% of daily work.
@@ -106,6 +122,7 @@ The agent reads the task, then decides which extra file to load:
106
122
  | Task the user asks for | File loaded |
107
123
  |------------------------|-------------|
108
124
  | "Fix this crash / build error" | `shared/error-recovery.md` |
125
+ | "Complex bug / long stack trace / investigate issue" | `shared/debugging-intelligence.md` |
109
126
  | "Read this screenshot / PDF / DOCX" | `shared/document-analysis.md` |
110
127
  | "Add analytics / logging / crash tracking" | `shared/anti-patterns.md` + `shared/observability.md` |
111
128
  | "Build a FlatList / animation" | `shared/performance-prediction.md` |
@@ -117,8 +134,24 @@ The agent reads the task, then decides which extra file to load:
117
134
  | "Setup CI/CD / GitHub Actions" | `shared/ci-cd.md` |
118
135
  | "Big feature / multi-screen" | `shared/ai-dlc-workflow.md` |
119
136
  | "Create/design screen / demo UI" | `shared/ui-ux-mobile.md` |
120
-
121
- **Load cost:** +500 to +3,500 tokens per on-demand file.
137
+ | "Storage / MMKV / SecureStore / save data" | `shared/storage-patterns.md` |
138
+ | "i18n / multi-language / translation / RTL" | `shared/i18n-localization.md` |
139
+ | "Review PR / review code / accessibility check" | `shared/code-review.md` (already loaded) + `shared/anti-patterns.md` |
140
+ | "Fix multiple / fix A then B / several places" | `shared/intent-analysis.md` (Task Extraction Protocol) |
141
+ | "Make it better / fix everything / vague request" | `shared/intent-analysis.md` (Scope Clarification Protocol) |
142
+ | "It's slow / doesn't work / non-technical description" | `shared/intent-analysis.md` (Intent Understanding Protocol) |
143
+ | "Build X like other apps / vague feature spec" | `shared/intent-analysis.md` (Spec Analysis Protocol) |
144
+ | "URGENT / production down / deadline / blocker" | `shared/intent-analysis.md` (Priority Detection) |
145
+ | "Build from spec / implement requirements" | `shared/spec-to-code.md` |
146
+ | "Setup state / Zustand / Redux / API client / forms" | `shared/code-generation-templates.md` |
147
+ | "Auth flow / deep links / modals / tabs / permissions" | `shared/navigation-patterns.md` |
148
+ | "Carousel / gestures / responsive / dark mode / a11y" | `shared/complex-ui-patterns.md` |
149
+ | "Pagination / optimistic / cache / WebSocket / offline" | `shared/data-flow-patterns.md` |
150
+ | "Error handling / retry / error boundary / toast" | `shared/error-handling.md` |
151
+ | "Component tests / unit tests / mock / factory" | `shared/testing-patterns.md` |
152
+ | "Compare options / best approach / upgrade vs stay" | SKILL.md Decision Matrix Protocol (already loaded) |
153
+
154
+ **Load cost:** +600 to +5,900 tokens per on-demand file.
122
155
 
123
156
  ---
124
157
 
@@ -126,7 +159,7 @@ The agent reads the task, then decides which extra file to load:
126
159
 
127
160
  **No automatic trigger.** Full load happens when the AI reads every file without being selective — either because it's over-eager, or because the user explicitly asks for it.
128
161
 
129
- **Total:** ~72,300 tokens (56.5% of 128K, 36% of 200K)
162
+ **Total:** ~100,880 tokens (78.8% of 128K, 50.4% of 200K)
130
163
 
