@cloudkinetix/bmad-enhanced/expansion-packs/ck-parallel-dev/utils/llm-dependency-analyzer.md

Cloud-Kinetix enhanced fork of BMAD-METHOD - Breakthrough Method of Agile AI-driven Development with robust versioning and unified validation.

---
name: LLM-Native Dependency Analyzer
version: 1.0.0
role: Analyze semantic dependencies for parallel development
description: Provides intelligent dependency analysis using LLM reasoning capabilities to understand code purpose, architectural patterns, and business logic connections
capabilities:
  - Semantic dependency detection
  - Hidden conflict identification
  - Risk-based planning
  - Platform-agnostic analysis
  - Architectural impact assessment
---

# LLM-Native Dependency Analyzer

## Purpose

Provides intelligent dependency analysis using LLM reasoning capabilities instead of script-based file scanning. This analyzer understands semantic dependencies, architectural patterns, and business logic connections between work items.

## Core Capabilities

### 1. Semantic Dependency Detection

- **Code Understanding**: Analyzes code purpose and functionality
- **API Contract Analysis**: Identifies interface dependencies
- **Data Flow Tracking**: Understands how data moves through the system
- **Business Logic Dependencies**: Recognizes functional relationships

### 2. Intelligent Conflict Prediction

- **Direct Conflicts**: Files/functions that will be modified
- **Indirect Conflicts**: Cascading effects of changes
- **Integration Points**: Shared interfaces and contracts
- **Test Dependencies**: Test suites affected by changes

### 3. Risk Assessment

- **Merge Conflict Probability**: Based on code overlap
- **Integration Risk**: Based on API changes
- **Testing Risk**: Based on test coverage impact
- **Architecture Risk**: Based on structural changes

## Analysis Framework

### Input Structure

```json
{
  "workItems": [
    {
      "id": "story-1",
      "description": "Add user authentication system",
      "type": "feature",
      "technicalDetails": "Implement JWT-based auth with refresh tokens"
    },
    {
      "id": "story-2",
      "description": "Update user profile management",
      "type": "enhancement",
      "technicalDetails": "Add profile picture upload and validation"
    }
  ],
  "codebaseContext": {
    "architecture": "microservices|monolithic|serverless",
    "primaryLanguage": "javascript|python|java",
    "testingFramework": "jest|pytest|junit",
    "keyComponents": ["auth-service", "user-service", "api-gateway"]
  }
}
```

### Analysis Prompt Template

```markdown
You are an expert software architect analyzing dependencies between work items for parallel development. Your goal is to identify all dependencies - both obvious and subtle - that could cause conflicts or issues if these items are developed simultaneously.

## Work Items to Analyze:

[List each work item with description and technical details]

## Codebase Context:

[Provide architecture, language, and component information]

## Analysis Tasks:

1. **File-Level Dependencies**
   - List specific files/modules each work item will likely modify
   - Identify shared files between work items
   - Note configuration files that might be touched

2. **Semantic Dependencies**
   - API contracts that multiple items depend on
   - Shared data models or schemas
   - Common utility functions or services
   - Database tables or collections affected

3. **Architectural Dependencies**
   - Service boundaries crossed
   - Communication patterns affected
   - Infrastructure components impacted
   - Security or authentication flows modified

4. **Testing Dependencies**
   - Test files that will need updates
   - Integration tests affected
   - Test data or fixtures that might conflict
   - Testing infrastructure changes needed

5. **Risk Assessment**
   For each dependency found:
   - Conflict probability: HIGH/MEDIUM/LOW
   - Impact severity: CRITICAL/MAJOR/MINOR
   - Mitigation strategy: How to resolve if both proceed

6. **Parallelization Recommendation**
   Based on your analysis, recommend:
   - Which items can safely run in parallel
   - Which must be sequenced and in what order
   - Which might benefit from closer coordination

Provide your analysis in a structured format that can be parsed programmatically.
```

### Output Structure

```json
{
  "dependencyMatrix": {
    "story-1": {
      "story-2": {
        "conflictLevel": "MEDIUM",
        "sharedResources": ["user-model", "auth-middleware"],
        "riskFactors": ["Both modify authentication flow"],
        "recommendation": "Sequence with story-1 first"
      }
    }
  },
  "fileConflicts": [
    {
      "file": "src/models/user.js",
      "affectedBy": ["story-1", "story-2"],
      "conflictType": "schema-modification",
      "severity": "HIGH"
    }
  ],
  "semanticDependencies": [
    {
      "type": "api-contract",
      "description": "User authentication endpoint",
      "affectedStories": ["story-1", "story-2"],
      "impact": "Breaking change requires coordination"
    }
  ],
  "executionPlan": {
    "waves": [
      {
        "wave": 1,
        "parallel": ["story-3", "story-4"],
        "reasoning": "No shared dependencies"
      },
      {
        "wave": 2,
        "parallel": ["story-1"],
        "reasoning": "Establishes auth foundation"
      },
      {
        "wave": 3,
        "parallel": ["story-2"],
        "reasoning": "Depends on auth from wave 2"
      }
    ]
  }
}
```

## Integration with Parallel Development

### 1. Pre-Execution Analysis

```markdown
Before starting parallel development:

1. Gather work item descriptions and technical details
2. Run LLM dependency analysis
3. Generate conflict matrix and risk assessment
4. Create optimized wave execution plan
5. Present findings for approval
```

### 2. Continuous Monitoring

```markdown
During parallel execution:

1. Monitor for unexpected dependencies
2. Re-analyze if scope changes
3. Adjust wave planning dynamically
4. Alert on emerging conflicts
```

### 3. Post-Execution Learning

```markdown
After completion:

1. Compare predicted vs actual conflicts
2. Update analysis patterns
3. Refine risk assessment models
4. Improve future predictions
```

## Platform Adaptation

### OpenAI/GPT

```javascript
const response = await openai.chat.completions.create({
  model: "gpt-4",
  messages: [
    {
      role: "system",
      content: analysisPrompt,
    },
  ],
  response_format: { type: "json_object" },
});
```

### Anthropic/Claude

```javascript
const response = await anthropic.messages.create({
  model: "claude-3-opus-20240229",
  messages: [
    {
      role: "user",
      content: analysisPrompt,
    },
  ],
});
```

### Google/Gemini

```javascript
const response = await model.generateContent({
  contents: [
    {
      parts: [
        {
          text: analysisPrompt,
        },
      ],
    },
  ],
});
```

## Benefits Over Script-Based Analysis

1. **Deeper Understanding**: Comprehends code purpose, not just file names
2. **Flexible Detection**: Finds non-obvious dependencies
3. **Risk-Based Planning**: Prioritizes based on actual impact
4. **Platform Agnostic**: Works with any LLM provider
5. **Continuous Learning**: Improves with feedback
6. **Context Aware**: Understands architectural patterns

## Usage Example

```bash
# Analyze dependencies for sprint work
llm-analyze-dependencies \
  --work-items "story-1.md,story-2.md,story-3.md" \
  --architecture "microservices" \
  --output "dependency-analysis.json"

# Generate execution plan
llm-plan-parallel \
  --dependency-analysis "dependency-analysis.json" \
  --max-parallel 4 \
  --risk-tolerance "medium" \
  --output "execution-plan.json"
```

This LLM-native approach provides intelligent, semantic analysis that goes far beyond simple file matching, enabling truly smart parallel development planning.