DevMate AI

A full-stack GitHub repository analyzer SaaS that uses machine learning to provide insights on issue difficulty, repository clustering, contribution success prediction, and personalized developer recommendations.

Tech Stack

Layer	Technology
Frontend	React 18, TypeScript, Vite, Material UI, Zustand, React Query
Backend	Node.js, Express, TypeScript, Prisma ORM
Database	PostgreSQL
ML Service	FastAPI, Python 3.11, scikit-learn, XGBoost
Auth	JWT (email + password)
DevOps	Docker, Docker Compose

---

Prerequisites

• Node.js v18+
• Python 3.11+
• PostgreSQL 15+ (or use Docker)
• Docker + Docker Compose (optional but recommended)

---

Option 1 — Run with Docker (Recommended)

The easiest way to run the entire project.

1. Clone the repository

git clone https://github.com/ManojLamani/DEVMATE_AI.git
cd DEVMATE_AI

2. Create a root .env file

cp .env.example .env

Edit .env and set at least JWT_SECRET. GITHUB_TOKEN and ANTHROPIC_API_KEY are optional.

3. Start all services

docker compose up --build

Database migrations run automatically when the backend starts.

Service	URL
Frontend	http://localhost:3000
Backend API	http://localhost:5000
ML Service	http://localhost:8000

If any of these ports are already in use, change FRONTEND_PORT, BACKEND_PORT, ML_PORT or POSTGRES_PORT in .env. Keep FRONTEND_URL and VITE_API_URL in step with them.

4. Run the smoke test (optional)

node scripts/smoke-test.mjs

This checks every API route, the ML service and the frontend against the running stack. It needs Node 18+. It reads the same .env ports, and you can pass a GitHub repo to analyse, for example node scripts/smoke-test.mjs pallets/click. It creates a throwaway smoke-…@example.com user each run.

---

Option 2 — Run Manually (Without Docker)

Step 1 — PostgreSQL

Create a database named devmate_db and note your credentials.

---

Step 2 — Backend

cd backend
npm install

Create .env in the backend/ folder:

DATABASE_URL="postgresql://postgres:your_password@localhost:5432/devmate_db"
JWT_SECRET="your_jwt_secret_key_here"
JWT_EXPIRES_IN="7d"
GITHUB_TOKEN=your_github_personal_access_token
PORT=5000
NODE_ENV="development"
FRONTEND_URL="http://localhost:3000"
ML_SERVICE_URL="http://localhost:8000"

Run migrations and start:

npm run db:generate
npm run db:migrate
npm run dev

Backend runs at http://localhost:5000

---

Step 3 — ML Service

cd ml-service
python -m venv venv

# Windows
venv\Scripts\activate

# macOS/Linux
source venv/bin/activate

pip install -r requirements.txt

Create .env in the ml-service/ folder:

ANTHROPIC_API_KEY=your_anthropic_api_key_here

Start the service:

uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

ML Service runs at http://localhost:8000

---

Step 4 — Frontend

cd frontend
npm install

Create .env in the frontend/ folder:

VITE_API_URL=http://localhost:5000/api

Start:

npm run dev

Frontend runs at http://localhost:3000

---

Getting API Keys

Key	Where to get
GITHUB_TOKEN	GitHub Settings → Developer settings → Personal access tokens — needs repo and read:user scopes
ANTHROPIC_API_KEY	Anthropic Console — required only for the AI repository explainer feature

---

Project Structure

DEVMATE_AI/
├── frontend/           # React + Vite frontend
│   └── src/
│       ├── pages/      # Route-level page components
│       ├── components/ # Shared UI components
│       ├── services/   # API service layer
│       └── store/      # Zustand state management
├── backend/            # Express + Prisma API
│   └── src/
│       ├── controllers/
│       ├── routes/
│       ├── middleware/
│       └── utils/
│   └── prisma/
│       └── schema.prisma
├── ml-service/         # FastAPI ML service
│   └── app/
│       ├── models/     # ML model classes
│       ├── schemas/    # Pydantic request schemas
│       ├── training/   # Model training scripts
│       └── utils/
│   └── saved_models/   # Pre-trained model files (.pkl)
├── scripts/
│   └── smoke-test.mjs  # End-to-end check of the running stack
├── .env.example        # Docker Compose configuration
└── docker-compose.yml

---

Deployment (Free)

Deploy the full stack for free using Vercel (frontend) + Render (backend + ML) + Supabase (database).

---

Step 1 — Database on Supabase (free)

1. Create a free account at supabase.com
2. Click New Project, set a name and password
3. Go to Settings → Database → Connection string → URI
4. Copy the DATABASE_URL — you'll need it in Step 2

---

Step 2 — Backend + ML Service on Render (free)

1. Create a free account at render.com
2. Click New → Blueprint and connect your GitHub repo (ManojLamani/DEVMATE_AI)
3. Render will detect the render.yaml and create both services automatically
4. Fill in the environment variables for each service in the Render dashboard:

devmate-backend:

Variable	Value
DATABASE_URL	Supabase connection string from Step 1
JWT_SECRET	Any long random string
GITHUB_TOKEN	Your GitHub personal access token
FRONTEND_URL	https://devmateai.com (or Vercel URL for now)
ML_SERVICE_URL	https://devmate-ml-service.onrender.com

devmate-ml-service:

Variable	Value
ANTHROPIC_API_KEY	From Anthropic Console (optional)

5. Click Deploy — Render will build and start both services
6. Note the backend URL (e.g. https://devmate-backend.onrender.com)

---

Step 3 — Frontend on Vercel (free)

1. Create a free account at vercel.com
2. Click Add New → Project and import ManojLamani/DEVMATE_AI
3. Set Root Directory to frontend
4. Add environment variables:

Variable	Value
VITE_API_URL	https://devmate-backend.onrender.com/api

5. Click Deploy
6. Vercel gives you a free URL like devmate-ai.vercel.app

---

Step 4 — Custom Domain (optional, ~$10/year)

To use a domain like devmateai.com:

1. Buy the domain at namecheap.com or cloudflare.com/registrar (~$10/year)
2. In Vercel: Project → Settings → Domains → Add Domain → enter devmateai.com
3. Vercel will show you DNS records — add them in your domain registrar's DNS settings
4. Update FRONTEND_URL in Render's backend env vars to https://devmateai.com

Note: Free Render services spin down after 15 minutes of inactivity and take ~30 seconds to wake up on the first request. Upgrade to Render's paid plan ($7/month) to keep services always-on.

---

Features

• Authentication — Email/password sign-in with JWT
• Repository Analysis — Fetch and store GitHub repo data (languages, contributors, issues, PRs, branches)
• Issue Difficulty Prediction — ML model classifies issues as Easy / Medium / Hard
• Repository Clustering — Groups repos by type (frontend, backend, data science, etc.)
• Contribution Success Prediction — Estimates likelihood of a PR being merged
• AI Repository Explainer — Claude-powered natural language explanation of a repository
• Developer Recommendations — Personalized repo and issue suggestions based on your profile
• Analytics Dashboard — Charts for commit activity, language breakdown, and PR trends