Lovable Prompt Creator - AI Prompt Generator

## APPLICATION OVERVIEW
This application is a web-based toolkit designed to evaluate the robustness of image-classification models against simple adversarial attacks. Its primary purpose is to help users identify and understand security risks in real-world deployments of these models, enabling them to enhance their defenses against potential vulnerabilities.

## CORE FEATURES
1. **Model Evaluation Dashboard**: An intuitive dashboard that allows users to upload and evaluate their image-classification models against various adversarial attacks. Results are displayed in an easy-to-understand format, highlighting vulnerabilities.
  
2. **Adversarial Attack Simulation**: A feature that enables users to select from different types of adversarial attacks (e.g., FGSM, PGD) and visualize the impact on model performance in real-time.

3. **Security Risk Assessment**: Automatically analyzes model performance metrics post-evaluation and generates a security risk report that outlines potential weaknesses and recommendations for improvement.

4. **User-Friendly Model Upload**: A simple drag-and-drop interface for users to upload their trained models and datasets, ensuring ease of use for all levels of technical expertise.

5. **Interactive Visualizations**: Provides interactive charts and graphs that illustrate model performance, attack effectiveness, and comparative analysis against baseline models.

6. **Documentation and Resources**: A dedicated section providing users with access to documentation, tutorials, and best practices for securing their image-classification models against adversarial attacks.

## DESIGN SPECIFICATIONS
- **Visual Style**: Minimalist design with a clean, simple aesthetic emphasizing functionality and usability. Ample white space ensures clarity and focus on content.
- **Color Mode**: Light theme featuring dark text on light backgrounds to enhance readability and reduce eye strain.
- **Layout**: 
  - A top navigation bar for easy access to core features.
  - A main content area divided into sections for model evaluation, attack simulation, and risk assessment.
  - Side panel for quick links to resources and documentation.
- **Typography**: 
  - Primary font: "Inter" for a modern and clean look.
  - Hierarchy: Use bold for headings, regular for body text, and italic for emphasis to maintain clarity and user engagement.

## TECHNICAL REQUIREMENTS
- **Framework**: React with TypeScript for building a robust and type-safe user interface.
- **Styling**: Tailwind CSS for utility-first CSS styling to facilitate rapid design changes and maintain consistency.
- **UI Components**: Utilize shadcn/ui for pre-built components that align with the minimalist aesthetic.
- **State Management**: Redux or Context API for managing application state effectively, ensuring smooth user experience.

## IMPLEMENTATION STEPS
1. **Set Up Development Environment**: Initialize a new React project with TypeScript and install Tailwind CSS and shadcn/ui.
2. **Create Layout Structure**: Implement the top navigation bar, main content area, and side panel using Tailwind CSS for styling.
3. **Develop Core Features**:
   - Build the Model Evaluation Dashboard component to handle model uploads and display results.
   - Implement the Adversarial Attack Simulation feature, including selectable attack types and real-time performance feedback.
   - Create the Security Risk Assessment logic to analyze and report vulnerabilities.
   - Develop user-friendly upload functionality with drag-and-drop capability.
   - Integrate interactive visualizations using a charting library (e.g., Chart.js or D3.js).
4. **Design UI Components**: Use shadcn/ui components for buttons, forms, and modals, ensuring a cohesive design language.
5. **Test Functionality**: Conduct unit and integration tests to ensure all features function correctly and provide a seamless user experience.
6. **Deploy the Application**: Host the application on a platform like Vercel or Netlify for public access.

## USER EXPERIENCE
Users will start at the dashboard, where they can easily upload their models and access various evaluation features. The drag-and-drop functionality streamlines the model upload process. Users select adversarial attacks from an intuitive menu, watch simulations in real-time, and receive detailed feedback on performance metrics. The security risk assessment provides actionable insights in a well-organized report format, while interactive visualizations help users grasp complex data effortlessly. Comprehensive documentation ensures users have the resources needed to enhance their image-classification models effectively.