RWWW - Reasoning with Werewolf

RWWW (Reasoning with Werewolf) is a Python-based simulation of the popular social deduction game Werewolf, where AI agents take on roles and interact within the game's logical framework. The project explores reasoning, deception, and strategic deduction in an AI-driven Werewolf game setting.

Project Overview

The project currently supports:

• Game Session Logic: The main game logic is encapsulated in game.py, which sets up a game session with different roles, prompts, and interactions.
• Daytime Discussion Phase: Players engage in discussions to deduce each other’s roles.
• Night Phase: Players conduct action and win condition is checked.
• Test Script: test.py allows you to run a sample game session.

Setup and Installation

1. Clone the Repository:

   git clone https://github.com/yourusername/RWWW-Reasoning-with-Werewolf.git
   cd RWWW-Reasoning-with-Werewolf

2. Install Dependencies: Use the requirements.txt file to install the necessary packages:

   pip install -r requirements.txt

3. Set Up OpenAI API Key: Set the OpenAI API key in your environment. You can do this by adding it to your .bashrc, .zshrc, or equivalent configuration file:

   export OPENAI_API_KEY="your_openai_api_key_here"

Running the Game

To start a game session and see the current implementation in action, run:

python test.py

The output will display players discussing who they suspect to be the Werewolf, based on the prompts generated in game.py.

Project Progress

Current Implementation

1. Daytime Discussion Phase:

   • Each player discusses and attempts to deduce who might be the Werewolf, based on role-specific prompts.

2. Role-Based Actions:

   • Players (AI agents) are assigned specific roles (Villager, Werewolf, Prophet), each with unique actions and win conditions.

3. Prompt Generation:

   • Initial prompts guide each role's behavior during discussions. This allows each AI agent to align its responses with its assigned role.

4. Night Phase:

   • Add Werewolf elimination logic and Prophet’s nightly revelation of one player’s identity. This also includes examine the win condition at the end of each night phase.

Next Steps

1. Game Logic Enhancements:

   • Develop voting mechanics and victory conditions to complete the gameplay loop.
   • Enable player elimination based on the discussions and voting outcomes.

2. Prompt Optimization:

   • Improve prompt structures to enhance each role’s reasoning and alignment with game strategies.
   • Experiment with prompt variations for better AI-driven interaction.

Contributing

Contributions to improve the game logic, prompts, or features are welcome! To contribute:

1. Fork the Repository and create a new branch for your changes.
2. Make your changes and ensure they align with the project goals.
3. Submit a Pull Request with a clear description of your updates.

License

This project is licensed under the MIT License.