Autogen Discord Bot

Autogen is an intelligent Discord bot powered by multiple LLM providers (LMStudio, OpenAI, Google Gemini), featuring memory management, web search capabilities, and natural conversation abilities. It maintains context through conversations and provides human-like responses while efficiently managing long messages and user interactions.

Features

• 🤖 Multiple LLM Provider Support:
  • LMStudio (local LLM)
  • OpenAI API
  • Google Gemini
• 🔍 Web Search Integration:
  • Search command: search for <query> or search: <query>
  • Customizable result limits (e.g., search for python: 5, search: latest AI news: 3)
  • Smart result formatting and summarization
• 💭 Long-term memory using ChromaDB with semantic search
• 🎯 Smart context retrieval for relevant responses
• 🔒 Server and channel-specific permissions
• 🚫 Automatic @ mention filtering
• ✨ Interactive response indicators
• 📝 Automatic handling of long messages
• 🧠 Improved memory recall with optimized thresholds

Prerequisites

• Python 3.10 or higher
• One of the following LLM providers:
  • LMStudio running locally (or another OpenAI-compatible API endpoint)
  • OpenAI API key
  • Google Gemini API key
• Discord Bot Token and Application
• CUDA-capable GPU (recommended for local LLM)

Installation

Option 1: Using Conda (Recommended)

1. Create a new conda environment:

conda create -n autogen python=3.12
conda activate autogen

2. Clone the repository:

git clone https://github.com/nvmax/AutogenDiscordbot.git
cd autogen-bot

3. Install dependencies:

pip install -r requirements.txt

Option 2: Using Python venv

1. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

2. Clone the repository and install dependencies:

git clone https://github.com/yourusername/autogen-bot.git
cd autogen-bot
pip install -r requirements.txt

Configuration

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

# Discord Configuration
DISCORD_TOKEN=your_discord_bot_token
ALLOWED_SERVER_ID=your_server_id
ALLOWED_CHANNEL_ID=your_channel_id

# LLM Configuration
LLM_PROVIDER=lmstudio  # Options: lmstudio, openai, gemini
LLM_BASE_URL=http://localhost:1234/v1  # For LMStudio
LLM_MODEL=your_model_name

# API Keys (if using OpenAI or Gemini)
OPENAI_API_KEY=your_openai_key
OPENAI_API_BASE=https://api.openai.com/v1  # Optional, for OpenAI-compatible APIs
GEMINI_API_KEY=your_gemini_key

# Memory Configuration
CHROMA_PERSIST_DIR=./data/chroma

# API Configuration
API_TIMEOUT=30
MAX_RETRIES=3

# Memory Settings
MAX_MEMORIES_PER_QUERY=15
CONTEXT_WINDOW_SIZE=10
MEMORY_SIMILARITY_THRESHOLD=0.15  # Threshold for memory relevance (0.0 to 1.0)
TOP_MEMORIES_TO_CONSIDER=8        # Number of top memories to consider

2. Update the configuration values:
   • Get your Discord bot token from Discord Developer Portal
   • Find your server ID by enabling Developer Mode in Discord and right-clicking your server
   • Find your channel ID by right-clicking the channel you want to use
   • If using OpenAI or Gemini, obtain API keys from their respective platforms

Using Different LLM Providers

LMStudio (Local)

1. Download and install LMStudio
2. Load your preferred model (recommended: deepseek-r1-distill-qwen-14b)
3. Start the local server with the following settings:
   • Host: assigned by LMStudio, enable network cors access
   • Port: 1234
   • Context Length: 4096 (or model maximum)
   • Temperature: 0.7
   • Max Tokens: 8196

OpenAI

1. Get an API key from OpenAI Platform
2. Set LLM_PROVIDER=openai in your .env file
3. Configure your OpenAI API key and model name

Google Gemini

1. Get an API key from Google AI Studio
2. Set LLM_PROVIDER=gemini in your .env file
3. Configure your Gemini API key

Using Web Search

The bot supports web searching with the following commands:

1. Basic search:

search for what is python programming

2. Search with limit:

search for best python frameworks: 5

or

search: latest AI news: 3

The bot will search the web and provide a formatted summary of the results.

Features in Detail

Interactive Response Indicators

• Shows thinking indicators while processing responses
• Variety of engaging status messages with emojis
• Updates messages in place for a clean chat experience

Smart Memory Management

• Improved memory recall with optimized similarity thresholds
• Retrieves up to 8 relevant memories per query
• Maintains conversation context effectively

Message Handling

• Automatically splits long responses into multiple messages
• Preserves message formatting and readability
• Handles messages up to Discord's character limit efficiently

@ Mention Handling

• Ignores messages containing @ mentions
• Allows for natural conversation flow without bot interference
• Perfect for mixed conversations with other users

Project Structure

autogen-bot/
├── bot/
│   ├── __init__.py
│   └── discord_bot.py      # Discord bot implementation with message handling
├── config/
│   ├── __init__.py
│   └── settings.py         # Configuration and environment variables
├── llm/
│   ├── __init__.py
│   └── llm_client.py       # LLM integration with improved response handling
├── memory/
│   ├── __init__.py
│   └── memory_manager.py   # Memory management with optimized recall
├── utils/
│   ├── __init__.py
│   └── embeddings.py       # Embedding utilities for semantic search
├── .env
├── main.py
├── requirements.txt
└── README.md

Usage

The bot provides a natural chat experience:

• Responds to messages in configured channels
• Shows thinking indicators while processing
• Automatically splits long responses
• Ignores @ mentions for better conversation flow
• Maintains context through ChromaDB memory system

Example interaction:

User: Tell me about quantum computing
Bot: 🤔 Let me think about that...
[Bot updates with comprehensive response]

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

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

Acknowledgments

• LMStudio for local LLM hosting
• ChromaDB for vector storage
• discord.py for Discord integration