Agents & Multi-Agent

MiroFish — Multi-agent scenario simulation

666ghj/MiroFish

Explore scenarios with interacting AI agents, knowledge graphs and generated reports.

★ 72KStars
⑂ 11.2KForks
125Open issues
PythonLanguage
AGPL-3.0License
Q85Editorial score

Overview

MiroFish turns supplied materials into a simulated social environment. Its agents interact through OASIS, enabling exploration of alternative outcomes.

Key features

  • Knowledge graphs
  • Agent interaction
  • Simulation reports

Requirements, installation and quick start

Requires Node.js 18+, Python 3.11–3.12 and uv. Clone the repository, copy .env.example to .env, configure LLM_API_KEY and ZEP_API_KEY, then run npm run setup:all and npm run dev. Follow the README for Docker deployment.

Usage

Upload seed documents, describe a scenario, configure agents, run a simulation and inspect its report. Start with a small experiment.

Model compatibility and use cases

Configurable LLM endpoint; consult the repository configuration.

License and risk notes

AGPL-3.0; see the repository LICENSE.

Documentation reviewed; simulation accuracy and runtime behavior have not been independently tested. External API usage may incur costs.

Firecrawl

firecrawl/firecrawl

★ 161.1KTypeScript

LangChain

langchain-ai/langchain

★ 143.6KPython

RAGFlow

infiniflow/ragflow

★ 86.7KGo