131
164
  **How it actually works:**
132
165
  - `@skill-mobile-mt` only injects SKILL.md into context
@@ -151,7 +184,7 @@ The agent reads the task, then decides which extra file to load:
151
184
  ```yaml
152
185
  skill:
153
186
  name: skill-mobile-mt
154
- version: "1.4.2"
187
+ version: "2.2.0"
155
188
  author: buivietphi
156
189
  category: engineering
157
190
  tags:
@@ -218,9 +251,10 @@ skill:
218
251
  java: ".java files in app/src/"
219
252
 
220
253
  context_budget:
221
- max_tokens: 77000
222
- smart_load_tokens: 40900
223
- savings: "~47%"
254
+ max_tokens: 166380
255
+ smart_load_tokens: 57350
256
+ fits_128k: "smart load only"
257
+ fits_200k: "full load (83%)"
224
258
  ```
225
259
 
226
260
  ---
@@ -266,6 +300,17 @@ Every agent MUST follow this loading sequence:
266
300
  - shared/observability.md (when adding logging, analytics, crash tracking)
267
301
  - shared/common-pitfalls.md (when encountering unfamiliar errors)
268
302
  - shared/release-checklist.md (when preparing for App Store/Play Store submission)
303
+ - shared/storage-patterns.md (when choosing or implementing local storage)
304
+ - shared/i18n-localization.md (when implementing multi-language or RTL)
305
+ - shared/debugging-intelligence.md (when investigating complex bugs or long stack traces)
306
+ - shared/intent-analysis.md (when input is multi-part, vague, non-technical, or ambiguous)
307
+ - shared/code-generation-templates.md (when setting up state management, API client, or forms)
308
+ - shared/spec-to-code.md (when building new feature from spec or requirements)
309
+ - shared/navigation-patterns.md (when implementing auth flow, deep links, modals, tabs, permissions)
310
+ - shared/complex-ui-patterns.md (when building carousel, gestures, responsive layout, dark mode, a11y)
311
+ - shared/data-flow-patterns.md (when implementing pagination, optimistic updates, cache, WebSocket)
312
+ - shared/error-handling.md (when implementing error handling, retry, error boundary)
313
+ - shared/testing-patterns.md (when writing component tests, hook tests, or setting up test factories)
269
314
 
270
315
  7. SKIP non-matching platform subfolders (saves ~66% context)
271
316
  ```
@@ -291,6 +336,17 @@ Priority 6 (ON-DEMAND): shared/testing-strategy.md — Detox + Maestro + XCUITes
291
336
  Priority 6 (ON-DEMAND): shared/ci-cd.md — GitHub Actions CI/CD templates
292
337
  Priority 6 (ON-DEMAND): shared/ai-dlc-workflow.md — AI-DLC structured workflow for complex features
293
338
  Priority 6 (ON-DEMAND): shared/ui-ux-mobile.md — Screen templates, design tokens, components, dark mode
339
+ Priority 6 (ON-DEMAND): shared/storage-patterns.md — MMKV, SecureStore, SQLite, WatermelonDB, Keychain
340
+ Priority 6 (ON-DEMAND): shared/i18n-localization.md — i18next, slang, .xcstrings, strings.xml, RTL
341
+ Priority 6 (ON-DEMAND): shared/intent-analysis.md — Task extraction, scope clarification, intent understanding, spec analysis
342
+ Priority 6 (ON-DEMAND): shared/code-generation-templates.md — Zustand/Redux/Riverpod, API client, forms, types
343
+ Priority 6 (ON-DEMAND): shared/spec-to-code.md — Spec → dependency graph → file plan → implementation
344
+ Priority 6 (ON-DEMAND): shared/navigation-patterns.md — Auth flow, deep links, modals, tabs, push, permissions
345
+ Priority 6 (ON-DEMAND): shared/complex-ui-patterns.md — Carousel, gestures, keyboard, responsive, dark mode, a11y
346
+ Priority 6 (ON-DEMAND): shared/data-flow-patterns.md — Pagination, optimistic updates, cache, WebSocket, offline queue
347
+ Priority 6 (ON-DEMAND): shared/error-handling.md — Error hierarchy, retry, error boundary, user messages
348
+ Priority 6 (ON-DEMAND): shared/testing-patterns.md — Component tests, hook tests, factories, snapshots
349
+ Priority 6 (ON-DEMAND): shared/debugging-intelligence.md — 30+ error patterns, git-aware debugging
294
350
  ```
295
351
 
