crispy-chiken/YugiohAi

Implements a dual-AI architecture with a Python deck-building optimizer and C# game AI that learns optimal play through self-training against itself. Integrates with Project Ignis EDOPro simulator and Windbot framework, exposing decision weights via HTTP server for real-time game decision-making. Supports multiple training modes including Monte Carlo Tree Search rollouts, manual intervention, and configurable win-threshold stopping criteria with persistent SQLite game state logging.