Skip to content

We build [idea] in only 45 minutes using claude code in Anthropic's Hackathon @umd. Everything here 100% prompt engineered(Exciting to see whats the come).

Notifications You must be signed in to change notification settings

R4ushan/promptEngineer

Repository files navigation

RizzX - AI-Powered Conversational Training Simulator | Anthropic Hackaton

An innovative chat application where cs students practice their conversational skills ("rizz") by messaging AI personas, with real-time AI coaching feedback after every message.

Core Concept

We challenged ourselves to fully prompt-engineer this project — building an app that relies entirely on AI instructions rather than hard-coded logic(was not easy!)

This approach turned out to be much harder than expected.
Through the process, we learned that:

  • Prompt engineering doesn’t replace real coding fundamentals.
  • Safe and structured AI use still requires technical competence.
  • AI is a powerful tool — but only in the hands of developers who understand it.

🧩 Setup Flow

  • Choose gender (Guy/Girl) to unlock gender-specific personas
  • Select from personas like Witty Barista ☕, Shy Classmate 🎓, or Dating App Match 💕
  • Set your goal: Get their number, Ask for a date, Be funny, or Make a new friend
  • Pick your difficulty: Easy / Medium / Hard
  • Choose your AI model: Claude Sonnet 4 or Claude Opus 4

🧑‍🏫 Coach-First Feedback System

  • Send a message → Instantly receive coaching before it’s sent
  • AI Coach gives:
    • Rizz Rating (X/10)
    • Detailed critique
    • Actionable improvement tips
  • Continue the chat and improve dynamically

🧠 Smart Outcome Detection

  • Detects successful goals (Victory screen 🎉)
  • Detects failures (Game Over screen 💀)
  • Dynamic, context-aware feedback based on conversation flow

🎨 Design & Experience

  • Instagram DM–style chat UI
  • Purple-pink gradient theme
  • Smooth animations and transitions
  • Fully mobile-responsive and modern design

Installation

  1. Clone or download this repository

  2. Install dependencies

    npm install
  3. Set up your API key

    Create a .env file in the root directory:

    cp .env.example .env

    Edit .env and add your Anthropic API key:

    VITE_ANTHROPIC_API_KEY=your_api_key_here
    
  4. Run the development server

    npm run dev
  5. Open your browser

    Navigate to the URL shown in your terminal)

📁 Project Structure

promptEngineer/
├── src/
│   ├── components/          # React components
│   │   ├── SetupScreen.tsx
│   │   ├── GenderSelect.tsx
│   │   ├── PersonaSelect.tsx
│   │   ├── GoalSelect.tsx
│   │   ├── DifficultySelect.tsx
│   │   ├── ModelSelect.tsx
│   │   ├── ChatScreen.tsx
│   │   ├── CoachModal.tsx
│   │   ├── VictoryScreen.tsx
│   │   └── GameOverScreen.tsx
│   ├── hooks/               # Custom React hooks
│   │   ├── useGameState.ts
│   │   └── useClaudeService.ts
│   ├── services/            # API services
│   │   └── claudeService.ts
│   ├── data/                # Static data
│   │   ├── personas.ts
│   │   └── goals.ts
│   ├── types/               # TypeScript types
│   │   └── index.ts
│   ├── App.tsx              # Main app component
│   ├── main.tsx             # Entry point
│   └── index.css            # Global styles
├── package.json
├── tsconfig.json
├── tailwind.config.js
├── vite.config.ts
└── README.md

🎮 How to Play

  1. Start the app and click "Start Training"
  2. Choose gender of the AI persona you want to talk to
  3. Select a persona from the available options
  4. Set your goal for the conversation
  5. Choose difficulty level (Easy, Medium, or Hard)
  6. Select AI model (Sonnet 4 or Opus 4)
  7. Start chatting! Type a message and send it
  8. Review AI Coach feedback after each message
  9. Continue the conversation and try to achieve your goal
  10. Win or learn from the experience!

🛠️ Tech Stack

  • Frontend Framework: React 18.2 with TypeScript
  • Styling: Tailwind CSS 3.3
  • Build Tool: Vite 5.0
  • AI Integration: Anthropic Claude API (Sonnet 4 & Opus 4)
  • Icons: Lucide React
  • State Management: Custom React hooks

🔒 Security

  • API keys are stored in .env files (never committed to git)
  • .gitignore configured to exclude sensitive files
  • Browser-based API calls (Note: For production, implement a backend proxy)

⚠️ Important Notes

  • API Usage: This app uses the Claude API which may incur costs. Monitor your usage.
  • Browser Warning: The app uses dangerouslyAllowBrowser: true for the Anthropic SDK. In production, you should implement a backend proxy to handle API calls securely.
  • Model Access: Ensure your API key has access to the selected Claude models.

🎨 Customization

Adding New Personas

Edit src/data/personas.ts:

{
  id: 'unique-id',
  name: 'Name',
  emoji: '☕',
  description: 'Description',
  gender: 'guy' | 'girl',
  avatar: '👨'
}

Adding New Goals

Edit src/data/goals.ts:

{
  id: 'goal_id',
  title: 'Goal Title',
  description: 'Goal description',
  icon: '📱'
}

Customizing Colors

Edit tailwind.config.js to change the color scheme.

🐛 Troubleshooting

Issue: API key not working

  • Ensure your .env file is in the root directory
  • Restart the dev server after adding the API key
  • Verify your API key is valid and has the correct permissions

Issue: Build errors

  • Delete node_modules and package-lock.json
  • Run npm install again
  • Ensure you're using Node.js v18 or higher

Issue: Models not responding

  • Check your API key has access to Claude Sonnet 4 and Opus 4
  • Check the browser console for error messages
  • Verify your internet connection

🙏 Acknowledgments

  • Built with Claude AI assistance
  • Powered by Anthropic's Claude API
  • Inspired by modern dating apps and conversation training tools

💡 Future Enhancements

  • Backend proxy for secure API calls
  • User accounts and progress tracking
  • More personas and scenarios
  • Conversation history and analytics
  • Multiplayer mode
  • Voice chat integration

Made with 💜 by the RizzX Team

Practice makes perfect. Build your confidence! 💪✨

About

We build [idea] in only 45 minutes using claude code in Anthropic's Hackathon @umd. Everything here 100% prompt engineered(Exciting to see whats the come).

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •