musubi-sdd 3.0.0 → 3.0.1

package/bin/musubi-validate.js

@@ -184,16 +184,6 @@ program
     }
   });
 
-// All validations (duplicate removed, keeping original)
-
-      displayResults('Complexity Validation', results, options);
-      process.exit(results.passed ? 0 : 1);
-    } catch (error) {
-      console.error(chalk.red('✗ Validation error:'), error.message);
-      process.exit(1);
-    }
-  });
-
 // All validations
 program
   .command('all')

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "musubi-sdd",
-  "version": "3.0.0",
+  "version": "3.0.1",
   "description": "Ultimate Specification Driven Development Tool with 27 Agents for 7 AI Coding Platforms + MCP Integration (Claude Code, GitHub Copilot, Cursor, Gemini CLI, Windsurf, Codex, Qwen Code)",
   "main": "src/index.js",
   "bin": {

package/src/templates/agents/claude-code/skills/ai-ml-engineer/mlops-guide.md

@@ -0,0 +1,350 @@
+# MLOps Guide
+
+## Overview
+
+Best practices for Machine Learning Operations (MLOps) in production systems.
+
+---
+
+## MLOps Lifecycle
+
+```
+┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐
+│  Data    │───▶│  Train   │───▶│  Deploy  │───▶│ Monitor  │
+└──────────┘    └──────────┘    └──────────┘    └──────────┘
+     │               │               │               │
+     └───────────────┴───────────────┴───────────────┘
+                    Continuous Loop
+```
+
+---
+
+## 1. Data Management
+
+### Data Versioning
+
+```bash
+# Using DVC (Data Version Control)
+dvc init
+dvc add data/training_data.csv
+git add data/training_data.csv.dvc .gitignore
+git commit -m "Add training data v1"
+```
+
+### Data Pipeline
+
+```python
+# data_pipeline.py
+from prefect import flow, task
+
+@task
+def extract_data(source: str) -> pd.DataFrame:
+    return pd.read_csv(source)
+
+@task
+def transform_data(df: pd.DataFrame) -> pd.DataFrame:
+    # Feature engineering
+    df['feature_1'] = df['col_a'] * df['col_b']
+    return df
+
+@task
+def validate_data(df: pd.DataFrame) -> bool:
+    # Data quality checks
+    assert df['feature_1'].isnull().sum() == 0
+    return True
+
+@flow
+def data_pipeline(source: str):
+    df = extract_data(source)
+    df = transform_data(df)
+    validate_data(df)
+    return df
+```
+
+---
+
+## 2. Experiment Tracking
+
+### MLflow Setup
+
+```python
+import mlflow
+from mlflow.tracking import MlflowClient
+
+# Set tracking URI
+mlflow.set_tracking_uri("http://mlflow-server:5000")
+mlflow.set_experiment("my-experiment")
+
+# Log experiment
+with mlflow.start_run():
+    # Log parameters
+    mlflow.log_param("learning_rate", 0.01)
+    mlflow.log_param("epochs", 100)
+    
+    # Train model
+    model = train_model(X_train, y_train)
+    
+    # Log metrics
+    mlflow.log_metric("accuracy", accuracy)
+    mlflow.log_metric("f1_score", f1)
+    
+    # Log model
+    mlflow.sklearn.log_model(model, "model")
+    
+    # Log artifacts
+    mlflow.log_artifact("feature_importance.png")
+```
+
+---
+
+## 3. Model Registry
+
+### Model Versioning
+
+```python
+from mlflow.tracking import MlflowClient
+
+client = MlflowClient()
+
+# Register model
+model_uri = f"runs:/{run_id}/model"
+mv = client.create_model_version(
+    name="my-model",
+    source=model_uri,
+    run_id=run_id
+)
+
+# Transition to staging
+client.transition_model_version_stage(
+    name="my-model",
+    version=mv.version,
+    stage="Staging"
+)
+
+# Promote to production
+client.transition_model_version_stage(
+    name="my-model",
+    version=mv.version,
+    stage="Production"
+)
+```
+
+### Model Metadata
+
+```yaml
+# model_metadata.yaml
+model:
+  name: fraud-detector
+  version: 2.1.0
+  framework: scikit-learn
+  
+training:
+  date: 2024-01-15
+  dataset_version: v1.2
+  metrics:
+    accuracy: 0.95
+    f1_score: 0.92
+    
+requirements:
+  - scikit-learn==1.3.0
+  - pandas==2.0.0
+  
+schema:
+  input:
+    - name: amount
+      type: float
+    - name: category
+      type: string
+  output:
+    - name: is_fraud
+      type: boolean
+    - name: confidence
