@buivietphi/skill-mobile-mt 1.4.0 → 1.4.1

package/AGENTS.md

@@ -62,7 +62,6 @@ skill-mobile-mt/
     ├── release-checklist.md         ← App Store/Play Store checklist (587 tokens)
     │
     ├── offline-first.md             ← Local-first + sync patterns (2,566 tokens)
-    ├── on-device-ai.md              ← Core ML / TFLite / llama.cpp patterns (700 tokens)
     │
     ├── ── TEMPLATES (copy to your project) ────────────────────
     ├── claude-md-template.md        ← CLAUDE.md for Claude Code (copy to project root)
@@ -110,7 +109,6 @@ The agent reads the task, then decides which extra file to load:
 | "Install this package / upgrade SDK" | `shared/version-management.md` |
 | "Prepare for App Store / Play Store" | `shared/release-checklist.md` |
 | "Weird issue, not sure why" | `shared/common-pitfalls.md` |
-| "On-device AI / ML / inference" | `shared/on-device-ai.md` |
 
 **Load cost:** +500 to +3,500 tokens per on-demand file.
 
@@ -178,7 +176,6 @@ skill:
         # - shared/document-analysis.md
         # - shared/release-checklist.md
         # - shared/common-pitfalls.md
-        # - shared/on-device-ai.md
 
     project:
       description: "Read current project, adapt to its framework and conventions"
@@ -261,7 +258,6 @@ Every agent MUST follow this loading sequence:
    - shared/observability.md          (when adding logging, analytics, crash tracking)
    - shared/common-pitfalls.md        (when encountering unfamiliar errors)
    - shared/release-checklist.md      (when preparing for App Store/Play Store submission)
-   - shared/on-device-ai.md           (when adding Core ML / TFLite / on-device inference)
 
 7. SKIP non-matching platform subfolders (saves ~66% context)
 ```
@@ -283,7 +279,6 @@ Priority 6 (ON-DEMAND): shared/observability.md — Sessions as 4th pillar
 Priority 6 (ON-DEMAND): shared/document-analysis.md — Parse images/PDFs → code
 Priority 6 (ON-DEMAND): shared/release-checklist.md — Pre-release verification
 Priority 6 (ON-DEMAND): shared/common-pitfalls.md — Known issue patterns
-Priority 6 (ON-DEMAND): shared/on-device-ai.md — Core ML / TFLite / llama.cpp
 ```
 
 ---

package/README.md

@@ -259,10 +259,9 @@ iOS only?
 | `shared/version-management.md` | 3,500 |
 | `shared/observability.md` | 3,000 |
 | `shared/offline-first.md` | 2,566 |
-| `shared/on-device-ai.md` | 700 |
 | `shared/claude-md-template.md` | ~500 |
 | `shared/agent-rules-template.md` | ~2,500 |
-| **Total** | **~49,500** |
+| **Total** | **~48,800** |
 
