ChessArena AI

Open-source Motia showcase for real-time LLM chess benchmarking using move quality, centipawn swing, and blunder detection.

ChessArena.ai is an open-source project from Motia that benchmarks how LLMs play chess. Instead of scoring only wins and losses, it evaluates move quality and game insight.

Repository: MotiaDev/chessarena-ai

Why this showcase matters

ChessArena demonstrates a practical Motia architecture with:

Real-time streaming updates for games and scores
Event-driven orchestration across HTTP entrypoints and worker logic
Multi-language runtime support (TypeScript + Python)
Objective move grading using Stockfish

The project README describes the core evaluation model:

Compare each move against Stockfish's recommended move
Track the delta as centipawn swing
Mark moves with swing greater than 100 centipawns as blunders

Repo structure (current)

The current repository includes:

api/ for backend and worker logic
app/ for frontend
types/ for shared types
public/images/ assets and demo media

This aligns with a split architecture where UI and workflow execution are separated, while streams/events keep state synchronized.

Tech and runtime prerequisites

According to the project README and package scripts, local setup requires:

Node.js 22+
PNPM
Python 3
uv
Stockfish

Install and run:

git clone https://github.com/MotiaDev/chessarena-ai.git
cd chessarena-ai
pnpm install
pnpm dev

Stockfish options in the project docs include:

brew install stockfish (macOS)
pnpm install-stockfish <platform>

What to learn from this project

If you are building similar workloads in Motia, ChessArena is a good reference for:

Low-latency real-time UX with stream updates
Queue-driven orchestration between gameplay steps
Integrating Python-based evaluation into a TypeScript-first codebase
Measuring AI quality with deterministic external scoring

Why this showcase matters

Repo structure (current)

Tech and runtime prerequisites

What to learn from this project

Links

On this page