🎮 Gaming RAG Assistant

An intelligent Q&A system specifically designed for game documentation, lore, and mechanics. Built as part of the NSK AI RAG Bootcamp Phase 1 project.

🌟 Features

• Multi-format Document Support: Text, PDF, and website content
• Intelligent Retrieval: Hybrid dense + sparse retrieval with domain filtering
• Persona-based Responses: Multiple assistant personalities (Guide, Sage, Expert, Casual)
• Difficulty Levels: Adjustable response detail from brief to comprehensive
• Sample Data: Pre-loaded example documents for immediate testing
• User-Friendly Interface: Clean Streamlit UI with guided onboarding
• Deployment Ready: Optimized for cloud deployment

🚀 Quick Start

Option 1: Try Online (Recommended)

https://gamingragassistant-xlr8.streamlit.app/

Option 2: Run Locally

1. Clone the repository

git clone https://github.com/your-username/game-rag-assistant.git
cd game-rag-assistant

2. Install dependencies

pip install -r requirements.txt

3. Run the application

streamlit run gameRAG.py

4. Open your browser to http://localhost:8501

📋 Requirements

Create a requirements.txt file with the necessary requirements similar to the ones in the repo

🎯 Tech Stack

• Framework: LangChain + Streamlit
• Embeddings: HuggingFace Sentence Transformers (all-MiniLM-L6-v2)
• Vector Store: FAISS (with Chroma fallback)
• Retrieval: Hybrid (Dense + BM25 sparse retrieval)
• Optional LLM: OpenAI GPT-3.5-turbo
• Document Processing: LangChain loaders for PDF, text, web content

📖 Usage Guide

1. Initialize the System

• Click "🚀 Initialize System" to set up the RAG pipeline
• Optionally configure OpenAI API key for enhanced answers

2. Add Game Documentation

Choose from three methods:

📝 Text Documents

• Paste your game documentation directly
• Perfect for mechanics, lore, and character descriptions

📄 PDF Files

• Upload game design documents, manuals, or guides
• Automatically extracts and processes all pages

🌐 Websites

• Load content from game wikis or documentation sites
• Great for community-maintained content

3. Ask Questions

• Use natural language queries about your game content
• Adjust persona and detail level in the sidebar
• Filter by content domain (lore, mechanics, code, assets)

💡 Example Queries

After loading sample data or your own documents, try:

• "How does the combat system work?"
• "Tell me about the fantasy world setting"
• "What are the main character classes?"
• "Explain the magic system"
• "How do I unlock new areas?"

🎮 Game-Specific Features

Content Domains

• Lore: World-building, characters, story
• Mechanics: Game rules, systems, progression
• Code: Technical documentation, APIs
• Assets: Art guides, audio, resources
• General: Mixed or uncategorized content

Assistant Personas

• Helpful Guide: Clear, professional explanations
• Wise Sage: Mystical, thoughtful responses
• Tech Expert: Developer-focused technical details
• Casual Gamer: Friendly, gaming-focused language

Response Levels

• Very Brief: One sentence answers
• Brief: Short paragraphs
• Balanced: Moderate detail (default)
• Detailed: Comprehensive explanations
• Very Detailed: In-depth analysis

🔧 Configuration

Environment Variables

# Optional: For enhanced answers
OPENAI_API_KEY=your_openai_api_key_here

🚀 Deployment

Streamlit Cloud

1. Push code to GitHub repository
2. Connect to Streamlit Cloud
3. Deploy directly from your repository
4. Set environment variables in Streamlit Cloud settings

📊 Project Structure

game-rag-assistant/
├── game_rag.py    # Main application
├── requirements.txt        # Python dependencies  
├── README.md              # This file
├── packages.txt            # For necessary FAISS packages
└── .streamlit/
    └── config.toml       # Streamlit configuration

🔍 How It Works

1. Document Ingestion: Loads and processes various document formats
2. Text Chunking: Splits documents into semantic chunks with overlap
3. Embedding: Converts chunks to vector representations
4. Indexing: Stores in FAISS vector database with BM25 sparse index
5. Retrieval: Hybrid search combining dense and sparse methods
6. Generation: Uses LLM or extractive summarization for answers
7. Context: Provides source citations and metadata

🐛 Known Limitations

• PDF processing may struggle with complex layouts
• Website loading depends on site structure and accessibility
• Large documents may require chunking parameter adjustment
• OpenAI integration requires valid API key and credits

🤝 Contributing

This project was created for the NSK AI RAG Bootcamp. Feel free to fork and enhance for your own games!

📄 License

MIT License - Feel free to use for your own game projects.

🙏 Acknowledgments

• NSK AI Community for the excellent bootcamp content
• LangChain team for the comprehensive RAG framework