🏰 Knowledge Dungeon 🏰

A gamified Q&A adventure built using Streamlit, CrewAI, and LangChain with a LLM-powered cast of agents guiding the player through trivia challenges. Players battle through levels, collect points, and survive with limited lives—all while being guided by a virtual narrator and hint system.

---

🚀 Features

• ✅ Dynamic Question Generation based on player level using LLMs.
• 🧠 Semantic Answer Checking using sentence embeddings (MiniLM) for flexible answer matching.
• 💡 Hint System for each question.
• 📜 Narrator Agent for immersive storytelling.
• 🎮 Gamified Progression: 10 levels, limited lives, and point tracking.
• 💾 Persistent Question History to avoid repetition across sessions.

---

🛠️ Tech Stack

• Streamlit – UI framework.
• LangChain with Groq – LLM agent orchestration.
• CrewAI – Agent management.
• sentence-transformers – Embedding-based answer validation.
• LLM Used: meta-llama/llama-4-maverick-17b-128e-instruct (via ChatGroq).

---

🎮 How to Play

1. Launch the app:

   streamlit run app.py

2. Read the narrator's introduction.

3. Answer general knowledge questions.

   • Use the hint if you're stuck (once per question).
   • Be careful—wrong answers cost a life!

4. Score points for each correct answer.

   • Every 3 correct answers = level up.
   • Reach Level 10 to win!

---

📁 File Overview

File	Purpose
app.py	Main game logic
history.txt	Stores asked questions (avoid repeats)
README.md	You are reading it 🙂

---

🧠 Agents Breakdown

Agent	Role
NarratorAgent	Welcomes the player with story context
QuestionAgent	Creates fresh, level-based trivia questions
HintAgent	Generates helpful but vague hints
AnswerCheckerAgent	Validates answers with semantic comparison

---

📦 Dependencies

Install with:

pip install streamlit langchain crewai sentence-transformers scikit-learn