QuestGen-AI

QuestGen-AI is an advanced exam question generation platform that leverages AI-Agent to automatically create customized question papers from user-uploaded PDF content. Built for educators and assessment creators, QuestGen-AI streamlines the often time-consuming process of creating high-quality assessment materials.

questgen.mp4

🚀 Features

• AI-Powered Question Generation: Creates high-quality, contextually relevant questions from PDF content using advanced language models
• Multiple Question Types: Supports various question formats including multiple-choice, true/false, short answer, and essay questions
• Customizable Parameters: Set question paper parameters such as marks, difficulty levels, and question distribution
• Institutional Branding: Include custom headers with institution name, course details, and exam information
• PDF Document Processing: Upload and extract content from PDF study materials, textbooks, or lecture notes
• Real-time Generation: Watch as your question paper is created through a streaming interface
• Multi-Agent Architecture: Utilizes specialized AI agents for extraction, creation, analysis, and formatting of questions
• Markdown Export: Generated question papers are formatted in clean, structured Markdown for easy export to other formats
• Cloud File Storage: Uses Convex for secure file uploads, storage, and management

🧠 How It Works

QuestGen-AI employs a sophisticated multi-agent system powered by LangChain and large language models:

1. Extractor Agent: Analyzes your input requirements to understand the exam type, total marks, question types, and subject areas

2. Question Creator Agent: Generates appropriate questions based on the extracted requirements and the content of the uploaded PDFs

3. Question Analysis Agent: Evaluates the generated questions for quality, clarity, relevance, and alignment with requirements

4. Decider Agent: Determines if the questions meet all requirements or need further refinement

5. Formatter Agent: Arranges the questions into a professionally formatted exam paper with clear sections and instructions

This multi-step process ensures high-quality, relevant questions that align with your specific needs and content.

🛠️ Technology Stack

• Frontend: Next.js, React, TailwindCSS, Framer Motion
• Backend: Next.js API routes
• AI Integration: LangChain, OpenRouter API (compatible with various LLM models)
• PDF Processing: PDF.js, pdf-parse
• Styling: Custom UI components with Tailwind CSS and Framer Motion animations
• Streaming: Server-Sent Events (SSE) for real-time content streaming
• Database & Storage: Convex for file uploads, storage, and management

📋 Requirements

• Node.js (v18.0 or later)
• npm or yarn package manager
• OpenRouter API key (for accessing AI models)
• Convex account (for file storage)
• Modern web browser

🔧 Installation & Setup

1. Clone the repository

   https://github.com/cRED-f/QuestGen-AI-Agent.git
   cd QuestGen-AI

2. Install dependencies

   npm install
   # or
   yarn install

3. Set up environment variables

   Create a .env.local file in the project root with:

   NEXT_PUBLIC_OPENROUTER_SITE_URL="http://localhost:3000"
   OPENROUTER_API_BASE="https://openrouter.ai/api/v1"
   NEXT_PUBLIC_CONVEX_URL="your_convex_deployment_url"
   

4. Initialize Convex

   npx convex init

   Follow the prompts to connect to your Convex account and create a new deployment.

5. Run the development server

   npm run dev
   # or
   yarn dev

6. Access the application

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

💻 Usage Guide

Generating an Exam Paper

1. Navigate to the Question Generator page

   From the homepage, click "Start For Free" or navigate to /chat route.

2. Upload Content

   Use the file upload area to drag and drop or select PDF documents containing the content from which questions should be generated.

3. Enter Question Paper Parameters

   • Question Header: Enter institutional information, exam details, and course information

     • Example: "University of Technology - Midterm Examination - CS101: Introduction to Programming - Spring 2023"

   • Question Description: Provide instructions and parameters for the question paper

     • Example: "Create a question paper with 5 multiple-choice questions (2 marks each), 3 short-answer questions (5 marks each), and 1 essay question (10 marks). Focus on Python basics and data structures."

4. Enter API Key

   Provide your OpenRouter API key in the designated field. This is used to access the AI models that power the question generation.

5. Select Model (Optional)

   Choose the AI model to use for generation. The default is "qwen/qwq-32b:free".

6. Generate Questions

   Click the "Generate Question Paper" button to start the process. The system will:

   • Process your PDF documents
   • Extract relevant content
   • Analyze requirements
   • Create appropriate questions
   • Format a complete exam paper

7. View and Export Results

   Once generation is complete, you can:

   • Review the generated questions
   • Copy the content
   • Print or save the question paper

📁 Project Structure

/
├── app/                      # Next.js app directory
│   ├── (root)/               # Root route group
│   │   ├── chat/             # Question generation page
│   │   └── page.tsx          # Home page
│   ├── api/                  # API routes
│   │   └── generate-questions/ # Question generation API
│   ├── assets/               # Static assets
│   ├── globals.css           # Global styles
│   └── layout.tsx            # Root layout component
├── components/               # Reusable React components
│   ├── eldoraui/             # Custom UI components
│   ├── ui/                   # Base UI components
│   └── About.tsx             # About component
├── convex/                   # Convex database schema and functions
│   ├── _generated/           # Auto-generated Convex type definitions
│   ├── files.ts              # File storage operations
│   └── schema.ts             # Database schema
├── lib/                      # Utility functions
│   ├── prompts/              # AI prompt templates
│   └── utils.ts              # Helper functions
├── public/                   # Public assets
├── services/                 # Core services
│   └── index.ts              # Multi-agent question generation service
├── temp/                     # Temporary directory for uploaded files
├── .env.local                # Environment variables
├── next.config.ts            # Next.js configuration
├── package.json              # Project dependencies
├── postcss.config.js         # PostCSS configuration
└── tailwind.config.ts        # Tailwind CSS configuration

🔄 API Documentation

POST /api/generate-questions

Uploads PDF files and initiates question generation.

Request:

• HTTP Method: POST
• Content-Type: multipart/form-data
• Body:
  • file-{index}: PDF files
  • questionHeader: String - Exam header details
  • questionDescription: String - Question paper requirements
  • apiKey: String - OpenRouter API key
  • modelName: String (optional) - AI model name

Response:

• Status: 200 OK
• Body:

  {
    "message": "success",
    "uploadedFiles": ["1234567890.pdf"]
  }

GET /api/generate-questions

Generates questions based on previously uploaded files.

Request:

• HTTP Method: GET
• Query Parameters:
  • questionHeader: String - Exam header details
  • questionDescription: String - Question paper requirements
  • apiKey: String - OpenRouter API key
  • modelName: String (optional) - AI model name
  • uploadedFiles: String - Comma-separated list of uploaded file names

Response:

• Content-Type: text/event-stream
• Events:
  • data: Contains generated content chunks
  • error: Contains error information if generation fails
  • complete: Signals generation completion

🔒 Security Considerations

• QuestGen-AI does not store your API keys permanently
• Uploaded PDF files are securely stored in Convex and automatically deleted once question generation is complete
• No user data is retained between sessions
• API keys are transmitted securely but should still be protected carefully

🔮 Future Improvements

• User accounts for saving and organizing generated question papers
• Additional question formats and customization options
• Pre-built templates for common exam types
• Batch generation of multiple question papers
• Integration with learning management systems
• Offline mode using local models

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

Built with ❤️ using Next.js, LangChain, Convex, and OpenRouter