Multi-agent AI sandbox simulating the 2026 Strait of Hormuz crisis. 4 nations × independent LLM agents × adjustable parameters = emergent geopolitics.
<div align="center">
<img src="frontend/public/logo.png" alt="Peakstone Labs" width="200" />
# Strait of Hormuz Simulator
### Multi-Agent Geopolitical Sandbox
*Each nation is an independent AI agent. Tune the parameters. Watch the game theory emerge.*
<img src="videos/demo.gif" alt="Demo" width="700" />
[](#license)
[](https://python.org)
[](https://vuejs.org)
[**▶ Play Now**](https://game.peakstone-labs.com) · [Report Bug](https://github.com/Peakstone-Labs/hormuz-agent-sandbox/issues) · [Add a Country](#add-a-new-country)
**[English](#what-is-this)** | **[中文](#zh-cn)**
</div>
---
## What Is This?
A real-time geopolitical simulation powered by multiple AI agents, set against the **2026 Strait of Hormuz crisis** — a real, ongoing conflict as of April 2026.
Each country (Iran, US, Israel, Gulf States) is controlled by an independent LLM agent with its own:
- **Strategic doctrine** — goals, red lines, behavioral tendencies
- **Constraints** — missile stockpiles, economic tolerance, domestic politics
- **Information** — shared situation briefing + rolling memory of prior rounds
You configure the chaos level, activate scenario modifiers ("tags"), adjust each nation's parameters, and watch how the AI agents interact — producing emergent geopolitical dynamics that no single agent was programmed to create.
### Key Features
- 🎭 **4 AI-driven nations** with deeply researched behavioral profiles
- 🏷️ **11 pluggable scenario tags** (Houthi Wildcard, Nuclear Brinkmanship, China Mediator...)
- 🛢️ **Real-time oil market agent** that prices in geopolitical risk
- 🎛️ **20+ adjustable parameters** — aggressiveness, constraints, chaos factor
- 🌊 **SSE streaming** — watch each agent "think" in real-time
- 🌐 **Bilingual** — full Chinese/English support (UI + LLM output)
- 📊 **SharHAL 分层混合模型工作流 — 强模型(Claude)负责理解/拆解/验收,低成本模型(DeepSeek)负责检索/提取/清洗。Hermes Agent skill。
An LLM agent fine-tuned on DeepSeek for spaced repetition, dynamically integrating knowledge points based on the Ebbinghaus forgetting curve.
基于 STM32F103 构建的端到端 AI 智能手表生态。自研“零重定位”原生机器码动态加载引擎与页面栈式 UI 框架;集成生产级 OTA 回滚保护机制与高带宽(921600 baud)串口协议栈。通过 Node.js 中继实现 DeepSeek AI 语义控制及 ASRPRO 语音全双工交互,是一个集成了分布式计算、现代存储管理与 AI Agent 的嵌入式全栈工程。
A Meta-Agent-Driven Self-Evolving Multi-Agent System for UAV Detection and Tracking
One command to run Hermes AI Agent with a browser UI. Zero prerequisites. 一行命令,AI 就位。
网页应用Agent,接入DeepSeek、Mimo等模型