Delivers persistent chat history, intelligent thread management, rich Markdown and LaTeX rendering, and shareable AI conversations. It features automatic title generation, graceful handling of AI provider rate limits, and a scalable full-stack architecture designed for a smooth and reliable user experience.
🌐 Live Demo: https://www.orvexa.xyz
|
|
|
|
|
|
| Category | Technologies |
|---|---|
| Frontend | React 19, Vite, React Router 7 |
| Styling | CSS3 |
| Markdown | React Markdown, Remark GFM |
| Math Rendering | KaTeX, Remark Math |
| Syntax Highlighting | Highlight.js, Rehype Highlight |
| Backend | Node.js, Express.js 5 |
| Database | MongoDB Atlas, Mongoose |
| AI Integration | OpenRouter API |
| Security | Helmet, CORS, Express Rate Limit |
| Deployment | Vercel, Render |
┌──────────────────────────┐
│ User Browser │
└─────────────┬────────────┘
│
▼
┌─────────────────────────────────────┐
│ React 19 + Vite Frontend │
│ │
│ • Chat Interface │
│ • Thread Management │
│ • Markdown & LaTeX Rendering │
│ • Animated AI Responses │
└─────────────┬───────────────────────┘
│
▼
┌─────────────────────────────────────┐
│ Express.js Backend │
│ │
│ • Chat Controller │
│ • Thread Management │
│ • System Prompt │
│ • AI Services │
│ • Rate Limiting │
│ • Logging & Error Handling │
└───────┬─────────────────┬───────────┘
│ │
▼ ▼
┌────────────────────┐ ┌────────────────────────┐
│ MongoDB Atlas │ │ OpenRouter AI │
│ │ │ │
│ • Threads │ │ • Chat Responses │
│ • Messages │ │ • Title Generation │
│ • Titles │ │ • Conversation Summary │
│ • Summaries │ │ │
│ • Thread Metadata │ │ │
└────────────────────┘ └────────────────────────┘
Rather than focusing only on features, Orvexa was designed around performance, user experience, and maintainability.
Conversation title generation and long-chat summarization run asynchronously, ensuring the user receives the assistant's response immediately without waiting for additional AI tasks.
Messages appear instantly in the interface while network requests continue in the background, making the application feel responsive even during slower API calls.
To prevent AI context windows from growing indefinitely, conversations are automatically summarized after every 20 message exchanges. Future prompts use the generated summary instead of replaying the entire history, reducing token usage while preserving important context.
Every conversation is stored in MongoDB, allowing users to:
- Resume previous chats
- Rename conversations
- Pin important threads
- Delete conversations
- Share conversations via unique URLs
Each conversation is assigned a unique thread ID, making every chat accessible through a dedicated URL that can be copied and shared.
Assistant responses support:
- GitHub-Flavored Markdown
- Syntax-highlighted code blocks
- Mathematical equations with KaTeX
- Tables, lists, and formatted documentation
Instead of exposing raw provider errors, Orvexa detects AI quota limits and displays clear, user-friendly messages while preserving the integrity of the conversation history.
The backend applies both short-term and daily request limits to prevent abuse and protect the shared AI quota without affecting normal usage.
The application follows a modular structure with clearly separated concerns:
- Components – User Interface
- Context – Global State Management
- Controllers – Business Logic
- Services – AI Integration
- Models – Database Layer
- Middleware – Security & Error Handling
- Routes – API Endpoints
This separation improves maintainability and makes future feature development straightforward.
- Frontend: Vercel
- Backend: Render
- Database: MongoDB Atlas
- AI Provider: OpenRouter
The architecture separates presentation, business logic, persistence, and AI integration into independent layers, making the system easier to scale and maintain.
Note
Orvexa uses OpenRouter to access free AI models. Since free models are community-shared and provider availability may vary, certain models can become temporarily unavailable or rate-limited during periods of high demand. The application detects these situations and displays user-friendly messages while preserving conversation history.
git clone https://github.com/nur-hasin/Orvexa.git
cd Orvexacd Backend
npm install
nodemon server.jsCreate a .env file:
PORT=8080
NODE_ENV=development
CLIENT_URL=http://localhost:5173
MONGODB_URI=your_mongodb_connection_string
OPENROUTER_API_KEY=your_openrouter_api_keycd Frontend
npm install
npm run devCreate a .env file:
VITE_API_URL=http://localhost:8080Visit:
http://localhost:5173
Orvexa/
│
├── Backend/
│ ├── config/ # Environment, database, and AI configuration
│ ├── constants/ # System prompts and application constants
│ ├── controllers/ # Request handlers and business logic
│ ├── middleware/ # Security, logging, rate limiting, and error handling
│ ├── models/ # MongoDB/Mongoose schemas
│ ├── routes/ # Express API routes
│ ├── services/ # AI integration and helper services
│ └── server.js # Express server entry point
│
├── Frontend/
│ ├── public/ # Static assets
│ ├── src/
│ │ ├── assets/ # Images, icons, and logos
│ │ ├── components/ # Reusable React components
│ │ ├── context/ # Global state management (React Context)
│ │ ├── styles/ # Component-specific stylesheets
│ │ ├── App.css # Global application styles
│ │ ├── App.jsx # Root application component
│ │ └── main.jsx # React application entry point
│ └── vite.config.js # Vite configuration
├── docs/ # Screenshots, GIFs, logo, and documentation assets
- 🔐 User Authentication
- ⚡ Streaming AI Responses
- 📂 File & Image Upload Support
- 🎙️ Voice Conversations
- 🔍 Conversation Search
- 📤 Export & Import Chats
- 🌙 Light / Dark Theme
- 📱 Progressive Web App (PWA)
- 👥 Multi-user Collaboration
CSE Undergraduate • Full-Stack Developer • AI Enthusiast
"Building intelligent, scalable, and user-centric web applications with modern technologies."