[Analysis Layer] Change Impact Analysis - Dependency Graph Awareness

[gloomcheng]

Problem Statement

When AI modifies a function/file, it doesn't know:

• What other files import it
• Will changes break downstream code
• Which tests need updating

This leads to incomplete refactors and hidden breakages.

Proposed Solution

Impact Analysis System

1. Dependency Graph

   • Build import/export graph of codebase
   • Track function/class references

2. Pre-change Analysis

   • Before modifying, show: "This affects 5 files"
   • List specific dependents

3. Change Validation

   • After change, verify dependents still work
   • Suggest necessary updates

Technical Approach

• Use TypeScript compiler API / tree-sitter for parsing
• Cache dependency graph, update incrementally
• Integrate with implement phase

Success Criteria

• AI sees impact before making changes
• Reduced broken imports after refactoring
• Automated suggestions for affected files

---

🤖 Generated with Claude Code

[gloomcheng]

✅ Issue #6 Completed: Change Impact Analysis - Dependency Graph Awareness

Successfully implemented dependency graph and change impact analysis system.

What was implemented:

1. ImpactAnalyzer Service - Core dependency graph engine

   • Parses TypeScript/JavaScript imports and exports
   • Builds bidirectional dependency graph (who imports what, who depends on this)
   • Recursive traversal to find all affected files
   • Breaking change detection (removed/renamed/signature-changed exports)
   • Smart caching (5-minute TTL) for performance
   • Supports configurable depth and exclude patterns

2. ImpactAnalysisHandler - MCP tool handler

   • 5 new MCP tools: impact-build-graph, impact-analyze, impact-preview, impact-validate, impact-stats
   • Rich impact reporting with levels (low → medium → high → critical)
   • Groups affected files by directory for clarity
   • Separates production code from test files
   • Generates actionable suggestions

3. Impact Analysis Features

   • Pre-change preview: See what will be affected before coding
   • Full analysis: Detailed breakdown of all dependents and usage
   • Post-change validation: Verify changes didn't break imports
   • Graph statistics: Understand codebase structure

Key Capabilities:

• Import/export dependency tracking
• Affected files detection with depth control
• Impact level calculation (based on affected file count)
• Breaking change identification
• Test vs production file separation
• Backward compatibility suggestions

Example Workflow:

1. impact-preview files=["src/utils/format.ts"] - Check before modifying
2. Make changes to the file
3. impact-validate files=["src/utils/format.ts"] - Verify nothing broke
4. Review affected files and update as needed

Impact Levels:

• Low (0-3 files): Safe to proceed
• Medium (4-10 files): Review affected files
• High (11-25 files): Consider backward compatibility
• Critical (26+ files): Major refactoring, deprecation recommended

This completes the first Analysis Layer feature, giving AI awareness of:

• Code dependencies and import relationships ✅
• Ripple effects of changes before/after modifications ✅
• Breaking changes and affected code paths ✅

Committed in: 046e7e4