 ## Installed Structure
 
@@ -294,7 +293,6 @@ iOS only?
         ├── observability.md           Sessions as 4th pillar
         ├── release-checklist.md       Pre-release verification
         ├── offline-first.md           Local-first + sync patterns
-        ├── on-device-ai.md            Core ML / TFLite / llama.cpp
         ├── claude-md-template.md      CLAUDE.md template for projects
         └── agent-rules-template.md    Rules templates for all agents
 ```
@@ -390,7 +388,6 @@ your-project/
 - **Anti-Pattern Detection** (`anti-patterns.md`): Detect PII leaks (CRITICAL), high cardinality tags, unbounded payloads, unstructured logs, sync telemetry on main thread — with auto-fix suggestions
 - **Performance Prediction** (`performance-prediction.md`): Calculate frame budget, FlatList bridge calls, and memory usage BEFORE writing code. Example: `50 items × 3 bridge calls × 0.3ms = 45ms/frame → 22 FPS ❌ JANK`
 - **Platform Excellence** (`platform-excellence.md`): iOS 18+ vs Android 15+ native UX standards — navigation patterns, typography, haptic feedback types, permission timing, ratings prompt flow, Live Activities/Dynamic Island, performance targets (cold start < 1s iOS, < 1.5s Android)
-- **On-Device AI** (`on-device-ai.md`): Decision matrix (API vs on-device), Core ML (iOS), ML Kit + MediaPipe (Android), llama.cpp cross-platform, TFLite Flutter, React Native ML Kit — with performance rules and model size guidance
 - **Version Management** (`version-management.md`): Full SDK compatibility matrix for RN 0.73-0.76, Expo 50-52, Flutter 3.22-3.27, iOS 16-18, Android 13-15. Check SDK compat BEFORE `npm install`. Release-mode testing protocol.
 - **Observability** (`observability.md`): Sessions as the 4th pillar (Metrics + Logs + Traces + **Sessions**). Session lifecycle, enrichment API, unified instrumentation stack, correlation queries. Every event carries `session_id` for full user journey reconstruction.
 

package/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: skill-mobile-mt
 description: "Master Senior Mobile Engineer. Patterns from 30+ production repos (200k+ GitHub stars: Ignite, Expensify, Mattermost, Immich, AppFlowy, Now in Android, TCA). Use when: building mobile features, fixing mobile bugs, reviewing mobile code, mobile architecture, React Native, Flutter, iOS Swift, Android Kotlin, mobile performance, mobile security audit, mobile code review, app release. Two modes: (1) default = pre-built production patterns, (2) 'project' = reads current project and adapts."
-version: "1.4.0"
+version: "1.4.1"
 author: buivietphi
 priority: high
 user-invocable: true
@@ -130,9 +130,6 @@ USER REQUEST                    → ACTION (Read tool required)
 "Offline / cache / sync"        → Read: shared/offline-first.md
                                   then: implement local-first architecture
 
-"On-device AI / ML / inference" → Read: shared/on-device-ai.md
-                                  then: choose Core ML / TFLite / llama.cpp per platform
-
 ```
 
 **⛔ NEVER start coding without identifying the task type first.**

package/bin/install.mjs

@@ -70,7 +70,7 @@ const fail = m => log(`  ${c.red}✗${c.reset} ${m}`);
 
 function banner() {
   log(`\n${c.bold}${c.cyan}  ┌──────────────────────────────────────────────────┐`);
-  log(`  │  📱 @buivietphi/skill-mobile-mt v1.4.0           │`);
+  log(`  │  📱 @buivietphi/skill-mobile-mt v1.4.1           │`);
   log(`  │  Master Senior Mobile Engineer                    │`);
   log(`  │                                                  │`);
   log(`  │  Claude · Cline · Roo Code · Cursor · Windsurf  │`);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@buivietphi/skill-mobile-mt",
-  "version": "1.4.0",
+  "version": "1.4.1",
   "description": "Master Senior Mobile Engineer skill for AI agents. Pre-built patterns from 18 production apps + local project adaptation. React Native, Flutter, iOS, Android. Supports Claude, Gemini, Kimi, Cursor, Copilot, Antigravity.",
   "author": "buivietphi",
   "license": "MIT",

