Skip to content

Latest commit

 

History

History
109 lines (81 loc) · 4.34 KB

README.md

File metadata and controls

109 lines (81 loc) · 4.34 KB

Gemini Discord Bot

Gemini Discord Bot allows you to converse on Discord using Google's Gemini API. It supports text-based conversations and responds to attached files.

Features

  • Text-based Conversation: Utilizes the Gemini API for natural language conversations.
  • Image Recognition: Recognizes and responds to the content of uploaded images.
  • File Recognition: Analyzes and responds to the content of uploaded files.
  • Conversation History: Maintains user-specific conversation history for context-aware responses.
  • History Reset: Resets the conversation history when the user sends RESET.
  • Discord Mentions and Special Characters Handling: Processes mentions and special characters appropriately and sends clean text to the Gemini API.
  • Long Message Splitting: Splits messages exceeding Discord's 2000-character limit into smaller chunks for seamless transmission.
  • Google Search Tool: Generates responses based on Google Search results.
  • Image Generation: Generates images based on text prompts (currently in beta).

Setup

  1. Create a Discord Bot:

  2. Obtain a Google AI Studio API Key:

    • Get an API Key from Google AI Studio.
    • Alternatively, if using Vertex AI API, configure your Google Cloud Platform project credentials.

  3. Set Environment Variables:

    • Create a .env file and set the following variables:

      DISCORD_BOT_TOKEN=Your Discord Bot Token
GOOGLE_AI_KEY=Your Google AI Studio API Key # Not needed for Vertex AI
# GCP_PROJECT_ID=Your Google Cloud Platform Project ID # Required for Vertex AI
# GCP_REGION=Your Google Cloud Platform Region # Required for Vertex AI
IMG_COMMANDS_ENABLED=True/False # Enable/Disable image generation commands (default: False)

  4. Install Dependencies:

    pip install -r requirements.txt

  5. Tools Configuration:

    To configure the generate_content_config in your code, follow these guidelines:

    • The tools parameter is included by default in the configuration:

      generate_content_config = types.GenerateContentConfig(
    temperature=1,
    top_p=0.95,
    max_output_tokens=8192,
    tools=tools,  # Comment out this line if you are using gemini-2.0-flash-thinking-exp
    # safety_settings=[...],  # Example safety settings (optional)
)

    • Important Note: When using gemini-2.0-flash-thinking-exp, ensure that the tools parameter is commented out. This configuration ensures compatibility with the specific version.

  6. Run the Bot:

    python GeminiDiscordBot.py

Usage

  • Text Conversation: Mention the bot or send a direct message (DM) to start a conversation.

  • File Recognition: Upload a file, with or without accompanying text, and the bot will analyze and respond to its content.

  • Reset Conversation History: Send RESET to clear the conversation history.

  • Image Generation: Use the !img command to generate images with the following format:

    !img <prompt>|<negative prompt>|<aspect ratio>
    • Prompt: Describe the image to generate.
    • Negative Prompt (optional): Specify what to exclude.
    • Aspect Ratio (optional): Specify aspect ratio (e.g., 1:1).

    Example:

    !img a cute cat|blurry|1:1

Important Notes

  • The Google AI Studio API has usage limits and may incur charges beyond the free tier. Refer to the Google AI Studio documentation for pricing details.
  • Use this bot in compliance with the Gemini API Terms of Service.
  • Unexpected errors or inappropriate responses may occur. Please use it with discretion.
  • If you encounter errors installing libmagic:
    • MacOS: brew install libmagic
    • Linux (Debian/Ubuntu): sudo apt-get install libmagic1
    • Windows: Download libmagic from the official website and add it to your environment variables.

Disclaimer

  • The developer is not responsible for any damages caused by the use of this bot.
  • This bot utilizes the Gemini API, and the developer is not responsible for the content of its responses.

License

This project is licensed under the MIT License.