Wintermute IRC Bot

An advanced AI-powered IRC bot inspired by William Gibson's Wintermute, featuring dynamic personality adaptation and intelligent conversation management.

Features

Core Functionality

• Dynamic Personality System: Automatically analyzes channel activity and adapts personality to match community tone.
• Intelligent Topic Tracking: Organizes conversations by topic with automatic expiration.
• Multi-LLM Support: Primary Anthropic Claude integration with OpenAI fallback.
• Context-Aware Responses: Maintains conversation context and references recent discussions.
• Admin Controls: Comprehensive management commands for bot configuration.

Advanced Capabilities

• Conversation Analysis: Deep sociolinguistic analysis of channel activity.
• Automatic Prompt Generation: Creates tailored personality directives based on channel behavior.
• Topic Threading: Groups related messages and maintains conversation flow.
• User Ignore System: Flexible user management with persistent ignore lists.
• Activity Logging: Comprehensive interaction logging for debugging and analysis.

Architecture

The bot consists of two main components:

1. Wintermute Bot (wintermute.py)

The main IRC bot that handles real-time interactions:

• Connects to IRC networks and manages channel presence.
• Processes messages and maintains conversation context.
• Generates responses using Claude (with OpenAI fallback).
• Manages topic threading and user interactions.

2. Prompt Generator (prompt_generator.py)

Analyzes channel activity and generates dynamic personality directives:

• Processes WeeChat IRC logs for recent activity.
• Performs sociolinguistic analysis of conversations.
• Generates contextual personality directives.
• Archives analysis results for reference.

Setup

Prerequisites

• Python 3.7+
• IRC network access
• API keys for Anthropic Claude and OpenAI

Installation

1. Clone the repository:

   git clone [https://github.com/yourusername/wintermute-irc-bot.git](https://github.com/yourusername/wintermute-irc-bot.git)
   cd wintermute-irc-bot

2. Install dependencies:

   pip install openai anthropic python-dotenv irc

3. Create a .env file with your configuration:

   # IRC Configuration
   IRC_SERVER=irc.yourserver.org
   IRC_PORT=6667
   IRC_CHANNELS=#yourchannel
   IRC_BOT_NICKNAME=wintermute
   IRC_BOT_PASSWORD=your_bot_password
   IRC_ACCOUNT_NAME=wintermute
   
   # API Keys
   ANTHROPIC_API_KEY=your_anthropic_api_key
   OPENAI_API_KEY_WINTERMUTE=your_openai_api_key
   OPENAI_API_KEY_PROMPT_GEN=your_openai_api_key_for_analysis

Configuration

Bot Configuration (wintermute.py)

• Modify admin username in the code (search for adminName).
• Adjust response parameters and personality settings.
• Configure channel-specific behaviors.

Prompt Generator Configuration (prompt_generator.py)

• Update WEECHAT_LOG_FILE_LOCAL_PATH to point to your IRC logs.
• Set CHANNEL_NAME_IN_LOG to match your channel.
• Adjust analysis parameters and token thresholds.

Usage

Running the Bot

python wintermute.py

Running Prompt Generation

python prompt_generator.py

For automated personality updates, set up a cron job:

# Run every hour
0 * * * * /path/to/python /path/to/prompt_generator.py

Bot Commands

User Commands

• wintermute: <message> - Direct conversation with the bot.
• wintermute: topics - Show active conversation topics.
• wintermute: help - Display available commands.

Admin Commands (require admin privileges)

• wintermute: clear topics - Clear conversation context.
• wintermute: set system_prompt <text> - Update bot personality.
• wintermute: show prompt - Display current system prompt.
• wintermute: ignore <user> - Add user to ignore list.
• wintermute: unignore <user> - Remove user from ignore list.
• wintermute: show ignored - List ignored users.

How It Works

Dynamic Personality System

• Log Analysis: Analyzes recent IRC activity using GPT models.
• Pattern Recognition: Identifies communication styles, emotional tones, and topics.
• Directive Generation: Creates personality instructions tailored to channel culture.
• Adaptive Response: Bot behavior automatically adjusts to match community.

Topic Management

• Messages are automatically categorized by topic using AI.
• Related conversations are grouped together.
• Topics expire after periods of inactivity.
• Context is maintained across topic switches.

Response Generation

• Primary responses use Anthropic Claude for high-quality conversation.
• OpenAI serves as a fallback for reliability.
• Responses are contextually aware and reference recent discussions.
• The bot maintains a consistent personality while adapting to the conversation flow.

File Structure

wintermute-irc-bot/
├── wintermute.py              # Main bot application
├── prompt_generator.py        # Dynamic personality generator
├── current_bot_directive.json # Current personality directive
├── directive_archive/         # Historical personality directives
├── wintermute_logs.txt        # Bot interaction logs
├── ignore_list.json           # User ignore list
├── archived_summaries.json    # Topic summaries
└── README.md                  # This file

Advanced Features

Conversation Analysis

The bot performs a deep analysis of channel activity including:

• Overall atmosphere and emotional tones.
• Communication patterns and formality levels.
• Key discussion topics and participant roles.
• Notable moments and memorable quotes.

Personality Adaptation

• Automatically adjusts to match channel culture.
• References past conversations and user quirks.
• Maintains a core identity while adapting its communication style.
• Preserves community-specific humor and references.

Contributing

• Fork the repository.
• Create a feature branch.
• Make your changes.
• Submit a pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Inspired by William Gibson's Wintermute AI from the Sprawl trilogy.
• Built with Anthropic Claude and OpenAI language models.
• Uses the irc library for IRC connectivity.

Support

If you encounter issues or have questions:

• Check the logs in wintermute_logs.txt.
• Review the configuration in your .env file.
• Open an issue on GitHub with relevant error messages.