smar-ai 🧠

Understand any code, fast.

Live app: https://smarai.rishmi5h.com

What is smar-ai?

smar-ai is an AI-powered GitHub repository analyzer that helps you understand, review, and work with any codebase. Paste a GitHub URL and get instant AI-generated analysis, architecture diagrams, README generation, PR/issue reviews, and copy-pasteable prompts you can give to any LLM to recreate or extend the project.

Powered by Groq (llama-3.3-70b-versatile) for blazing-fast inference.

Key Features

• 📊 Smart Analysis — AI-powered code understanding with 3 analysis modes (Overview, Detailed, Learning Guide)
• 🏗️ Architecture Diagrams — Auto-generated interactive dependency graphs with Mermaid.js and AI explanations
• 📝 README Generator — One-click professional README.md generation from repository code
• 🧩 Prompt Generator — Generate copy-pasteable prompts to recreate, extend, review, or migrate any codebase
• 🔀 PR & Issue Analysis — Paste a PR or issue URL for targeted AI review with diff analysis
• 📈 Codebase Evolution — Compare how code changed between any two commits
• 💬 Interactive Q&A — Chat with the AI about the analyzed repository
• ⚡ Real-time Streaming — See results as they're generated via Server-Sent Events
• 🎨 Beautiful UI — Modern, responsive interface with syntax highlighting and dark theme
• 📥 Export Options — Copy or download analysis, README, and prompts

Quick Start

Prerequisites

• Node.js 20+
• A Groq API key (free tier available)
• A GitHub token (optional, for higher rate limits)

Setup

# Clone the repository
git clone https://github.com/rishmi5h/smar-ai.git
cd smar-ai

# Backend
cd server
cp .env.example .env
# Edit .env with your GROQ_API_KEY
npm install
npm run dev

# In a new terminal — Frontend
cd client
npm install
npm run dev

Visit http://localhost:5173

How It Works

graph LR
    A[GitHub URL] -->|Fetch Code| B[GitHub API v3]
    B -->|Code + Metadata| C[Groq LLM]
    C -->|Stream via SSE| D[React Frontend]
    D -->|Display| E[Analysis / Architecture / README / Prompts]

Loading

1. Paste a URL — Repository, PR, or Issue URL
2. Choose analysis type — Overview, Detailed, or Learning Guide
3. Explore tabs — Analysis, Changes, Architecture, README, Prompts
4. Export — Copy, download, or use generated prompts with any LLM

Features in Detail

Analysis Modes

Mode	What You Get
Code Overview	Purpose, tech stack, architecture, key components
Detailed Explanation	File structure, data flow, design patterns, configuration
Learning Guide	Prerequisites, step-by-step learning path, hands-on activities

Architecture Diagrams

Automatically maps file dependencies across your codebase. Supports imports from JavaScript, TypeScript, Python, Go, Java, and Rust. Renders interactive Mermaid diagrams with zoom, pan, and SVG download.

README Generator

One-click generation of a professional README.md including project overview, features, tech stack, setup instructions, project structure, and contributing guidelines. Streams in real-time with copy and download options.

Prompt Generator

Generate prompts you can paste into any LLM (Claude, ChatGPT, Cursor, Copilot) to:

Prompt Type	Description
Recreate This Project	Full blueprint with tech stack, file structure, build order, and implementation patterns
Add a Feature	Convention-aware guide specifying files to create, patterns to follow, and wiring instructions
Code Review	Context-rich review prompt with focus areas, checklists, and output format
Convert / Migrate	Migration strategy with pattern mapping, file-by-file plan, and dependency replacements

Migration presets: JS → TS, React → Next.js, Express → Fastify, REST → GraphQL, Flask → FastAPI, CJS → ESM, or custom targets.

PR & Issue Analysis

Paste a PR URL to get AI-powered diff review with file-by-file analysis. Paste an issue URL for context-aware discussion and suggested approaches.

Codebase Evolution

Compare any two commits to see how the codebase evolved — structural changes, new patterns introduced, and code quality shifts.

Interactive Q&A

After analysis, chat with the AI about the repository. Ask follow-up questions about architecture, patterns, or specific files.

API Reference

All streaming endpoints use Server-Sent Events (SSE) for real-time responses.

Analyze Repository

POST /api/analyze-stream

curl -X POST http://localhost:5050/api/analyze-stream \
  -H "Content-Type: application/json" \
  -d '{"repoUrl": "https://github.com/facebook/react", "analysisType": "overview"}'

Analyze PR

POST /api/analyze-pr

curl -X POST http://localhost:5050/api/analyze-pr \
  -H "Content-Type: application/json" \
  -d '{"prUrl": "https://github.com/owner/repo/pull/123"}'

Analyze Issue

POST /api/analyze-issue

curl -X POST http://localhost:5050/api/analyze-issue \
  -H "Content-Type: application/json" \
  -d '{"issueUrl": "https://github.com/owner/repo/issues/456"}'

Generate Architecture

POST /api/architecture

curl -X POST http://localhost:5050/api/architecture \
  -H "Content-Type: application/json" \
  -d '{"repoUrl": "https://github.com/owner/repo"}'

Generate README

POST /api/readme

curl -X POST http://localhost:5050/api/readme \
  -H "Content-Type: application/json" \
  -d '{"repoUrl": "https://github.com/owner/repo"}'

