Slay

Hex-based strategy game recreation with AI player development.

Repo: https://github.com/chrisaacson69/slay Links: Slay-C (C port), Slay — Evaluation & Search (the 1v1 eval + search theory), The Multiplayer Coalition Problem (the N≥3 coalition wall Slay’s AI hits), Gaming, Vault Index

Overview

Recreation of Slay, a turn-based hex strategy game where players expand territories, manage economies, and command units to eliminate opponents. The goal is a clean headless engine suitable for AI self-play and tree-based search.

Reference: OpenSlay (Java) — ~65-70% complete recreation, used for rules reference.

Key design: Engine is fully separated from renderer. The engine can run headless for self-play, AI training, and testing. GameState.clone() supports tree-based search.

Roadmap

• Core engine (units, grid, territories, combat, actions)
• Pygame-ce renderer with hex grid display
• Random AI baseline
• Greedy heuristic AI
• Alpha-beta tree search AI
• AI vs AI arena
• Improve map generation (multi-player, balanced starts)
• Polish renderer (sprites, animations, territory info panel)
• Headless self-play harness for AI training
• Performance optimization for deep search

See full architecture, mechanics, and running instructions in the repo README.

Tags

python, ai, games, game-ai