Back to browse
GitHub Repository

An LLM-powered soccer simulation where every player on the field is an AI agent.

11 starsHTML

A Soccer Game Simulator Played by AI Agent

by gangtao·May 4, 2026·5 points·2 comments

AI Analysis

●●SolidRabbit HoleBig Brain

Every player is an LLM agent that evolves strategies between matches autonomously.

Strengths
  • Code generation pipeline writes player logic in Python, JS, or Rust with language-specific sandboxes.
  • Post-match evolution pipeline feeds match logs back to LLMs for iterative strategy improvement.
  • Supports full tournament modes: arena, cup brackets, and round-robin leagues with standings.
Weaknesses
  • Requires API keys for LLM providers; local Ollama is the only free option but slower.
  • Niche appeal limits broader adoption beyond AI experimentation circles.
Category
Target Audience

AI researchers, simulation enthusiasts, game developers

Similar To

Google Research Football · Neural MMO · PettingZoo

Similar Projects

Gaming●●Solid

Multi Agent World Cup Simulator

Four-agent match simulation with in-browser TTS commentary via WebGPU.

Crowd PleaserCozy
tantara
107d ago