Add new papers: LLM Economist, PokeAgent Challenge, and PokéChamp

[sethkarten]

Summary

This PR adds three new papers to the LLM-based Game Agents survey:

New Papers Added:

1. LLM Economist: Large Population Models and Mechanism Design in Multi-Agent Generative Simulacra

   • Section: Simulation Games > Human/social Simulation
   • Links: Paper + Code repository
   • Description: Framework for mechanism design using LLMs with multi-agent reinforcement learning

2. The PokeAgent Challenge: Competitive and Long Context Learning at Scale

   • Section: Competition Games
   • Links: Paper + Competition website
   • Description: NeurIPS 2025 competition benchmark featuring competitive Pokémon battling and RPG speedrunning

3. PokéChamp: An Expert-level Minimax Language Agent

   • Section: Competition Games
   • Links: Paper + Code repository
   • Description: ICML 2025 spotlight paper on competitive Pokémon agents using minimax reasoning

Changes Made:

• Added all three papers with proper publication dates
• Included relevant links (papers, code repositories, websites)
• Maintained consistent formatting with existing entries
• Placed papers in appropriate sections based on their content

Why These Papers:

These papers represent significant advances in:

• Multi-agent economic simulation with LLMs
• Competitive game AI and strategic reasoning
• Long-context planning and decision-making
• Integration of language models with game-theoretic approaches

All papers are from 2025 and include proper citations and implementation resources.