Generate Prompt

POST /api/generate-prompt

curl -X POST http://localhost:5050/api/generate-prompt \
  -H "Content-Type: application/json" \
  -d '{"repoUrl": "https://github.com/owner/repo", "promptType": "recreate"}'

Prompt types: recreate, feature, review, migrate

Compare Changes

POST /api/changes

Chat with Repository

POST /api/chat

Repository Info

GET /api/repo-info?repoUrl=owner/repo

Project Structure

smar-ai/
├── server/
│   ├── src/
│   │   ├── index.js                  # Express server entry point
│   │   ├── services/
│   │   │   ├── groqService.js        # Groq AI — analysis, README, prompts, PR/issue review
│   │   │   ├── githubService.js      # GitHub API — repos, PRs, issues, trees, diffs
│   │   │   ├── importParser.js       # Multi-language import/dependency parser
│   │   │   ├── aiService.js          # AI service abstraction layer
│   │   │   └── ollamaService.js      # Ollama integration (legacy)
│   │   └── routes/
│   │       └── analyze.js            # All API endpoints
│   ├── .env.example
│   ├── Dockerfile
│   └── package.json
├── client/
│   ├── src/
│   │   ├── components/
│   │   │   ├── RepoAnalyzer.jsx      # Main orchestrator — URL detection, routing
│   │   │   ├── SearchBar.jsx         # URL input with type detection (repo/PR/issue)
│   │   │   ├── AnalysisResults.jsx   # Tab container — Analysis, Changes, Architecture, README, Prompts
│   │   │   ├── ArchitecturePanel.jsx # Mermaid diagram + AI analysis
│   │   │   ├── ChangesPanel.jsx      # Commit comparison / codebase evolution
│   │   │   ├── ReadmePanel.jsx       # README generator
│   │   │   ├── PromptPanel.jsx       # Prompt generator (4 types)
│   │   │   ├── ChatPanel.jsx         # Interactive Q&A
│   │   │   ├── PRAnalysisResults.jsx # PR diff review
│   │   │   ├── IssueAnalysisResults.jsx # Issue analysis
│   │   │   ├── MarkdownRenderer.jsx  # Markdown + syntax highlighting + Mermaid
│   │   │   └── LoadingSpinner.jsx    # Loading states
│   │   ├── App.jsx
│   │   └── index.css
│   ├── .env.example
│   ├── Dockerfile
│   ├── nginx.conf
│   ├── netlify.toml
│   └── vite.config.js
├── docker-compose.yml
├── DEPLOYMENT.md
└── README.md

Environment Variables

Server (server/.env):

PORT=5050
GROQ_API_KEY=your_groq_api_key_here
GROQ_MODEL=llama-3.3-70b-versatile
GITHUB_TOKEN=ghp_xxx  # Optional, for higher GitHub API limits

Client (client/.env):

VITE_API_URL=http://localhost:5050/api

Tech Stack

Frontend:

• React 19 + Vite
• Mermaid.js (architecture diagrams)
• react-syntax-highlighter
• Modern CSS3 with CSS variables

Backend:

• Node.js 20 + Express
• Groq SDK (llama-3.3-70b-versatile)
• GitHub REST API v3
• Server-Sent Events (SSE) for streaming

AI:

• Groq cloud inference (primary)
• Ollama local inference (legacy support)

Deployment:

• Docker & Docker Compose
• Railway (backend)
• Netlify (frontend)
• Nginx

Deployment

Option 1: Docker (Recommended)

./scripts/deploy-docker.sh start

Full Docker guide →

Option 2: Railway + Netlify (Cloud)

Backend (Railway):

1. Connect GitHub repository to Railway
2. Set environment variables:

   GROQ_API_KEY=gsk_...
   GROQ_MODEL=llama-3.3-70b-versatile
   GITHUB_TOKEN=ghp_... (optional)
   PORT=5050
   

Frontend (Netlify):

1. Connect GitHub repository to Netlify
2. Set build settings:
   • Base directory: client
   • Build command: npm run build
   • Publish directory: dist
3. Set environment:

   VITE_API_URL=https://your-railway.railway.app/api
   

Full Railway guide → Full Netlify guide →

Option 3: Traditional VPS

Supports AWS, DigitalOcean, Linode, etc.

Full VPS guide →

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create feature branch: git checkout -b feature/amazing-feature
3. Commit changes: git commit -m "Add amazing feature"
4. Push to branch: git push origin feature/amazing-feature
5. Open Pull Request

Troubleshooting

Q: "Invalid GitHub URL" A: Use format https://github.com/owner/repo, https://github.com/owner/repo/pull/123, or https://github.com/owner/repo/issues/456

Q: Rate limit errors A: Add GITHUB_TOKEN for higher GitHub API limits

Q: Docker won't start A: Make sure Docker is running and ports 3000, 5050 are available

Q: Architecture diagram shows no connections A: The repo may have very few source files, or use an unsupported import syntax

License

ISC License - See LICENSE file for details

Support

• 📖 Full Documentation
• 🚀 Deployment Guide
• 🐛 Report Issues
• 💬 Discussions

Acknowledgments

• Powered by Groq for fast LLM inference
• Built with React and Vite
• Architecture diagrams by Mermaid.js
• GitHub integration via GitHub REST API
• Deployed on Railway, Netlify, and Docker

---

Made with ❤️ for code understanding

Get Started Now →