// agents · 666ghj/MiroFish

MiroFish

MiroFishは、666ghjが公開しているオープンソースのAIエージェントプロジェクトです(GitHub: 666ghj/MiroFish)。GitHubスターは85,663、フォークは12,992で、AGPL-3.0ライセンスの下で公開されています。GitHubでの最終更新は2026年10月16日です。

A swarm-intelligence prediction engine: thousands of agents rehearse the future in a digital world

★ 85.7k スター⑂ 13.0k フォークAGPL-3.0更新 2026年10月
▦ ビジュアル — このプロジェクトのREADMEには埋め込み可能なスクリーンショットがないため、
ここではプレースホルダーを表示しています。下のガイド自体は完全な内容です。
// プロジェクトの基本情報

プロジェクト基本情報

リポジトリ
666ghj/MiroFish
GitHubスター
85,663
フォーク
12,992
ライセンス
AGPL-3.0
GitHubでの最終更新
2026年10月16日
掲載タイプ
完全なインストールガイド
公式サイト
Official site
// 仕組み

仕組み

  1. Clone the repo

    Get the source, which needs Node 18+, Python 3.11–3.12 and uv.

  2. Configure your keys

    Copy the example environment file and fill in your LLM API key.

  3. Install everything

    One command installs root and frontend Node dependencies and the Python backend.

  4. Start it

    Run frontend and backend together, then open the app.

// 機能

選ばれる理由

A parallel digital world

Seed information becomes a high-fidelity simulated world for agents to live in.

Thousands of agents

Agents with independent personalities, long-term memory and behavioral logic, evolving socially.

God’s-eye variables

Inject variables dynamically to steer the simulation and deduce trajectories.

Prediction reports

A detailed report plus an interactive digital world to explore.

Any OpenAI-format LLM

Works with any LLM API that speaks the OpenAI SDK format.

Rehearse, don’t guess

A rehearsal laboratory for decision-makers: run the future many times before deciding.

// 自分で実行

自分で実行してみる

  1. Step 1 — Clone the repo

    Download the source and move into the folder.

    git clone https://github.com/666ghj/MiroFish.git
    cd MiroFish
  2. Step 2 — Configure keys

    Copy the example environment file, then fill in your LLM API key.

    cp .env.example .env
  3. Step 3 — Install and start

    Install all dependencies, then start frontend and backend.

    npm run setup:all
    npm run dev
// 活用例

作れるもの

  • Rehearse a decision in a simulated world before committing to it.
  • Explore how a policy draft or news event might play out through agent interaction.
  • Forecast financial signals with a swarm instead of a single model.
  • Build a digital sandbox where thousands of agents evolve socially.

For comparison, the same shelf also holds deepseek-harness — DeepSeek's plugin-based agent harness where and graphify — Turn any codebase, docs and schemas into a. The complete list is the AI Agents shelf. Before running a model locally, check how much VRAM you need for local LLMs — model size has to match your hardware.

// よくある質問

よくある質問

What is MiroFish?

A multi-agent prediction engine: it builds a parallel digital world from real seed information and lets thousands of agents interact to rehearse possible futures.

Are its predictions reliable?

No — they are simulations, not validated forecasts. Treat them as rehearsals for decisions, not certainties.

What do I need to run it?

Node.js 18+, Python 3.11–3.12, uv, and an LLM API key in the OpenAI SDK format.

Is it free?

Yes, AGPL-3.0 licensed. A free monthly quota is sufficient for simple usage, per the README.

Is the README in English?

The README is Chinese-first with an English version; the interface is in English.

// さらに探索

似ているツール

AIエージェント

deepseek-harness

DeepSeek's plugin-based agent harness where everything is a plugin.

★ 245.5k スターMIT
ガイドを見る →
開発ツール

graphify

Turn any codebase, docs and schemas into a queryable knowledge graph for your coding assistant

★ 124.7k スターApache-2.0
ガイドを見る →
開発ツール

orca

The agent development environment for running a fleet of parallel coding agents in worktrees

★ 87.8k スターMIT
ガイドを見る →
AIエージェント

MiroFish-Offline

Offline multi-agent simulation and prediction engine: Neo4j plus Ollama, local.

★ 2.6k スターAGPL-3.0
ガイドを見る →
AIエージェント

gortex

Gortex, in Go — built for high-performance code-intelligence engine for ai agents.

★ 1.9k スターApache-2.0
ガイドを見る →
AIエージェント

rill

Blazing-fast business intelligence built for humans and AI agents alike.

★ 2.9k スターApache-2.0
ガイドを見る →
// 代替候補

代替候補: AIエージェント

LibreAIカタログに掲載されたAIエージェントプロジェクトの中で最もスターの多いもの — MiroFishが合わなかった場合に検討したい、現実的な候補リストです。

  • openclaw★ 391.6k

    A personal AI that really does things — on any OS, with your data staying yours.

  • Skills For Real Engineers★ 322.7k

    Matt Pocock's engineering skills for coding agents — grill the spec first, then build it right, MIT

  • superpowers★ 294.9k

    An agentic skills framework & software development methodology that works.

  • ECC★ 275.2k

    Agent harness optimization system with skills, memory and security for Claude Code, Codex and more

  • hermes-agent★ 252.0k

    The agent that grows with you — a Python framework for adaptive AI agents.

// さらに見る agents

さらに見る: AIエージェント

カタログの中でMiroFishとスター数が近いAIエージェントプロジェクト — 前後の棚もそのまま見て回れます。

  • OpenHands★ 90.3k

    Open-source AI-driven development platform: autonomous software agents in a sandbox

  • crawl4ai★ 84.7k

    Open-source web crawler and scraper for LLMs and AI agents: any website into clean, LLM-ready Markdown. Run it yourself, or use Crawl4AI Cloud with one key.

  • servers★ 91.1k

    The official collection of reference Model Context Protocol servers.

  • deer-flow★ 83.4k

    An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message…

  • Agent-Reach★ 94.0k

    Give your AI agent eyes on the whole internet: Twitter, Reddit, YouTube, GitHub and more, one CLI

  • AppFlowy★ 77.1k

    Bring projects, wikis, and teams together with AI. AppFlowy is the AI collaborative workspace where you achieve more without losing control of your data. The…

MiroFishを入手する

無料でオープン — ソースコードを入手して、リポジトリにスターを付け、作り始めましょう。