Multi-Agent Werewolf

Already contacted

Implemented a project where multiple agents play Werewolf together. Different agents play different roles (werewolf, villager, seer, witch), thinking and acting based on their settings and field information during different game phases.

Created two Agents. PlayerAgent can load different prompts at runtime to become different roles. This LLM-based agent is the game player, thinking and acting based on instructions, field info, and role settings. GodAgent communicates with all PlayerAgents, issues instructions, and advances game progress.

① PlayerAgent: Game player, thinks and acts based on instructions and field info. ② GodAgent: Advances game progress. ③ Role Prompts: Different roles have different Prompts, loaded during gameplay based on GodAgent's role assignment. ④ Environment: Records game info and PlayerAgent memories, serving as basis for thinking and action. Agent actions also affect Environment. Implemented with Redis database. ⑤ Tools: Designed for PlayerAgent to interact with Environment. ⑥ Main Game Logic: Game logic designed per Werewolf rules, advanced by GodAgent.