package/shared/on-device-ai.md

@@ -1,175 +0,0 @@
-# On-Device AI — Mobile ML Integration
-
-> On-demand. Load when: "on-device AI", "ML model", "Core ML", "TFLite", "MediaPipe", "llama", "inference", "local model"
-> Source: llama.cpp, Core ML, MediaPipe, TensorFlow Lite
-
----
-
-## Decision Matrix
-
-```
-Use case                           Solution
-───────────────────────────────────────────────────────────
-Image classification / OCR         Core ML (iOS) / ML Kit (Android)
-Text classification / sentiment     Core ML NLP / ML Kit
-Face detection / pose estimation   Vision (iOS) / MediaPipe
-On-device LLM chat (<7B params)    llama.cpp / llama.rn / executorch
-Cloud LLM (>7B / latest models)    API call — don't run on device
-Real-time object detection         Core ML / TFLite + MediaPipe
-Speech to text (on-device)         SFSpeechRecognizer (iOS) / ML Kit (Android)
-
-Rule: If model > 500MB → use API. If latency > 3s acceptable → use API.
-```
-
----
-
-## iOS — Core ML
-
-```swift
-// 1. Import model (drag .mlpackage into Xcode)
-import CoreML
-import Vision
-
-// 2. Image classification
-let model = try VNCoreMLModel(for: MyClassifier(configuration: .init()).model)
-let request = VNCoreMLRequest(model: model) { request, _ in
-    guard let results = request.results as? [VNClassificationObservation] else { return }
-    let top = results.first!
-    print("\(top.identifier): \(top.confidence)")
-}
-let handler = VNImageRequestHandler(cgImage: image, options: [:])
-try handler.perform([request])
-
-// 3. NLP text classification
-import NaturalLanguage
-let classifier = NLModel(mlModel: SentimentClassifier().model)
-let label = classifier.predictedLabel(for: "This is great!")
-
-// Model conversion: use coremltools Python package
-// coremltools.convert(pytorch_model, inputs=[...])
-```
-
----
-
-## Android — ML Kit + MediaPipe
-
-```kotlin
-// ML Kit — text recognition (no model download needed)
-dependencies {
-    implementation("com.google.mlkit:text-recognition:16.0.0")
-    implementation("com.google.mlkit:face-detection:16.1.5")
-}
-
-val recognizer = TextRecognition.getClient(TextRecognizerOptions.DEFAULT_OPTIONS)
-recognizer.process(inputImage)
-    .addOnSuccessListener { result -> result.text }
-    .addOnFailureListener { e -> /* handle */ }
-
-// MediaPipe — pose / hand / face landmark detection
-dependencies {
-    implementation("com.google.mediapipe:tasks-vision:0.10.14")
-}
-
-val handLandmarker = HandLandmarker.createFromOptions(context,
-    HandLandmarkerOptions.builder()
-        .setBaseOptions(BaseOptions.builder().setModelAssetPath("hand_landmarker.task").build())
-        .setNumHands(2)
-        .build()
-)
-```
-
----
-
-## On-Device LLM — llama.cpp (Cross-Platform)
-
-```
-Model sizes (GGUF Q4_K_M quantization):
-  Llama 3.2 3B   → ~2GB RAM  ✅ Phone-friendly
-  Llama 3.1 8B   → ~5GB RAM  ⚠️ High-end only (iPhone 15 Pro, Pixel 9)
-  Llama 3.1 70B  → ~40GB RAM ❌ Not feasible on device
-
-Download: huggingface.co/models?search=gguf
-```
-
-```swift
-// iOS — llama.swift
-// https://github.com/ggerganov/llama.cpp (Swift bindings included)
-import llama
-
-let model = llama_load_model_from_file(modelPath, llama_model_default_params())
-let ctx = llama_new_context_with_model(model, llama_context_default_params())
-// Tokenize + run inference on background thread
-```
-
-```javascript
-// React Native — llama.rn
-// npm install llama.rn
-import { LlamaContext } from 'llama.rn';
-
-const context = await LlamaContext.create({
-  model: `${RNFS.DocumentDirectoryPath}/model.gguf`,
-  n_ctx: 2048,
-  n_threads: 4,
-});
-const result = await context.completion({ prompt: 'Hello!', n_predict: 100 });
-```
-
-```dart
-// Flutter — flutter_llama (or use Platform.channel to llama.cpp)
-// For production: use executorch (Meta) or llama.cpp via FFI
-```
-
----
-
-## React Native — ML Kit (via react-native-mlkit)
-
-```javascript
-// npm install @infinitered/react-native-mlkit-core
-// npm install @infinitered/react-native-mlkit-object-detection
-
-import { ObjectDetectionCamera } from '@infinitered/react-native-mlkit-object-detection';
-
-// Image labeling
-import MLKitImageLabeling from '@react-native-ml-kit/image-labeling';
-const labels = await MLKitImageLabeling.label(imageUri);
-// Returns: [{ text: 'Cat', confidence: 0.95 }]
-```
-
----
-
-## Flutter — tflite_flutter
-
-```dart
-// pubspec.yaml: tflite_flutter: ^0.10.4
-import 'package:tflite_flutter/tflite_flutter.dart';
-
-final interpreter = await Interpreter.fromAsset('model.tflite');
-final input = [imageData];  // pre-processed tensor
-final output = List.filled(1000, 0).reshape([1, 1000]);
-interpreter.run(input, output);
-// output[0] = probability for each class
-```
-
----
-
-## Performance Rules
-
-```
-1. NEVER run inference on the main thread
-   iOS: DispatchQueue.global(qos: .userInitiated).async { ... }
-   Android: viewModelScope.launch(Dispatchers.Default) { ... }
-   RN: run on JS thread or use NativeModule
-
-2. Load model ONCE — cache in memory
-   ❌ Load model on every inference call
-   ✅ Load at app start or first use, keep reference
-
-3. Batch requests when possible
-   - Process images in background queue, not per-tap
-
-4. Show progress for operations >500ms
-   - Spinner or progress bar — user expects AI to take a moment
-
-5. Fallback to API if device is low on memory
-   let memoryPressure = ProcessInfo.processInfo.isLowPowerModeEnabled
-```