296
352
  ---
@@ -469,10 +525,10 @@ npx @buivietphi/skill-mobile-mt --init all # → all files
469
525
  {
470
526
  "id": "skill-mobile-mt",
471
527
  "name": "skill-mobile-mt",
472
- "version": "2.0.0",
528
+ "version": "2.2.0",
473
529
  "author": "buivietphi",
474
530
  "category": "engineering",
475
- "description": "Master Senior Mobile Engineer. Patterns from 30+ production repos (200k+ GitHub stars) + research from top 53k+ star skill repos. Cardinal rules, self-critique loops, leverage pyramid, verification-first. React Native, Flutter, iOS, Android.",
531
+ "description": "Master Senior Mobile Engineer. Patterns from 30+ production repos (200k+ GitHub stars) + research from top 53k+ star skill repos. Cardinal rules, self-critique loops, leverage pyramid, verification-first, decision matrix, codebase scan strategy, grounded code review (anti-false-positive), git-aware debugging, task extraction protocol, multi-fix execution, UI fix protocol, completion re-check. React Native, Flutter, iOS, Android.",
476
532
  "risk": "low",
477
533
  "source": "buivietphi (MIT)",
478
534
  "platforms": ["react-native", "flutter", "ios", "android"],
package/README.md CHANGED
@@ -305,43 +305,55 @@ iOS only?
305
305
 
306
306
  | Scenario | Tokens | % of 128K | % of 200K |
307
307
  |----------|-------:|----------:|----------:|
308
- | SKILL.md only | ~15,800 | 12.3% | 7.9% |
309
- | + 1 platform + core shared/ | ~42,000 | 32.8% | 21.0% |
310
- | Cross-platform (RN/Flutter + iOS + Android) | ~58,000 | 45.3% | 29.0% |
311
- | All files loaded | ~79,000 | 61.7% | 39.5% |
312
- | **Smart load (recommended)** | **~42,000** | **32.8%** | **21.0%** |
308
+ | SKILL.md only | ~25,150 | 19.6% | 12.6% |
309
+ | **Smart load (recommended)** | **~53,350 – 57,350** | **42% – 45%** | **27% – 29%** |
310
+ | Cross-platform (RN+Flutter+iOS+Android) | ~95,350 | 74.5% | 47.7% |
311
+ | All files loaded | ~166,380 | 130.0% | 83.2% |
312
+ | | | *exceeds 128K* | *fits 200K* |
313
313
 
