# geek-ai/MAgent

A Platform for Many-Agent Reinforcement Learning

Repository: https://github.com/geek-ai/MAgent
Canonical: https://ross.abutalabs.com/products/magent
Language: Python
License: MIT
License Family: permissive
Topics: reinforcement-learning, multi-agent, deep-learning
Last push: 2022-10-22T00:21:27+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3222, "days_push": 1412, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1762, forks 334 (observed 2026-08-28T04:05:32.828459+00:00)

## What it is
MAgent is a research platform for many-agent reinforcement learning, supporting environments that scale from hundreds to millions of agents in gridworld settings like pursuit, gathering, and battle. It includes baseline algorithms (DQN, DRQN, A2C) implemented in TensorFlow and MXNet.

## Use cases
- simulate multi-agent reinforcement learning with millions of agents
- train agents in pursuit, gathering, or battle gridworld environments
- research emergent collective behavior in large agent populations
- benchmark multi-agent RL algorithms like DQN and A2C
- run an interactive battle game against trained agents

## When to choose
- you need a many-agent RL environment scaling to very large agent counts
- you want gridworld multi-agent benchmarks with baseline implementations
- you are reproducing research from the AAAI 2018 MAgent paper

## When to avoid
- you need maintained software - use the Farama-Foundation/MAgent2 fork instead
- you need Windows support or pip installation
- you work with continuous action spaces rather than gridworld settings

## Facets
- artifact type: library
- maturity: abandoned
- function: reinforcement-learning, simulation, machine-learning
- domain: reinforcement-learning, artificial-intelligence, simulation
- platform: python
- tags: multi-agent, gridworld, deep-learning, research-platform, unmaintained, linux, macos

## Member repositories
- geek-ai/MAgent (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:32.828459+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:27:01.798110+00:00, confidence not recorded.
  - readme: https://github.com/geek-ai/MAgent (fetched 2026-08-28T04:05:32.828459+00:00, sha 7217fd34d3c2)
- Data as of 2026-08-30T08:39:29.467469+00:00.