🧠 UltimateTicTacAI

An AI Agent for Ultimate Tic Tac Toe, developed using Monte Carlo Tree Search (MCTS) and Alpha-Beta Pruning algorithms.

---

🚀 Play Now (Web UI)

Try the AI directly in your browser without installing anything!
👉 https://shinnei.github.io/UltimaTTT-Agent/

The web version is fully serverless, running Python natively in your browser using Pyodide and Web Workers.

---

📌 Project Overview

This project is part of the Introduction to AI course (Semester 241) at HCMUT, where each group was required to build an AI agent that competes in a round-robin tournament playing Ultimate Tic Tac Toe.

Our goal was to explore and compare different decision-making techniques in adversarial games by implementing:

• MCTS – a search-based AI using simulations and playouts.
• Alpha-Beta Pruning – a classic adversarial search algorithm with custom evaluation functions.

---

📖 Game Rules

For English readers, please refer to Wikipedia for an overview of the game mechanics:
🔗 https://en.wikipedia.org/wiki/Ultimate_tic-tac-toe

---

🗂 What's in the Repo?

• state.py — Core logic for handling the Ultimate Tic Tac Toe board and rules.
• test_game.py — Entry point for running and visualizing AI matches.
• Alpha_Beta.py — Alpha-Beta Pruning-based AI agent with heuristic evaluations.
• MCTS.py — MCTS-based AI agent.
• index.html, style.css, script.js, worker.js — Frontend and Pyodide Web Worker files for the browser-based UI.

---

▶️ How to Run Locally

If you prefer running the code locally, there are multiple ways:

1. Web Version (Local Server)

python -m http.server 8000

Then navigate to http://localhost:8000/index.html.

2. Python Script

Use the main() function in main.py or run test_game.py to simulate a match between two agents in the terminal.

---

📬 Contact Info

• Name: Nguyen Huu Huy Thinh
• Email: dst15092004@gmail.com
• Website: https://shinnei.github.io