AI Junction

---

AI Junction exposes a Discord interface for generating text and images using LMStudio (or any OpenAI API interface) and ComfyUI back-ends. The bot allows users to manage their own configurations for both LLM prompting and image generation, as well as generate content in-server using custom prompts.

Support Discord: https://discord.gg/7B2ZQhAB2m

Features

• LLM-based text generation using LMStudio
• Image generation using ComfyUI
• Custom configuration management for both GPT and image generation settings
• User-friendly Discord interface for managing settings and generating content

Prerequisites

Before installing the AI Junction bot, ensure you have the following:

• Node.js (v16.20.1 or higher)
• npm
• A Discord account and a registered Discord application/bot
• Access to LMStudio server and/or
• Access to ComfyUI server
• MongoDB database (get a free hosted instance here https://www.mongodb.com/cloud/atlas/register)

---

Installation

1. Clone the repository:

git clone https://github.com/p4stoboy/AI-Junction.git
cd AI-Junction

2. Install dependencies:

npm install

3. Rename .env.example to .env and fill in the required environment variables:

DISCORD_TOKEN=your_discord_bot_token
CLIENT_ID=your_discord_app_client_id
MONGODB_URI=your_mongodb_connection_string
LMSTUDIO_URL=http://localhost:1234/v1
COMFY_URL=http://127.0.0.1:8188
COMFY_WS=ws://127.0.0.1:8188
ENABLE_GPT=true
ENABLE_COMFY=true
GLOBAL_ADMIN_ID=your_discord_user_id

4. Remove .example and edit model mappings in model_mappings.example.json to match the models available in your ComfyUI server:

• display_name: The name that will be displayed in the bot's settings menus
• model: The exact name of the model as it appears in the ComfyUI server
• default_model: The model that will be selected by default when creating a new image generation configuration

all fields are required

{
    "maps": [
        {
            "model": "anythingMidjourneyV4_v1.ckpt",
            "display_name": "Anything Midjourney V4"
        },
        {
            "model": "sd_xl_turbo_1.0_fp16.safetensors",
            "display_name": "SD XL Turbo"
        },
        {
            "model": "unfiltredai.safetensors",
            "display_name": "Unfiltered AI"
        }
    ],
    "default_model": "sd_xl_turbo_1.0_fp16.safetensors"
}

4. Build the project:

npm run build

5. Deploy commands to Discord:

npm run deploy-commands

note: Ensure that the ENABLE_COMFY / ENABLE_GPT variables are set to false if you are not supporting the corresponding back-end, this will disable the corresponding commands.

6. Start the bot:

npm start

---

Usage

GPT Commands

• /gptsettings: Manage GPT configuration settings

• This command opens an interactive menu to create, edit, or select GPT configurations.

• /gpt <prompt>: Generate text using GPT

• Example: /gpt Write a short story about a robot learning to paint

Image Generation Commands

• /imaginesettings: Manage image generation configuration settings

  • This command opens an interactive menu to create, edit, or select image generation configurations.
  • When creating or editing a configuration, you'll set a positive prompt template. This template can include a {prompt} wildcard, which will be replaced with the user's input when using the /imagine command.
  • Example template: "A detailed painting of {prompt} in the style of Van Gogh"

• /imagine <prompt> [options]: Generate an image

  • Basic usage: /imagine A serene lake with mountains
  • The <prompt> will replace the {prompt} wildcard in your active configuration's positive prompt template.
  • Optional override parameters:
    • positive_prompt: Override the entire positive prompt template
    • negative_prompt: Override the negative prompt
    • steps: Override the number of generation steps (1-50)
    • cfg: Override the CFG scale (0-20)
    • model: Override the selected model

  Examples with override parameters:

  • /imagine A serene lake with mountains steps:30 cfg:7.5
  • /imagine A futuristic cityscape model:"SD XL Turbo" negative_prompt:"clouds, birds"
  • /imagine An alien landscape positive_prompt:"Otherworldly terrain with {prompt}, vibrant colors" cfg:8

Configuration Quick-Edit Commands

• /cfg <value>: Update the CFG value in your active Imagine config

• Example: /cfg 1.7

• Valid range: 0-20

• /steps <value>: Update the steps value in your active Imagine config

• Example: /steps 20

• Valid range: 1-50

• /model <name>: Update the model in your active Imagine config

• Example: /model "SD XL Turbo"

• Use the exact name of the model as it appears in the settings menu

Configuration Options

GPT Settings

• Config Name: A unique name for your GPT configuration
• System Prompt: The initial prompt that sets the context for the AI

Image Settings

• Config Name: A unique name for your image configuration
• Positive Prompt: The main prompt for image generation
• Negative Prompt: Prompt for elements to avoid in the image
• Steps: Number of generation steps (1-50)
• CFG: Classifier Free Guidance scale (0-20)
• Model: Select from available image generation models

Troubleshooting

• If commands are not working, ensure you've run the deploy-commands script and that your bot has the necessary permissions in your Discord server.
• Check the console logs for any error messages if the bot is not responding as expected.
• Ensure that both LMStudio and ComfyUI servers are running and accessible at the URLs specified in your .env file.
• If you're experiencing issues with a specific configuration, try creating a new one to see if the problem persists.

Contributing

Contributions are welcome. Please feel free to submit a Pull Request.