Discord LLM Family - Proof of Concept

A minimal proof of concept for a family of Discord bots that simulate conversations between different AI models in the Pollinations Discord community. Each bot represents a different language model with a simple unique personality.

Features

• Multiple Discord bots managed from a single codebase
• Integration with Pollinations.AI text generation API
• Simple personality framework for each bot
• Dynamic bot avatars generated using Pollinations.AI image API
• Basic conversation capabilities
• Conversation channels where bots respond to all messages (not just mentions)
• Conversation history feature for bots to see and respond to the full conversation context

Getting Started

Prerequisites

• Node.js 20+
• Discord bot tokens (create them on the Discord Developer Portal)
• Pollinations API key (if required)

Discord Bot Setup

1. Create a bot on the Discord Developer Portal
2. In the Bot section, enable all Privileged Gateway Intents:
   • Presence Intent
   • Server Members Intent
   • Message Content Intent
3. Set the bot permissions to 247872 which includes:
   • Read Messages/View Channels
   • Send Messages
   • Embed Links
   • Attach Files
   • Read Message History
   • Mention Everyone
   • Add Reactions
4. Copy the bot token for use in your .env file

Installation

1. Clone the repository
2. Install dependencies:

   npm install
   

3. Create a .env file based on .env.example and add your Discord bot tokens and Pollinations API key:

   BOT_TOKEN_1=your-discord-bot-token-1
   BOT_TOKEN_2=your-discord-bot-token-2
   POLLINATIONS_API_KEY=your-pollinations-api-key
   

Running the Application

Development mode with auto-reload:

npm run dev

Production mode:

npm run build
npm run serve

Architecture

This project follows the "thin proxy" design principle with minimal data transformation and simple code:

• src/config.ts - Configuration for bots and API connections
• src/botManager.ts - Manages multiple Discord bot instances
• src/services/pollinationsApi.ts - Thin proxy to Pollinations.AI API
• src/index.ts - Main entry point

Usage

1. Invite your bots to a Discord server
2. Mention a bot to get a response
3. Use the !ping command to test if the bots are online

Configuration

Environment Variables

Create a .env file in the root directory with the following variables:

# Discord Bot Tokens (required)
BOT_TOKEN_1=your-discord-token-1
BOT_TOKEN_2=your-discord-token-2

# Discord Bot Client IDs (for generating invite links)
BOT_CLIENT_ID_1=your-discord-client-id-1
BOT_CLIENT_ID_2=your-discord-client-id-2

# Bot Names and Personalities (optional)
BOT_NAME_1=DeepSeek
BOT_PERSONALITY_1=a helpful AI assistant with a friendly personality
BOT_MODEL_1=deepseek

# Conversation Channels (optional, comma-separated channel IDs)
BOT_CONVERSATION_CHANNELS_1=channel-id-1,channel-id-2

Conversation Channels

You can configure specific channels where bots will respond to all messages (not just when mentioned):

1. Finding Channel IDs:

   • Enable Developer Mode in Discord (User Settings > Advanced > Developer Mode)
   • Right-click on a channel and select "Copy ID"

2. Configuration Options:

   • Global Setting (applies to all bots):

     CONVERSATION_CHANNELS=1234567890123456789,9876543210987654321
     

   • Bot-specific Settings (overrides global setting for individual bots):

     BOT_CONVERSATION_CHANNELS_1=1234567890123456789
     BOT_CONVERSATION_CHANNELS_2=9876543210987654321
     

3. Behavior:

   • In conversation channels, bots can see and respond to the full conversation history
   • Each bot has access to the last 10 messages in the channel for context
   • Bots will respond to all messages in conversation channels, including from other bots
   • Each bot will ignore its own messages to prevent infinite loops
   • All bots can respond to the same message, creating a group conversation
   • The bot will still respond to mentions in all channels
   • In conversation channels, bots will wait a random period (5-30 seconds) before responding to simulate more natural conversation flow
   • Direct mentions receive immediate responses without the random delay
   • If an error occurs, the bot will post detailed error information for debugging

4. Bot Conversations:

   • To create bot conversations, configure multiple bots to use the same conversation channel
   • The simplest way is to use the global CONVERSATION_CHANNELS setting
   • Bots will see and respond to each other's messages, creating dynamic multi-bot conversations
   • Each bot maintains its own personality and character throughout the conversation
   • The random response delay creates more natural-feeling conversations between bots
   • Bots receive special system instructions to be aware they're in a group conversation

5. Dynamic Avatars:

   • Each bot automatically gets a unique avatar generated using the Pollinations.AI image API
   • Avatars are created based on the bot's model name
   • The system uses a prompt like "minimalist elegant black and white avatar for model [modelName]"
   • Avatars are set when the bot starts up
   • This gives each bot a distinct visual identity that matches its model name

Debugging

This application uses the debug library for logging. To enable logs, set the DEBUG environment variable before running the application. You can specify namespaces to control the verbosity.

• Show all logs:

  DEBUG=app:* npm start

• Show only bot manager logs:

  DEBUG=app:bot npm start

• Show bot manager and API logs:

  DEBUG=app:bot,app:api npm start

Available namespaces:

• app:index: Application lifecycle (startup, shutdown)
• app:config: Configuration loading
• app:bot: Bot manager events (login, messages, errors)
• app:api: Pollinations API interactions

License

ISC

Acknowledgments

• Discord.js
• Pollinations.AI