# JackHopkins/factorio-learning-environment

A non-saturating, open-ended environment for evaluating LLMs in Factorio

Repository: https://github.com/JackHopkins/factorio-learning-environment
Canonical: https://ross.abutalabs.com/products/factorio-learning-environment
Homepage: https://jackhopkins.github.io/factorio-learning-environment/versions/0.3.0.html
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-06-11T22:52:19+00:00

## Health v2 (maintenance only)
Score: 82/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 87, release rhythm 66, longevity 100
- inputs: {"age_days": 1901, "days_push": 83, "days_rel": 149, "gap_med": 35.5, "n_releases_24m": 7}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1156, forks 95 (observed 2026-08-28T04:03:47.949133+00:00)

## What it is
An open-source framework for developing and evaluating LLM agents in the game of Factorio, providing an open-ended, non-saturating benchmark environment. Agents interact via Python code synthesis through a REPL pattern, with OpenAI Gym compatibility, headless scaling, and MCP support.

## Use cases
- evaluate LLM agents on long-horizon planning tasks
- benchmark frontier models in an open-ended game environment
- run multi-turn interactive agent experiments with a gym interface
- research multimodal agents with pixel observations
- test agent recovery and adaptation in dynamic environments
- connect Claude Code or other agents to Factorio via MCP

## When to choose
- you need a challenging, non-saturating eval for frontier LLM agents
- you want a gym-compatible interactive environment for agent research
- you need scalable headless game-based evaluation with Docker

## When to avoid
- you need a simple static benchmark rather than an interactive game environment
- you cannot run Docker or manage Factorio server clusters
- your focus is not agent evaluation or long-horizon reasoning research

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, machine-learning, benchmarking, simulation, sdk, cli, mcp
- domain: artificial-intelligence, large-language-models, reinforcement-learning, developer-tools
- platform: python, windows, cli
- tags: llm-evaluation, factorio, agent-benchmark, open-ended-eval, gym-environment, code-synthesis, research, ai-agents, game-development, docker, linux, macos

## Member repositories
- JackHopkins/factorio-learning-environment (main) score 82

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:47.949133+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 2026-08-30T06:32:05.488306+00:00, confidence not recorded.
  - readme: https://github.com/JackHopkins/factorio-learning-environment (fetched 2026-08-28T04:03:47.949133+00:00, sha 220f37957b57)
  - homepage: https://jackhopkins.github.io/factorio-learning-environment/versions/0.3.0.html (fetched 2026-08-29T12:37:05.854989+00:00, sha 9aed546e0d18)
- Data as of 2026-08-30T08:39:29.467469+00:00.