🧠 CodeMind AI — Code Complexity Reviewer

AI-powered code analysis with multi-tier intelligence — Gemini → AWS Bedrock → Offline fallback.

---

🌐 Live Demo

🔗 https://codemind-frontend.onrender.com

Tip

The app auto-warms all backend services on first load — no manual steps needed. Just click and wait ~30s on first visit (Render free tier cold start).

---

✨ Features

Feature	Description
🤖 Multi-Tier AI Router	Gemini → AWS Bedrock → Offline fallback for 100% uptime
⚡ Request Caching	SHA-256 hash lookup in MongoDB — cache hits respond in ~5ms
📊 Complexity Analysis	Big-O time & space complexity, warnings, and improvement tips
💬 AI Chat Assistant	Ask follow-up questions about your code in real time
🔥 Warmup Screen	Auto-pings all services on load — no more cold start failures
🦴 Skeleton Loaders	Shimmer placeholders while AI analysis runs
🛡️ Robust Error Handling	Standardized error responses with graceful degradation
🧪 Unit Test Coverage	12 Pytest cases covering all offline analysis branches
🔄 Full CI/CD Pipeline	GitHub Actions: test → build → deploy to Render automatically
🏓 Keep-Alive Cron	Pings all 3 Render services every 10 min — prevents cold starts

---

🏗️ Architecture

CodeMind AI/
├── .github/
│   └── workflows/
│       ├── frontend.yml      ← CI: React build check
│       ├── backend.yml       ← CI: Node.js syntax check
│       ├── ml-service.yml    ← CI: Pytest suite (12 tests)
│       ├── deploy.yml        ← CD: Auto-deploy all 3 to Render
│       └── keep-alive.yml    ← Cron: ping services every 10 min
├── frontend/                 ← React 19 + Vanilla CSS dark theme
│   └── src/
│       ├── components/
│       │   ├── WarmupScreen.js     # Cold-start handler
│       │   ├── CodeEditor.js       # Code input area
│       │   ├── ResultPanel.js      # Analysis report
│       │   ├── SkeletonLoader.js   # Shimmer placeholders
│       │   ├── AiAssistant.js      # Chat interface
│       │   ├── ComplexityGraph.js  # Big-O visualizer
│       │   ├── Header.js
│       │   └── Sidebar.js          # History panel
│       └── index.css
├── backend/                  ← Node.js/Express — API gateway + cache
│   └── server.js             # MD5 cache, MongoDB, proxy to ML service
├── ml-service/               ← Python Flask — AI orchestration
│   ├── app.py                # 3-tier router + offline engine
│   ├── pytest.ini
│   └── tests/
│       └── test_analyze_offline.py
└── README.md

Intelligence Router Flow

User Request
     │
     ▼
┌─────────────┐    ✅ success    ┌──────────────┐
│  Google     │──────────────►  │   Response   │
│  Gemini     │                 └──────────────┘
└─────────────┘
     │ ❌ timeout / 429
     ▼
┌─────────────┐    ✅ success    ┌──────────────┐
│    AWS      │──────────────►  │   Response   │
│  Bedrock    │                 └──────────────┘
└─────────────┘
     │ ❌ error
     ▼
┌─────────────┐
│   Offline   │──────────────►  Static heuristic analysis
│  Analyzer   │
└─────────────┘

---

🚀 Quick Start (Local)

Prerequisites

• Node.js v18+
• Python 3.10+
• MongoDB (local or Atlas)
• Git

1. Clone the Repository

git clone https://github.com/Meenbudha/code-complexity-reviewer.git
cd code-complexity-reviewer

2. Set Up the Backend (Node.js)

cd backend
npm install

Create backend/.env:

PORT=5000
MONGO_URL=mongodb://localhost:27017/codemind
ML_SERVICE_URL=http://localhost:8000

node server.js

3. Set Up the ML Service (Python Flask)

cd ml-service
python -m venv venv

# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activate

pip install -r requirements.txt

Create ml-service/.env:

GEMINI_API_KEY=your_gemini_api_key_here
AWS_ACCESS_KEY_ID=your_aws_access_key
AWS_SECRET_ACCESS_KEY=your_aws_secret_key
AWS_REGION=us-east-1

python app.py

4. Set Up the Frontend (React)

cd frontend
npm install --legacy-peer-deps

Create frontend/.env:

REACT_APP_BACKEND_URL=http://localhost:5000
REACT_APP_ML_URL=http://localhost:8000

npm start

Open http://localhost:3000 in your browser.