+      type: float
+```
+
+---
+
+## 4. Model Deployment
+
+### Serving with FastAPI
+
+```python
+from fastapi import FastAPI
+import mlflow
+
+app = FastAPI()
+
+# Load model on startup
+model = mlflow.sklearn.load_model("models:/my-model/Production")
+
+@app.post("/predict")
+async def predict(features: dict):
+    df = pd.DataFrame([features])
+    prediction = model.predict(df)
+    probability = model.predict_proba(df)
+    
+    return {
+        "prediction": int(prediction[0]),
+        "confidence": float(probability[0].max())
+    }
+
+@app.get("/health")
+async def health():
+    return {"status": "healthy", "model_version": "2.1.0"}
+```
+
+### Deployment Strategies
+
+| Strategy | Description | Use Case |
+|----------|-------------|----------|
+| Shadow | Run parallel to existing | Validate new model |
+| Canary | Gradual traffic shift | Safe rollout |
+| Blue-Green | Full switch | Quick rollback |
+| A/B Test | Split traffic | Compare models |
+
+---
+
+## 5. Monitoring
+
+### Prediction Logging
+
+```python
+import logging
+from datetime import datetime
+
+def log_prediction(request, response, latency_ms):
+    logging.info({
+        "timestamp": datetime.utcnow().isoformat(),
+        "request_id": request.id,
+        "features": request.features,
+        "prediction": response.prediction,
+        "confidence": response.confidence,
+        "latency_ms": latency_ms,
+        "model_version": "2.1.0"
+    })
+```
+
+### Data Drift Detection
+
+```python
+from scipy import stats
+
+def detect_drift(reference_data, current_data, threshold=0.05):
+    """Detect distribution drift using KS test."""
+    drifted_features = []
+    
+    for column in reference_data.columns:
+        statistic, p_value = stats.ks_2samp(
+            reference_data[column],
+            current_data[column]
+        )
+        
+        if p_value < threshold:
+            drifted_features.append({
+                "feature": column,
+                "p_value": p_value,
+                "statistic": statistic
+            })
+    
+    return drifted_features
+```
+
+### Performance Metrics
+
+```python
+# Prometheus metrics for ML
+from prometheus_client import Counter, Histogram, Gauge
+
+prediction_counter = Counter(
+    'model_predictions_total',
+    'Total predictions',
+    ['model_version', 'prediction_class']
+)
+
+prediction_latency = Histogram(
+    'model_prediction_latency_seconds',
+    'Prediction latency',
+    ['model_version']
+)
+
+model_accuracy = Gauge(
+    'model_accuracy',
+    'Current model accuracy',
+    ['model_version']
+)
+```
+
+---
+
+## 6. CI/CD for ML
+
+### GitHub Actions Pipeline
+
+```yaml
+# .github/workflows/ml-pipeline.yml
+name: ML Pipeline
+
+on:
+  push:
+    paths:
+      - 'models/**'
+      - 'data/**'
+
+jobs:
+  train:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      
+      - name: Setup Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.10'
+      
+      - name: Install dependencies
+        run: pip install -r requirements.txt
+      
+      - name: Pull data
+        run: dvc pull
+      
+      - name: Train model
+        run: python train.py
+      
+      - name: Evaluate model
+        run: python evaluate.py
+      
+      - name: Register model
+        if: github.ref == 'refs/heads/main'
+        run: python register_model.py
+```
+
+---
+
+## 7. Best Practices
+
+### Reproducibility Checklist
+
+- [ ] Code versioned in Git
+- [ ] Data versioned with DVC
+- [ ] Dependencies pinned (requirements.txt)
+- [ ] Random seeds set
+- [ ] Experiments logged in MLflow
+- [ ] Model artifacts stored
+
+### Model Validation Checklist
+
+- [ ] Performance metrics acceptable
+- [ ] No data leakage
+- [ ] Fairness metrics checked
+- [ ] Edge cases tested
+- [ ] Latency requirements met
+- [ ] Memory usage acceptable
+
+### Production Checklist
+
+- [ ] Model card documented
+- [ ] API versioned
+- [ ] Health checks implemented
+- [ ] Monitoring in place
+- [ ] Rollback procedure defined
+- [ ] A/B test framework ready

