🚀 Live Demo | 🔬 Scientific Audit
The world's first code quality analyzer with time-travel capabilities. Track how your code evolved across Git commits.
Every code quality tool tells you what's wrong with your code right now. But they don't answer the most important question:
"How did my code get this way?"
Refactor Codex is the world's first code analyzer with time-travel capabilities. Track quality metrics across Git commit history, identify regressions, and see exactly when technical debt was introduced.
Built by stitching together incompatible technologies into something alive:
┌─────────────────────────────────────────────┐
│ React Frontend (Vite + React 19) │
│ ↓ HTTP/REST │
│ Express Backend (Node.js) │
│ ├─ JavaScript/TypeScript Analyzer (Babel) │
│ ├─ Python Analyzer (subprocess) │
│ ├─ GitHub API Integration │
│ └─ Gemini AI Integration │
│ ↓ stdio │
│ MCP Server (Python) │
│ │ │
└─────────────────────────────────────────────┘
Technologies Stitched Together:
- 🐍 Python + ⚡ JavaScript/TypeScript (dual runtime analysis)
- 🧬 AST Analysis + 🤖 AI-Powered Refactoring (Babel + Gemini)
- 🔌 MCP Protocol + 🌐 REST API + ⚛️ React (three architectures, unified)
- ⏰ Historical Analysis + 📊 Real-time Metrics (Git + GitHub API)
- 🎨 Three.js + 💀 Gothic Horror Theme (because refactoring is reanimating dead code)
UNIQUE - No other tool does this. Track code quality evolution across Git commit history:
- Historical Timeline - Quality scores across up to 20 commits
- Regression Detection - Pinpoint commits that introduced technical debt
- Trend Analysis - Visualize improvement/decline patterns
- Best/Worst Commits - Identify quality peaks and valleys
- Commit-by-Commit Breakdown - Detailed metrics for each version
- Developer Journey - Track coding skill improvement over time
Real Example: Analyze lodash/array.js over 10 commits:
Commit abc123: Quality dropped 15 points (regression detected)
Commit def456: Quality improved 8 points (refactoring success)
Overall trend: Improving (+12 points over 6 months)
True multi-language support - JavaScript, TypeScript, AND Python with scientifically accurate metrics:
- Quality Score (0-100) - Weighted combination of complexity, smells, and maintainability
- McCabe Cyclomatic Complexity - Industry-standard metric (M = decision_points + 1)
- Toxicity Score (0-100) - Severity-weighted code smell density
- Maintainability Index - Research-based formula:
MI = 0.5×Q + 0.3×(100-T) + 0.2×(100-5C) - Technical Debt - SQALE method: 15min per smell, severity-weighted
- Function-Level Analysis - Detailed breakdown of every function
- Code Smell Detection - 12+ smell types with remediation suggestions
Intelligent refactoring suggestions with context:
- Extract Function Refactoring - Identify code blocks that should be separate functions
- Before/After Diffs - Side-by-side comparison with syntax highlighting
- Risk Assessment - Safety ratings (Low/Medium/High risk)
- AI Explanations - Two AI personalities:
- Friendly Assistant - Encouraging, educational explanations
- Mr. Smith - Direct, technical analysis for serious issues
- Step-by-Step Implementation - Detailed guidance for each refactoring
- Parameter Detection - Automatic identification of function parameters and return values
Analyze entire codebases with scientific precision:
- Batch Analysis - Scan up to 30 files simultaneously
- Repository Health Score - Aggregate quality metrics with statistical analysis
- Worst Files First - Prioritized refactoring recommendations
- Language Detection - Automatic JS/TS/Python identification
- Rate Limit Handling - Smart GitHub API management with caching
- Smell Density Metrics - Issues per 1000 lines (industry standard)
- Technical Debt Estimation - Total remediation time across codebase
Every other tool shows you code quality NOW. We show you the STORY.
- See how quality evolved over 10+ commits
- Find the exact commit that introduced tech debt
- Track your improvement as a developer
- Visualize your coding journey
SonarQube costs $$$$ for this. We do it free.
┌─────────────────────────────────────────────┐
│ React Frontend (JavaScript) │
│ ↓ HTTP │
│ Express Backend (Node.js) │
│ ↓ Child Process │
│ Python Analyzer (Python AST) │
│ ↓ stdio │
│ MCP Server (Python) │
│ │
└─────────────────────────────────────────────┘
- Multi-language support (JS/TS/Python)
- GitHub API integration with rate limiting
- Error handling at every layer
- Batch processing with async operations
- Extensible architecture
- Developers spend 60% of time reading code, 40% writing
- Technical debt costs $3.61 per line of code
- Code reviews take 4-8 hours per week
- Refactor Codex automates the analysis and suggests fixes
# 1. Setup API Key (REQUIRED for AI features)
cd backend
cp .env.example .env
# Edit .env and add your Gemini API key:
# GEMINI_API_KEY=your_key_here
# Get free key: https://aistudio.google.com/app/apikey
# 2. Start backend
npm install
npm start
# 3. Start frontend (new terminal)
cd frontend
npm install
npm run dev
# 4. Open http://localhost:5173The Mr. Smith AI analysis requires a Google Gemini API key:
- Get your FREE API key: https://aistudio.google.com/app/apikey
- Open
backend/.envfile - Add your key:
GEMINI_API_KEY=your_actual_key_here - Restart the backend server
Why? The API key is stored in .env which is in .gitignore - it won't be uploaded to GitHub, keeping your key safe!