nickslevine

nickslevine / budok-ai

Public

LLM fighting game mod of YOMI Hustle

34
3
100% credibility
Found Mar 27, 2026 at 34 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

budok-ai enables large language models to compete in the turn-based fighting game YOMI Hustle by modding the game client and using a Python daemon to solicit AI decisions.

How It Works

1
🔍 Discover AI Fighters

You stumble upon a fun project where smart AI assistants battle in a clever turn-based fighting game called YOMI Hustle.

2
🛒 Grab the Game

You buy YOMI Hustle on Steam because it's a blast even without AIs, and now you're ready to mod it for AI showdowns.

3
⚙️ Set Up Your Playground

You install a few easy tools on your Mac to run the game in a virtual playground where AIs can play safely.

4
🤖 Pick Your AI Warriors

You choose which smart AI helpers will fight, like Claude or GPT, and connect them so they can think and decide moves.

5
⚔️ Launch the Battle

With one simple command, you start the match—the AIs take turns picking moves in this rock-paper-scissors chess game.

6
📱 Watch It Unfold

A handy screen shows live stats, AI thoughts, and decisions as the fight plays out turn by turn.

🏆 Relive the Victory

The battle ends with a winner, full stats, and a video replay saved so you can share the epic AI duel.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 34 to 34 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is budok-ai?

budok-ai mods YOMI Hustle, a turn-based 2D fighting game like chess meets rock-paper-scissors, so LLMs battle head-to-head. A Python daemon coordinates responses from Anthropic, OpenAI, or OpenRouter APIs, running the game headlessly in a Linux VM on MacOS via OrbStack. One command like `scripts/run_match.sh` launches LLM vs LLM fights, records replay videos, and saves prompts/decisions for analysis.

Why is it gaining traction?

It replays LLM decisions frame-by-frame in the real game engine, exposing flaws like endless blocking or whiffed attacks—perfect for spotting strategic gaps beyond chat benchmarks. Baselines (random, greedy) work without API keys, plus a live TUI streams reasoning traces and stats. As a github llm-resources gem, it offers llm github local integration for quick experiments, no cloud needed.

Who should use this?

AI researchers benchmarking LLMs on timing/decision games, or devs building llm github actions/copilot extensions for game AI. Ideal for llm github simonw fans wanting reproducible tournaments via JSON configs, or anyone downloading YOMI Hustle to test models like Sonnet vs Opus in budok fighting mod scenarios.

Verdict

Fun llm github repository for python-based game experiments, but 34 stars and 1.0% credibility score mean it's raw—MacOS-only setup, own the $20 Steam game first. Worth a spin for llm github integration proofs if you're into yomi hustle mods; otherwise, watch for RL upgrades.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.