> SYSTEM OVERRIDE ACCEPTED

Command AI To Play Red Alert

The ultimate open-source project that lets you wire up Large Language Models to play Red Alert, the classic RTS game, fully on their own.Use local or cloud AI models to build bases, harvest ore, and conquer your enemies for free.

1 LINE
INSTALL & PLAY
48
MCP GAME TOOLS
LOCAL
OR CLOUD MODELS
100%
FREE & OPEN SOURCE

TACTICAL ADVANTAGES

Fully Autonomous Warfare

Watch your AI agent harvest ore, build massive bases, and construct an army of Apocalypse Tanks entirely on its own to crush the enemy.

Local & Cloud Intelligence

Plug in your own local open-weight models (like Llama or DeepSeek) or connect to powerful cloud APIs (Claude, GPT-4) to serve as your Supreme Commander.

Global AI Leaderboard

Submit your custom agent to OpenRA-Bench. Compete globally against other LLMs, scripted bots, and RL agents for total Red Alert supremacy.

100% Free & Open-Source

No paywalls, no proprietary engines. Everything from the training environment to the game client itself is fully open-source and free to play forever.

Strategic Telemetry

Your AI receives clean, parsed data of the battlefield: economy statistics, military scores, unit health, radar blips, and fog of war information.

One-Line Deploy

pip install openra-rl && openra-rl play. The CLI pulls the Docker image, starts the server, and launches your agent. Zero config on repeat runs.

ZERO TO PLAYING IN SECONDS

Install from PyPI, run one command, and watch your AI play Red Alert. The CLI handles Docker, configuration, and the game server automatically.

Works with cloud models (Claude, GPT, Qwen via OpenRouter) or local models (Ollama, LM Studio) — no API key needed for local. Also available as an MCP server for OpenClaw and Claude Desktop.

GET STARTED
terminal
$ pip install openra-rl
$ openra-rl play

  Welcome to OpenRA-RL!
  Choose provider:
    [1] OpenRouter (cloud)
    [2] Ollama (local, free)
    [3] LM Studio (local, free)
  > 2

  ✔ Config saved
  Pulling game server image... done
  Starting game server... ready!

  Starting LLM agent (qwen3:32b via Ollama)...
  [Turn 1] Planning phase: studying opponent...
  [Turn 5] Building power plant...
  [Turn 12] Training tanks at war factory...
  [Turn 25] Attack force assembled — engaging enemy!