Skip to content
@go-mirofish

Go-Mirofish

Upload documents, describe what you want to predict, and get a full simulation report on your device.
go-mirofish logo

Go-MiroFish, lightweight and local-first

GitHub Stars License: AGPL-3.0 Go Version Python Version GitHub Sponsors Buy Me A Coffee Discord X go-mirofish - Fast local Go AI swarm engine – predict anything with agents | Product Hunt Upload documents, describe what you want to predict, and get a full simulation report on your laptop.

Quick start

Canonical development: Go gateway in Docker on :3000; Vue locally via Vite on :5173. You need Docker, Node 18+, and a one-time npm run setup.

  1. Clone

    git clone https://github.com/go-mirofish/go-mirofish.git
    cd go-mirofish
  2. Configure and install

    cp .env.example .env
    npm run setup

    Edit .env and set LLM_API_KEY and ZEP_API_KEY.

  3. Start the API (Docker)

    make up
  4. Start the UI (local — second terminal)

    npm run dev

What go-mirofish vs MiroFish

MiroFish (upstream) go-mirofish (this repo)
Control plane Python / Flask (plus JS frontend) Go (gateway/) — all /api/* routes
Local dev Python venv, Flask, often multi-service Docker gateway + local Vite (make up + npm run dev)
Simulation worker Python-side integration Go-native worker in the gateway process
Benchmarks & examples Mixed scripts Go go-mirofish-examples + bench tools + mirofish-hybrid helpers
Product Python Required on the hot path Removed (no backend/.venv in this tree)
Design goal Full MiroFish upstream feature set Local-first: lower moving parts, one gateway binary, fewer host dependencies

Note

RAM/startup “targets” depend on model provider, graph size, and simulation profile. For supported setup, see Installation.

🌐 Live Demo

📸 Screenshots


Home / entry

Simulation run

Report generation

Report timeline / tools

Simulation history

Deep interaction

Split: graph, workbench & system terminal

Graph view & node details

Hardware compatibility

Device RAM Works?
Desktop / laptop 8GB Yes
Desktop / laptop 4GB Yes (smaller simulations)
Raspberry Pi 5 4GB Yes (light workloads; validate locally)
Raspberry Pi 4 4GB Limited (expect tight headroom)

Warning

Large graphs, long simulations, or heavy models can exceed 4GB systems. Start with short runs and smaller seeds.

Contributing

Issues and PRs are welcome. Use this repo for go-mirofish changes; upstream product discussion stays with MiroFish. Longer guides for contributors and the Phase 1–6 roadmap will live on go.mirofish.ai as the docs site grows.

License

AGPL-3.0.

Acknowledgments

Derived from MiroFish. Simulation is powered by OASIS—thanks to the CAMEL-AI team.

Pinned Loading

  1. go-mirofish go-mirofish Public

    Forked from 666ghj/MiroFish

    A Simple and Universal Swarm Intelligence Engine, Predicting Anything. A hybrid Go + Python system built on top of MiroFish

    Vue 4 2

Repositories

Showing 2 of 2 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…