Dr. Aria - AI Psychologist

Dr. Aria is a highly sophisticated, AI-driven psychological chat application. It provides real-time, clinically informed therapeutic conversations using a dual-model LLM architecture. Designed with strict privacy in mind, the platform supports both server-side and end-to-end encryption.

🧠 Core Features

• Real-Time Therapy Chat: Low-latency, token-by-token streaming responses using Server-Sent Events (SSE).
• Dual-Model AI Architecture:
  • Uses a heavy model (e.g., gpt-4o) for high-quality, empathetic primary responses.
  • Uses a lighter model (e.g., gpt-4o-mini) for asynchronous background tasks.
• Asynchronous Clinical Subsystems:
  • Observer: Silently analyzes conversations to extract clinical insights, emotional states, and behavioral patterns.
  • Compressor: Summarizes past conversational chapters to prevent context window overflow.
  • Synthesizer: Builds and maintains a longitudinal profile of the user's mental health journey.
• Dual-Mode Privacy:
  • Mode A (Server-Side Encryption): The server securely encrypts and decrypts messages in RAM.
  • Mode B (End-to-End Encryption): The client encrypts messages locally. The server only sees ciphertext for storage and transient plaintext in RAM strictly during the request lifecycle.
• Built-in CLI Tool: Includes a powerful chat.js CLI script for interacting directly with the backend without needing the frontend.

🛠 Tech Stack

• Frontend: Next.js 16, React 19, Tailwind CSS (Found in the Frontend/ folder).
• Backend: Node.js, Express.js.
• Database: MongoDB (Mongoose).
• AI Integration: OpenAI SDK.
• Security: JWT for stateless auth, bcrypt for password hashing, AES for message encryption.

📂 Project Structure

├── Frontend/           # Next.js 16 Client Application
├── ai/                 # AI prompt builders, wrappers, and asynchronous processors
│   ├── groq.js               # OpenAI API bindings & streaming logic
│   ├── observer.js           # Silent clinical insights generator
│   ├── chapterCompressor.js  # Context summarization
│   └── profileSynthesizer.js # Longitudinal user profiling
├── config/             # DB & environmental config
├── controllers/        # Express route handlers (Auth, Chat, Privacy)
├── middlewares/        # JWT Authentication and Route Protection
├── models/             # Mongoose schemas (User, Message, Profile, etc.)
├── routes/             # Express API routing and rate limiting
├── utils/              # Encryption utilities, mailing, and cron jobs
├── chat.js             # Interactive CLI client for local testing
└── index.js            # Main backend Express server entry point

🚀 Getting Started

Prerequisites

• Node.js (v18+)
• MongoDB instance (local or Atlas)
• OpenAI API Key

1. Backend Setup

1. Clone the repository and navigate to the project root.
2. Install dependencies:

   npm install

3. Create a .env file in the root directory:

   PORT=4000
   FRONTEND_URL=http://localhost:3000
   MONGODB_URL=mongodb+srv://<user>:<password>@cluster.mongodb.net/dr-aria
   SECRET_KEY=your_super_secret_jwt_key
   OPENAI_API_KEY=sk-proj-...
   
   # Optional: Model overrides
   OPENAI_MAIN_MODEL=gpt-4o
   OPENAI_FAST_MODEL=gpt-4o-mini

4. Start the backend server:

   npm run dev
   # or
   node index.js

2. Frontend Setup

1. Navigate to the frontend directory:

   cd Frontend

2. Install dependencies:

   npm install

3. Copy the example .env variables (ensure it points to the backend API):

   NEXT_PUBLIC_API_URL=http://localhost:4000/api/v1

4. Start the Next.js development server:

   npm run dev

5. Access the app at http://localhost:3000.

💻 Using the CLI (chat.js)

You can test the AI logic, SSE streams, and authentication locally without starting the frontend using the built-in CLI tool.

Ensure the backend server is running, then run:

node chat.js

CLI Options:

• node chat.js [email protected] (Login with specific email)
• node chat.js --signup (Force a fresh user creation)
• node chat.js --verbose (Show full error stack traces and JWT debug info)

In-Chat Commands:

• /history - Show your last 6 messages
• /me - View current user statistics and encryption mode
• /clear - Refresh session JWT
• /quit - Exit the CLI

🛡 Security & Rate Limiting

The application uses multiple strict limiters to prevent abuse:

• OTP Limiter: Protects SMS/Email gateways.
• Auth Limiter: Prevents brute forcing on /login and /signup.
• Chat Limiter: Limits users to a strict conversational pace (e.g., 20 requests per minute) to protect OpenAI API quotas.

🔄 Automated Tasks

• Weekly Reporter (utils/weeklyReportCron.js): A cron job runs every Sunday to compile insights and longitudinal profiles generated by the observer into structured weekly summaries.

---

Disclaimer: Dr. Aria is an AI tool designed for psychological support simulation and is not a replacement for professional human psychiatric or medical help.