MCP Code Analysis Server

# Domain Analysis Tools - Implementation Summary ## What Was Implemented ### 1. Pattern Analyzer Module (`src/domain/pattern_analyzer.py`) A comprehensive analysis engine that provides value beyond basic domain extraction: #### Key Features: - **Cross-Context Coupling Analysis**: Measures and analyzes dependencies between bounded contexts - **Anti-Pattern Detection**: Identifies common DDD anti-patterns in the codebase - **Context Split Suggestions**: Uses graph algorithms to suggest how to split large contexts - **Evolution Analysis**: Tracks how the domain model changes over time #### Implementation Details: - Uses SQL queries to analyze relationships stored during indexing - Implements graph algorithms for community detection - Provides actionable recommendations based on patterns found - Calculates metrics like cohesion scores and coupling factors ### 2. MCP Tool Wrappers (`src/mcp_server/tools/analysis_tools.py`) Exposes the pattern analyzer functionality through MCP tools: #### New Tools: 1. **`analyze_coupling`**: Provides detailed coupling analysis with metrics 2. **`detect_anti_patterns`**: Finds 6 types of DDD anti-patterns 3. **`suggest_context_splits`**: Suggests how to break up large contexts 4. **`analyze_domain_evolution`**: Shows domain changes over time 5. **`get_domain_metrics`**: Provides overall domain health score ### 3. Integration Points - Added pattern analyzer to domain module exports - Registered analysis tools in the MCP server - Created comprehensive documentation - Added test coverage for key functionality ## How It Works ### During Indexing: 1. Domain entities and relationships are extracted by LLM 2. Data is stored in PostgreSQL with relationships intact 3. Bounded contexts are detected using Leiden algorithm ### During Analysis (Value-Add Tools): 1. **Coupling Analysis**: Queries relationship data to find inter-context dependencies 2. **Anti-Pattern Detection**: Analyzes entity metadata for problematic patterns 3. **Split Suggestions**: Uses graph algorithms on existing data 4. **Evolution Tracking**: Compares timestamps to show changes ## Key Benefits 1. **Beyond Indexing**: These tools provide insights that aren't captured during indexing 2. **Actionable Results**: Each tool provides specific recommendations 3. **Health Metrics**: Overall scoring helps track improvement 4. **Pattern Recognition**: Identifies issues that are hard to spot manually ## Example Usage ```bash # Get overall domain health mcp call get_domain_metrics # Find architectural issues mcp call detect_anti_patterns # Analyze coupling problems mcp call analyze_coupling # Plan refactoring mcp call suggest_context_splits --min_entities 20 ``` ## Architecture Decision The analysis tools are separate from indexing because: 1. They need the full graph to provide meaningful insights 2. They're computationally different (analysis vs extraction) 3. They can be run on-demand without re-indexing 4. They provide cross-cutting concerns across the entire codebase This approach allows users to: - Index once, analyze many times - Run different analyses for different purposes - Track improvements over time - Focus on specific areas of concern