Reproduce “CriticaLLM: Using LLMs as Autonomous Game Master for Tabletop Role-Playing Games” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Gaëtan Berlaimont, Justin Vanwichelen, Benoît Ronval, Hélène Verhaeghe

CriticaLLM coordinates a tabletop role-playing adventure through a GM LLM that runs the story and a Judge LLM that screens player actions against game state, rules, and theme; summaries and location-specific memory manage context. In a three-room, D&D-inspired system, nine experts often rated theme and coherence positively but reported limited choice impact. Rewording tests also showed model-dependent Judge failures.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.