---

🧪 Running Tests

cd ml-service
python -m pytest tests/ -v

Expected output: 12 passed covering O(1), O(n), O(n²), O(log n), O(n log n), O(2^n), recursion, hashmap, string concat, deep nesting, Java syntax.

---

☁️ Deployment on Render

All 3 services are deployed on Render free tier.

Service	Render Config
Frontend	Root: frontend · Build: npm install --legacy-peer-deps && npm run build · Publish: build/
Backend	Root: backend · Build: npm install · Start: node server.js
ML Service	Root: ml-service · Build: pip install -r requirements.txt · Start: gunicorn app:app --workers 2 --timeout 60 --bind 0.0.0.0:$PORT

Environment Variables on Render

Frontend service:

REACT_APP_BACKEND_URL = https://your-backend.onrender.com
REACT_APP_ML_URL      = https://your-ml-service.onrender.com

Important

REACT_APP_* variables must be set in Render before building. React bakes them into the static bundle at build time — they are not runtime variables.

Backend service:

MONGO_URL       = mongodb+srv://... (MongoDB Atlas)
ML_SERVICE_URL  = https://your-ml-service.onrender.com
PORT            = 5000

ML service:

GEMINI_API_KEY       = ...
AWS_ACCESS_KEY_ID    = ...
AWS_SECRET_ACCESS_KEY= ...
AWS_REGION           = us-east-1
FLASK_ENV            = production

---

🔄 CI/CD Pipeline

Every git push to main triggers a full automated pipeline:

Push to main
     │
     ├── ⚛️ Frontend CI   (npm build)
     ├── 🟡 Backend CI    (node --check)
     └── 🐍 ML Service CI (pytest 12 tests)
           │ all pass ✅
           ▼
     ├── 🌐 Deploy Frontend → Render
     ├── 🟡 Deploy Backend  → Render
     └── 🤖 Deploy ML       → Render
           │
           ▼
     ✅ Deployment Summary printed

Keep-Alive: A separate cron workflow pings all 3 services every 10 minutes to prevent Render free-tier spin-down.

GitHub Secrets Required

Important

All sensitive values must go in the Secrets tab — NOT the Variables tab.

Secret	Purpose
GEMINI_API_KEY	Gemini AI provider
AWS_ACCESS_KEY_ID	AWS Bedrock fallback
AWS_SECRET_ACCESS_KEY	AWS Bedrock fallback
RENDER_DEPLOY_HOOK_FRONTEND	Auto-deploy frontend
RENDER_DEPLOY_HOOK_BACKEND	Auto-deploy backend
RENDER_DEPLOY_HOOK_ML	Auto-deploy ML service
RENDER_BACKEND_URL	Keep-alive ping
RENDER_ML_URL	Keep-alive ping

---

🛠️ Tech Stack

Layer	Technology
Frontend	React 19, Vanilla CSS, Custom Dark Theme
Backend	Node.js 18, Express 5, Mongoose
ML Service	Python 3.10, Flask, Gunicorn
AI — Primary	Google Gemini 2.0 Flash
AI — Fallback	AWS Bedrock (Amazon Nova Micro)
AI — Offline	Custom heuristic regex engine
Database	MongoDB Atlas (cache + history)
Testing	Pytest (12 offline unit tests)
CI/CD	GitHub Actions (4 workflows)
Hosting	Render (all 3 services)

---

🗺️ Roadmap

• Deploy all 3 services to Render
• Full CI/CD pipeline with GitHub Actions
• Cold start WarmupScreen
• Keep-alive cron job
• Unit tests (12 Pytest cases)
• Add input validation + rate limiting
• Support more languages (JavaScript, Go, Rust)
• Dark/Light mode persistence (localStorage)
• VS Code extension

---

📄 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

Built with ❤️ by Meenbudha

⭐ Star this repo if you find it useful!

📖 Full Architecture Docs →