314
314
  ### Per-file token breakdown
315
315
 
316
316
  | File | Tokens |
317
317
  |------|-------:|
318
- | `SKILL.md` | 8,700 |
319
- | `AGENTS.md` | 1,600 |
320
- | `react-native/react-native.md` | 5,108 |
321
- | `flutter/flutter.md` | 2,100 |
322
- | `ios/ios-native.md` | 1,452 |
323
- | `android/android-native.md` | 4,400 |
324
- | `shared/code-review.md` | 865 |
325
- | `shared/bug-detection.md` | 499 |
326
- | `shared/prompt-engineering.md` | 5,200 |
327
- | `shared/architecture-intelligence.md` | 4,500 |
328
- | `shared/common-pitfalls.md` | 1,160 |
329
- | `shared/error-recovery.md` | 2,435 |
330
- | `shared/document-analysis.md` | 1,200 |
331
- | `shared/release-checklist.md` | 587 |
332
- | `shared/anti-patterns.md` | 2,800 |
333
- | `shared/performance-prediction.md` | 1,500 |
334
- | `shared/platform-excellence.md` | 2,200 |
335
- | `shared/version-management.md` | 3,500 |
336
- | `shared/observability.md` | 3,000 |
337
- | `shared/offline-first.md` | 2,566 |
338
- | `shared/testing-strategy.md` | 2,200 |
339
- | `shared/ci-cd.md` | 2,500 |
340
- | `shared/ai-dlc-workflow.md` | 2,500 |
341
- | `shared/ui-ux-mobile.md` | 5,500 |
342
- | `shared/claude-md-template.md` | ~500 |
343
- | `shared/agent-rules-template.md` | ~2,500 |
344
- | **Total** | **~59,100** |
318
+ | `SKILL.md` | ~25,150 |
319
+ | `AGENTS.md` | ~6,440 |
320
+ | `react-native/react-native.md` | ~5,840 |
321
+ | `flutter/flutter.md` | ~2,350 |
322
+ | `ios/ios-native.md` | ~1,660 |
323
+ | `android/android-native.md` | ~4,370 |
324
+ | `shared/code-review.md` | ~11,600 |
325
+ | `shared/bug-detection.md` | ~7,290 |
326
+ | `shared/debugging-intelligence.md` | ~8,730 |
327
+ | `shared/prompt-engineering.md` | ~7,440 |
328
+ | `shared/architecture-intelligence.md` | ~3,860 |
329
+ | `shared/common-pitfalls.md` | ~1,330 |
330
+ | `shared/error-recovery.md` | ~2,780 |
331
+ | `shared/document-analysis.md` | ~1,520 |
332
+ | `shared/release-checklist.md` | ~670 |
333
+ | `shared/anti-patterns.md` | ~2,910 |
334
+ | `shared/performance-prediction.md` | ~1,310 |
335
+ | `shared/platform-excellence.md` | ~1,890 |
336
+ | `shared/version-management.md` | ~3,660 |
337
+ | `shared/observability.md` | ~5,750 |
338
+ | `shared/offline-first.md` | ~2,930 |
339
+ | `shared/testing-strategy.md` | ~2,500 |
340
+ | `shared/ci-cd.md` | ~2,830 |
341
+ | `shared/ai-dlc-workflow.md` | ~1,950 |
342
+ | `shared/ui-ux-mobile.md` | ~6,910 |
343
+ | `shared/storage-patterns.md` | ~2,760 |
344
+ | `shared/i18n-localization.md` | ~3,140 |
345
+ | `shared/intent-analysis.md` | ~6,250 |
346
+ | `shared/code-generation-templates.md` | ~5,370 |
347
+ | `shared/spec-to-code.md` | ~2,640 |
348
+ | `shared/navigation-patterns.md` | ~2,900 |
349
+ | `shared/complex-ui-patterns.md` | ~3,810 |
350
+ | `shared/data-flow-patterns.md` | ~3,250 |
351
+ | `shared/error-handling.md` | ~2,960 |
352
+ | `shared/testing-patterns.md` | ~3,410 |
353
+ | `shared/claude-md-template.md` | ~1,050 |
354
+ | `shared/agent-rules-template.md` | ~2,870 |
355
+ | `humanizer/humanizer-mobile.md` | ~2,630 |
356
+ | **Total** | **~166,380** |
345
357
 
346
358
  ## Installed Structure
347
359
 
@@ -360,7 +372,8 @@ iOS only?
360
372
  │ └── android-native.md Android Kotlin + Java patterns
361
373
  └── shared/
362
374
  ├── code-review.md Senior review checklist
363
- ├── bug-detection.md Auto bug scanner
375
+ ├── bug-detection.md Auto bug scanner (4 modes + git-aware)
376
+ ├── debugging-intelligence.md 30+ error patterns + root cause tracing
364
377
  ├── prompt-engineering.md Auto-think + prompt templates
365
378
  ├── architecture-intelligence.md Patterns from 30+ production repos
366
379
  ├── common-pitfalls.md Problem → Symptoms → Solution
@@ -373,6 +386,8 @@ iOS only?
373
386
  ├── observability.md Sessions as 4th pillar
374
387
  ├── release-checklist.md Pre-release verification
375
388
  ├── offline-first.md Local-first + sync patterns
389
+ ├── storage-patterns.md Storage selection matrix (AsyncStorage/MMKV/SQLite/etc.)
390
+ ├── i18n-localization.md Internationalization + RTL + locale patterns
376
391
  ├── testing-strategy.md Detox + Maestro + XCUITest + Espresso E2E
377
392
  ├── ci-cd.md GitHub Actions CI/CD templates
378
393
  ├── ai-dlc-workflow.md AI-DLC structured dev workflow
@@ -409,15 +424,37 @@ your-project/
409
424
  - Single responsibility per file (max 300 lines)
410
425
  - Dependency injection, no hardcoded singletons
411
426
 
