The first AI history strategy simulator that abolishes hardcoded branching stories. Uses an LLM (DeepSeek) directly as the game's logic simulation engine.

See how the LLM generates narratives and executes logic under the hood
Testing the boundaries of LLMs acting as autonomous strategic game brains
Abolishes hardcoded story trees. AI dynamically calculates faction growth, military campaigns, and diplomatic choices based on current database state.
Allows players to type any strategic command. The AI parses the text, calculates its logical viability, and executes the results.
The simulator measures how far you deviate from actual history. As divergence rises, the AI shifts to sandboxed logical simulation.
Faction databases and storylines are completely modular. Players can inject custom campaigns by editing simple JSON config files.
Leverages ultra-low LLM pricing. Playing a full 100-turn historical campaign costs less than 5 cents (USD 0.007).
Play in your browser, CLI, or Feishu. Future releases will support WeChat and Telegram via the OpenClaw protocol.
git clone https://github.com/emergencescience/histrategy.git cd histrategy pip install -e .
export DEEPSEEK_API_KEY="sk-you...-key"
histrategy
Custom scenario? AI-written faction chronicle? Bespoke general cards? Post a bounty on Emergence Science and let the agent network build it for you.
Copy the command below and paste it to your AI assistant (OpenClaw / Hermes Agent). It will install the game and start a campaign โ no manual setup needed.
Also try: "Play histrategy โ rome-triumvirate as Octavian, in English"
Or have your AI assistant learn the full installation & gameplay guide:
Once our GitHub repository reaches 100 stars, Emergence Science will release the OpenClaw client, allowing you to play directly via instant messengers.