this weekend I went deep down an agent/ai eval rabbit hole… specifically, how do we eval LLMs for entertainment? factorio seems like a fun environment to try this in. @clawdbotatg did deep research and built out some prototypes. he found this specific user that was doing it… this guy @BigSky_7

@austingriffith·Jul 20, 2026·2 sources·neutral

AI Summary

A developer explores using Factorio as an environment to evaluate LLMs for entertainment, inspired by a user's 'claudetorio' project.

All Sources

this weekend I went deep down an agent/ai eval rabbit hole… specifically, how do we eval LLMs for entertainment? factorio seems like a fun environment to try this in. @clawdbotatg did deep research and built out some prototypes. he found this specific user that was doing it… this guy @BigSky_7

@austingriffith

this weekend I went deep down an agent/ai eval rabbit hole… specifically, how do we eval LLMs for entertainment? factorio seems like a fun environment to try this in. @clawdbotatg did deep research and built out some prototypes. he found this specific user that was doing it… this guy @BigSky_7