MiroFish: A Universal Swarm Intelligence Engine for Predicting Anything

Introduction

MiroFish is a cutting-edge AI prediction engine designed to simulate and forecast future events across diverse domains. It utilizes multi-agent technology to create high-fidelity parallel digital worlds. Within these simulations, thousands of intelligent agents, each with independent personalities, long-term memory, and behavioral logic, freely interact and undergo social evolution.

By extracting seed information from the real world, such as news, policy drafts, or financial signals, MiroFish allows users to inject dynamic variables from a "God's-eye view." This capability enables precise deduction of future trajectories, effectively allowing you to "rehearse the future in a digital sandbox and win decisions after countless simulations." From serious predictions for decision-makers to playful simulations for individual creativity, MiroFish aims to make it possible to predict anything.

Installation

MiroFish offers flexible deployment options, with source code deployment being the recommended method. Ensure you have the necessary prerequisites before starting.

Prerequisites

• Node.js: Version 18+ (includes npm)
• Python: Version ?3.11, ?3.12
• uv: Latest version (Python package manager)

1. Configure Environment Variables

First, copy the example configuration file and fill in your API keys:

cp .env.example .env

Edit the .env file to include your LLM API configuration (supports any OpenAI SDK format API, Alibaba Qwen-plus recommended) and Zep Cloud configuration.

2. Install Dependencies

Install all project dependencies with a single command:

npm run setup:all

3. Start Services

Start both the frontend and backend services from the project root:

npm run dev

The frontend will be available at http://localhost:3000 and the backend API at http://localhost:5001.

Docker Deployment

Alternatively, you can deploy MiroFish using Docker:

# 1. Configure environment variables (same as source deployment)
cp .env.example .env

# 2. Pull image and start
docker compose up -d

This will read the .env file from the root directory and map ports 3000 (frontend) and 5001 (backend).

Examples

MiroFish offers compelling demonstrations of its predictive capabilities through a live demo and detailed video examples.

• Live Demo: Explore the online demo environment for a public opinion event simulation: MiroFish Live Demo
• Wuhan University Public Opinion Simulation: Watch a complete demo video for prediction using a generated public opinion report: Watch on Bilibili
• Dream of the Red Chamber Lost Ending Simulation: See MiroFish's deep prediction of a lost ending based on hundreds of thousands of words from the novel: Watch on Bilibili

Why Use MiroFish?

MiroFish stands out as a versatile tool for both strategic decision-making and creative exploration. For decision-makers, it acts as a rehearsal laboratory, enabling risk-free testing of policies and public relations strategies. At the micro level, it serves as a creative sandbox for individual users, allowing exploration of imaginative scenarios and novel endings, making everything fun, playful, and accessible.

Its ability to capture collective emergence triggered by individual interactions breaks through the limitations of traditional prediction, offering deep insights into "what if" scenarios across various fields, from financial prediction to political news.

Links

• GitHub Repository: 666ghj/MiroFish
• Live Demo: MiroFish Live Demo
• Join Discord: MiroFish Discord
• Follow on X (Twitter): MiroFish AI on X