Discord Bot with Ollama & ComfyUI Integration

A simple Python-based Discord bot that combines a local LLM (via Ollama) with ComfyUI for image generation. It also features news analysis, tarot readings, voice channel notifications, and an interactive chat command with persistent conversation history.

Table of Contents

• Features
• Architecture Overview
• Prerequisites
• Installation
• Configuration
• Usage
• Available Commands
• File Structure
• Contributing

---

Features

1. LLM Queries (.llm)

   • Query a local Ollama instance for text-based responses.

2. Image Generation (.img)

   • Generate images via ComfyUI based on custom workflow templates.

3. Chat with Memory (.chat / .clearchat)

   • Engage in a multi-turn conversation with the bot, which stores history in a local SQLite database.

4. News Analysis (.news)

   • A multi-agent system retrieves the latest news on a topic, synthesizes the key points, and summarizes them.

5. Tarot Readings (.tarot)

   • Get a one-card Tarot reading with interpretation and guidance.

6. Vision Model (.vision)

   • Analyze images using Ollama’s vision model by attaching an image and providing a prompt.

7. Voice Channel Notifications

   • Posts notifications in a specific Discord channel whenever someone joins, leaves, or switches voice channels (excluding any channels you mark as excluded in the .env).

---

Architecture Overview

Discord Bot        Ollama (Local LLM)        ComfyUI (Image Gen)
   |                       |                         |
   | Discord.py           | requests                | WebSocket + REST
   +----------------------|------------------------- |-------------------> 
   | Chat commands        |.llm, .vision            | .img generation
   | newsagent,           |                         | 
   | tarotagent           |                         |

• Discord.py: Manages bot commands, events, and server interactions.
• Ollama: Serves as the LLM backend for text queries and image analysis (vision model).
• ComfyUI: Workflow-driven image generation connected via a combination of REST/POST and WebSocket for job queueing.

---

Prerequisites

1. Python 3.8+
2. Ollama running locally (for text and vision queries).
3. ComfyUI running and reachable via HTTP/WS.
4. Discord Bot Token: Create an application and bot on Discord's Developer Portal.

---

Installation

1. Clone this repository (or copy the files into a local folder):

   git clone https://github.com/swiftraccoon/simpleDiscordBot.git
   cd simpleDiscordBot

2. Install Python dependencies:

   pip install discord.py requests websocket-client duckduckgo-search swarm python-dotenv ollama websockets

   Tip: You may want to use a virtual environment:

   python -m venv venv
   source venv/bin/activate
   pip install ...

3. Set up Ollama:
   Follow Ollama’s installation instructions to ensure it is running locally (default port 11434).

4. Set up ComfyUI:
   Ensure ComfyUI is installed and running (default port 8188).

---

Configuration

1. Copy the example env file:

   cp .env.example .env

2. Edit your .env to add your own credentials and settings:

   • DISCORD_BOT_TOKEN = Your Discord bot token.
   • DISCORD_NOTIFICATION_CHANNEL_ID = The channel ID where voice join/leave notifications should appear.
   • DISCORD_EXCLUDED_VOICE_CHANNEL_IDS = A comma-separated list of channel IDs you don’t want to track.
   • OLLAMA_API_URL = URL for Ollama’s /api/generate endpoint. (Default: http://localhost:11434/api/generate)
   • COMFYUI_SERVER_ADDRESS = Address of your ComfyUI instance, e.g. localhost:8188.
   • Additional model names (DEFAULT_LLM_MODEL, CHAT_MODEL, VISION_MODEL, etc.) can be overridden as needed.

   For example:

   DISCORD_BOT_TOKEN=your_discord_bot_token
   DISCORD_NOTIFICATION_CHANNEL_ID=123456789012345678
   DISCORD_EXCLUDED_VOICE_CHANNEL_IDS=987654321098765432,123456789012345679
   OLLAMA_API_URL=http://localhost:11434/api/generate
   COMFYUI_SERVER_ADDRESS=localhost:8188
   ...

3. (Optional) Adjust Bot Prefix:
   Change DISCORD_COMMAND_PREFIX in your .env if you want something other than ..

---

Usage

1. Run the Bot:

   python main.py

   • By default, the bot uses the --model specified in .env or you can override it:

     python main.py --model "some-other-model"

2. Disable Image Generation:
   If you do not want to generate images, add --no-img:

   python main.py --no-img

3. Invite the Bot to Your Server:

   • Go to the Discord Developer Portal → Your Application → OAuth2 → URL Generator.
   • Select bot scope and any necessary permissions (e.g., for voice states, message content, manage messages, etc.).
   • Paste the generated invite link into your browser and select the server to add it to.

4. Confirm It’s Online:

   • You should see a log message: "Logged in as: YourBotName".
   • In your server, type .help (or your custom prefix + help) to see available commands.

---

Available Commands

Below are the default commands (. is the default prefix; adjust if you changed DISCORD_COMMAND_PREFIX):

• .llm <prompt>
  Query the local LLM (Ollama). Example:

  .llm What is the capital of France?
  

• .img <prompt>
  Generate an image via ComfyUI using a preconfigured workflow. Example:

  .img a vivid painting of a dragon made of stardust
  

  This requires ComfyUI to be running.

• .vision <prompt> (attach an image)
  Send an image prompt to Ollama’s vision model. Example:

  .vision What is the primary object in this photo?
  

  Make sure to attach an image in Discord along with your text prompt.

• .news <topic>
  Get a summary of the latest news on a topic. Example:

  .news AI breakthroughs
  

• .tarot
  Draw a single Tarot card for a quick reading. Example:

  .tarot
  

• .chat <message>
  Start or continue a chat conversation with the bot. It maintains context across messages in a server-specific database. Example:

  .chat Hey, how’s it going?
  

• .clearchat
  Clear the stored chat history for the current server. This requires the Manage Messages permission. Example:

  .clearchat
  

---

File Structure

.
├── .env.example             # Environment variable template
├── chat_db.py              # Chat database interactions (SQLite)
├── comfyui_workflows.py    # Workflow templates for image generation in ComfyUI
├── main.py                 # The main entry point (Discord bot logic)
├── newsagent.py            # News agent multi-stage processing
├── tarotagent.py           # Tarot reading agent
└── README.md               # This file

• .env.example
  Contains sample environment variables. Copy and rename to .env, then fill in your own details.
• main.py
  The primary script that starts the Discord bot, loads environment variables, and registers commands.
• chat_db.py
  Handles saving and retrieving conversation history from a SQLite database.
• comfyui_workflows.py
  Provides predefined JSON-based workflows for ComfyUI image generation.
• newsagent.py
  Uses a multi-agent approach for searching and synthesizing recent news.
• tarotagent.py
  Randomly draws a Tarot card and provides a short reading using an LLM agent approach.

---

Contributing

1. Fork or Clone this repository.
2. Create a new branch for your feature or bugfix.
3. Submit a Pull Request with a clear description of your changes.

All contributions are welcome—whether it’s improving documentation, fixing bugs, or adding new features!

Enjoy building your AI-powered Discord experience!