412
- ### Code Review Protocol
413
- - Architecture, correctness, performance, security, platform checklists
414
- - Severity levels: Critical / High / Medium / Low
415
- - Auto-fail patterns (console.log in production, hardcoded secrets, empty catch blocks)
427
+ ### Code Review Protocol (12-category, PR-level)
428
+ - **PR-Level Review**: size check, single responsibility, test accompaniment, commit hygiene — BEFORE code review
429
+ - **12 categories**: architecture, correctness, boundary conditions, test quality, readability, performance, security (expanded), accessibility (WCAG 2.1), breaking changes, platform, documentation, i18n
430
+ - **Severity levels**: Critical / High / Medium / Low with structured output format
431
+ - **Auto-fail patterns**: 10 code + 4 PR-level + 7 AI-specific grounding checks
432
+ - **Review output**: PR-level summary → findings by severity → verdict (APPROVE / CHANGES REQUESTED)
433
+ - **Boundary conditions**: null, empty, off-by-one, numeric limits, unicode, timezone — dedicated dimension
434
+ - **Accessibility**: touch targets (44pt/48dp), contrast ratios, screen reader, font scaling, focus management
435
+ - **Breaking change detection**: public API, deep links, DB schema, push payload, analytics events
436
+ - **Security depth**: certificate pinning, JWT lifecycle, bridge security, biometric auth, dependency audit
437
+ - **Grounded review (anti-false-positive)**: verify every finding before flagging — check actual installed APIs, don't flag from memory, confidence levels (high/medium/low)
438
+ - **Practical usage review**: 25 production crash patterns — 5 cross-platform + 5 React Native + 5 Flutter + 5 iOS Swift + 5 Android Kotlin
439
+ - **Library-specific traps**: React Native, Flutter, iOS Swift, Android Kotlin — real gotchas per platform
440
+ - **Cross-platform examples**: before/after code for all 4 platforms (not just RN)
416
441
 
417
442
  ### Bug Detection Scanner
418
- - Crash risks, memory leaks, race conditions
419
- - Security vulnerabilities, performance issues, UX problems
420
- - Platform-specific detection rules
443
+ - 4 debug modes: Error Analysis, Fix Bug, Investigate, Diagnostic Scan
444
+ - Git-aware debugging: checks recent commits before deep-diving (auto-skips if no git)
445
+ - 4-phase fix protocol: Root Cause → Pattern Analysis → Single Hypothesis → Defense in Depth
446
+ - 45+ error pattern database with exact search strategies — RN (7) + Flutter (6) + iOS Swift (5) + Android Kotlin (6) + Network (4) + State (3) + Navigation (2) + Build (4) + Platform (3)
447
+ - Anti-rationalization table: catches self-deception during debugging
448
+
449
+ ### Decision Matrix Protocol
450
+ - Structured comparison format when multiple valid approaches exist
451
+ - Estimation protocol: classify effort XS/S/M/L/XL with risk assessment
452
+ - Migration/upgrade decision protocol: breaking changes analysis + impact scan
453
+
454
+ ### Codebase Scan Strategy
455
+ - 3 scan levels: Quick (~5 reads), Standard (~15 reads), Deep (~30+ reads)
456
+ - Monorepo strategy: focus target package, scan shared deps
457
+ - Multi-module strategy: JS/TS vs Native vs Cross-layer task detection
421
458
 
422
459
  ### Grounding Protocol (Anti-Hallucination)
423
460
  - **Source hierarchy**: project code > skill files > official docs > production repos > AI knowledge