This repository is a curated, continuously updated list of academic papers focusing on Large Language Model (LLM)-based game agents. It serves researchers and developers exploring the intersection of LLMs and game AI, providing a structured overview of advancements across various game genres and agent capabilities.

How It Works

The collection categorizes papers by game type (e.g., Adventure, Simulation, Competition, Communication, Action) and specific game titles (e.g., Minecraft, NetHack, StarCraft II). Each entry typically includes a link to the paper and, where available, associated code repositories, facilitating direct access to research and implementation details.

Quick Start & Requirements

This is a curated list of papers, not a software package. No installation or execution is required.

Highlighted Details

• Comprehensive coverage of LLM applications in diverse game genres, from text adventures to complex 3D environments.
• Includes papers on foundational LLM agent techniques like ReAct, Reflexion, and planning strategies.
• Features research on embodied agents, multi-agent coordination, and LLMs in competitive and cooperative gaming scenarios.
• Regularly updated, with the last update noted as March 27, 2025.

Maintenance & Community

The project is maintained by git-disl and welcomes Pull Requests for new paper submissions. Contact information for Sihao Hu (sihaohu@gatech.edu) is provided for suggestions. The project also encourages citing their survey paper for updates.

Licensing & Compatibility

The repository itself does not have a specific license mentioned, but it links to external academic papers and code repositories, which will have their own respective licenses.

Limitations & Caveats

This resource is a survey and does not provide executable code or a unified framework. Users must consult individual paper links for specific implementation details, dependencies, and licensing.