package/src/templates/agents/claude-code/skills/ai-ml-engineer/model-card-template.md

@@ -0,0 +1,246 @@
+# Model Card Template
+
+## Overview
+
+Template for documenting machine learning models following best practices.
+
+---
+
+## Model Card Document
+
+```markdown
+# Model Card: [Model Name]
+
+## Model Details
+
+### Basic Information
+| Field | Value |
+|-------|-------|
+| Model Name | [Name] |
+| Version | [X.Y.Z] |
+| Type | [Classification/Regression/etc.] |
+| Framework | [TensorFlow/PyTorch/scikit-learn] |
+| Date | YYYY-MM-DD |
+| Authors | [Team/Names] |
+
+### Description
+[Brief description of what the model does]
+
+### Intended Use
+- **Primary Use Cases**: [What it's designed for]
+- **Intended Users**: [Who should use it]
+- **Out-of-Scope Uses**: [What it shouldn't be used for]
+
+---
+
+## Model Architecture
+
+### Overview
+[Description of model architecture]
+
+### Inputs
+| Name | Type | Shape | Description |
+|------|------|-------|-------------|
+| feature_1 | float | (1,) | Transaction amount |
+| feature_2 | int | (1,) | Category code |
+
+### Outputs
+| Name | Type | Shape | Description |
+|------|------|-------|-------------|
+| prediction | int | (1,) | Class label |
+| probability | float | (n_classes,) | Class probabilities |
+
+### Hyperparameters
+| Parameter | Value |
+|-----------|-------|
+| learning_rate | 0.001 |
+| batch_size | 32 |
+| epochs | 100 |
+
+---
+
+## Training Data
+
+### Dataset Description
+| Field | Value |
+|-------|-------|
+| Name | [Dataset name] |
+| Version | [Version] |
+| Size | [N samples] |
+| Date Range | [Start] to [End] |
+
+### Data Distribution
+| Feature | Distribution |
+|---------|--------------|
+| Class 0 | 85% |
+| Class 1 | 15% |
+
+### Preprocessing
+- [Step 1]: [Description]
+- [Step 2]: [Description]
+
+### Data Splits
+| Split | Size | Purpose |
+|-------|------|---------|
+| Train | 70% | Model training |
+| Validation | 15% | Hyperparameter tuning |
+| Test | 15% | Final evaluation |
+
+---
+
+## Evaluation
+
+### Metrics
+| Metric | Value | Threshold |
+|--------|-------|-----------|
+| Accuracy | 0.95 | > 0.90 |
+| Precision | 0.92 | > 0.85 |
+| Recall | 0.88 | > 0.80 |
+| F1 Score | 0.90 | > 0.85 |
+| AUC-ROC | 0.97 | > 0.90 |
+
+### Confusion Matrix
+```
+              Predicted
+              0     1
+Actual  0   850    50
+        1    30   120
+```
+
+### Performance by Subgroup
+| Subgroup | Accuracy | Size |
+|----------|----------|------|
+| Group A | 0.96 | 400 |
+| Group B | 0.94 | 350 |
+| Group C | 0.93 | 250 |
+
+---
+
+## Fairness & Bias
+
+### Evaluation
+| Metric | Group A | Group B | Threshold |
+|--------|---------|---------|-----------|
+| TPR | 0.89 | 0.87 | Δ < 0.05 ✅ |
+| FPR | 0.08 | 0.09 | Δ < 0.05 ✅ |
+| PPV | 0.91 | 0.89 | Δ < 0.05 ✅ |
+
+### Mitigation Steps
+- [Step taken to address bias]
+
+### Known Limitations
+- [Limitation 1]
+- [Limitation 2]
+
+---
+
+## Ethical Considerations
+
+### Potential Risks
+- [Risk 1]: [Mitigation]
+- [Risk 2]: [Mitigation]
+
+### Use Cases to Avoid
+- [Should not be used for X]
+
+---
+
+## Deployment
+
+### Requirements
+```
+python>=3.9
+tensorflow==2.12.0
+numpy==1.24.0
+```
+
+### Resource Requirements
+| Resource | Minimum | Recommended |
+|----------|---------|-------------|
+| CPU | 2 cores | 4 cores |
+| Memory | 2 GB | 4 GB |
+| GPU | - | NVIDIA T4 |
+
+### Latency
+| Percentile | Latency |
+|------------|---------|
+| p50 | 15ms |
+| p95 | 45ms |
+| p99 | 80ms |
+
+### Endpoints
+| Endpoint | Method | Description |
+|----------|--------|-------------|
+| /predict | POST | Get prediction |
+| /health | GET | Health check |
+
+---
+
+## Monitoring
+
+### Metrics to Track
+- Prediction distribution
+- Latency percentiles
+- Error rate
+- Data drift indicators
+
+### Alerting Thresholds
+| Metric | Warning | Critical |
+|--------|---------|----------|
+| Latency p99 | > 100ms | > 200ms |
+| Error Rate | > 1% | > 5% |
+| Drift Score | > 0.1 | > 0.2 |
+
+### Retraining Triggers
+- [Trigger 1]: [Condition]
+- [Trigger 2]: [Condition]
+
+---
+
+## Version History
+
+| Version | Date | Changes |
+|---------|------|---------|
+| 1.0.0 | YYYY-MM-DD | Initial release |
+| 1.1.0 | YYYY-MM-DD | Added feature X |
+| 2.0.0 | YYYY-MM-DD | Major architecture change |
+
+---
+
+## References
+
+- [Link to training code]
+- [Link to data documentation]
+- [Link to related papers]
+- [Link to API documentation]
+
+---
+
+## Contact
+
+For questions or issues:
+- Team: [Team name]
+- Email: [Contact email]
+- Slack: [Channel]
+```
+
+---
+
+## Quick Checklist
+
+### Before Release
+- [ ] Model architecture documented
+- [ ] Training data described
+- [ ] Evaluation metrics included
+- [ ] Fairness analysis completed
+- [ ] Ethical risks assessed
+- [ ] Deployment requirements listed
+- [ ] Monitoring plan defined
+- [ ] Version history updated
+
+### Review Questions
+1. Is the intended use clearly defined?
+2. Are limitations and risks documented?
+3. Can another team reproduce training?
+4. Are bias metrics acceptable?
+5. Is there a monitoring plan?