Discord AI Bot with Gemini

This is a Python-based Discord bot that uses Google's Gemini API to provide AI-powered responses. It can be configured to interact in specific channels continuously or respond to keywords in other channels.

Features

• Configurable Command Prefix: Change the bot's command prefix (e.g., $ to !!).
• Gemini API Integration: Leverages Google's Gemini models for generating responses.
• Contextual Conversations:
  • Set Channels: Designate specific channels where the bot will respond to every message, maintaining a conversation history for that channel.
  • Keyword Replies: Responds to messages containing specific keywords (e.g., "ai", "bot", its own name) in any channel (unless ignored), using user-specific conversation history.
• Customizable AI Contexts: Define specific personas or topic expertise for channels by placing .txt files in a context/ directory.
• Persistent Data: Saves channel settings, conversation histories, and user contexts to a JSON file (bot_data.json).
• Channel Management:
  • setchannel/unsetchannel: Manage channels for continuous conversation.
  • ignore/unignore: Control where keyword-based replies are active.
  • setcontext/unsetcontext: Assign or remove a custom AI context for a channel.
• Rate Limiting: Prevents spamming AI responses (both keyword and set channel triggers) and warns users when they exceed limits. Configurable via .env file.
• Customizable Persona: Define the bot's core behavior and topic expertise via a system prompt in the .env file, or through custom context files.
• Utility Commands: Includes help and time (system uptime).

Prerequisites

• Python 3.8 or higher
• pip (Python package installer)
• A Discord Bot Token
• A Google Gemini API Key

Setup Instructions

1. Clone or Download the Repository:

   # If using git
   git clone <repository_url>
   cd <repository_directory>

   If you downloaded a ZIP, extract it and navigate into the directory.

2. Create a Virtual Environment (Recommended):

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

3. Install Dependencies:

   pip install  discord.py python-dotenv google-generativeai

4. Configure Environment Variables: Create a .env file in the root directory of the project and populate it with your credentials and desired settings. Use the provided .env file in the chat as a template:

   DISCORD_TOKEN=YOUR_DISCORD_BOT_TOKEN_HERE
   GEMINI_API_KEY=YOUR_GEMINI_API_KEY_HERE
   COMMAND_PREFIX=$
   BOT_KEYWORDS=ai,bot,assistant,your_bot_name # Comma-separated list of keywords
   GEMINI_MODEL_NAME=gemini-1.5-flash          # e.g., gemini-pro, gemini-1.0-pro, gemini-1.5-flash
   SYSTEM_PROMPT="You are a helpful AI assistant. Your primary topic is general knowledge but you adapt based on context."
   CONTEXT_DIR=context                         # Directory containing .txt files for custom AI contexts/personas
   
   # Rate Limiting for AI Responses
   RATE_LIMIT_MAX_PROMPTS=3 # Max number of AI prompts allowed within the time window per user
   RATE_LIMIT_SECONDS=8     # Time window in seconds for the rate limit

   • DISCORD_TOKEN: Your Discord bot's token. (Obtain from the Discord Developer Portal)
   • GEMINI_API_KEY: Your API key for Google Gemini. (Obtain from Google AI Studio)
   • COMMAND_PREFIX: The prefix for bot commands (e.g., $, !, !!).
   • BOT_KEYWORDS: Comma-separated list of words that will trigger the bot in non-set channels. The bot's actual name will also be added to this list automatically when it starts.
   • GEMINI_MODEL_NAME: The specific Gemini model you want to use (e.g., gemini-pro, gemini-1.5-flash).
   • SYSTEM_PROMPT: The initial instruction given to the AI to define its persona and general behavior. This is used if no specific context is set for a channel.
   • CONTEXT_DIR: The directory where custom AI context .txt files are stored. Defaults to context.
   • RATE_LIMIT_MAX_PROMPTS: The maximum number of AI prompts a user is allowed to make within the RATE_LIMIT_SECONDS window. Defaults to 3 if not set.
   • RATE_LIMIT_SECONDS: The time window in seconds for the rate limit. Defaults to 8 if not set.

5. Create the context/ directory and add context files (Optional but Recommended): Create a folder named context (or whatever you set CONTEXT_DIR to) in the root directory of your project. Inside this folder, you can create .txt files, where each file contains a system prompt for a specific AI persona or topic. The filename (without the .txt extension) will be the context_name you use with the setcontext command.

   Example: context/general.txt

   You are a helpful AI assistant. Your primary topic is general knowledge but you adapt based on context.
   

   context/tech_support.txt

   You are a friendly and knowledgeable tech support assistant. Focus on providing clear, step-by-step solutions to common technical issues. Ask clarifying questions if needed.
   

6. Run the Bot:

   python bot.py

   You should see log messages in your console indicating the bot has connected to Discord and the Gemini API is configured.

Bot Usage

How the Bot Interacts

• Set Channels: If you use the {COMMAND_PREFIX}setchannel command in a channel, the bot will respond to every message sent in that channel. It maintains a shared conversation history for this channel.
• Keyword Replies: In any other channel (that is not explicitly ignored using {COMMAND_PREFIX}ignore), the bot will listen for messages containing any of the defined BOT_KEYWORDS (or its own name). If a keyword is detected, it will generate a response using a conversation history specific to the user who triggered it.
• Custom AI Contexts: You can set a specific AI context for a channel using {COMMAND_PREFIX}setcontext <context_name>. This context will override the default SYSTEM_PROMPT from the .env file for all AI interactions in that channel (both set-channel and keyword-based replies).
• Rate Limit Notifications: If a user triggers AI responses too frequently (either via keywords or in a set channel), they will be temporarily rate-limited and notified by the bot.
• Commands: Messages starting with the COMMAND_PREFIX are treated as commands.

Available Commands

(Replace {COMMAND_PREFIX} with the prefix you set in your .env file, e.g., $)

• {COMMAND_PREFIX}help:

  • Shows a help message listing all available commands and their usage.

• {COMMAND_PREFIX}setchannel:

  • Sets the current channel for continuous conversation. The bot will respond to all messages in this channel.
  • Usage: {COMMAND_PREFIX}setchannel

• {COMMAND_PREFIX}unsetchannel:

  • Unsets the current channel from continuous conversation. The bot will no longer respond to every message here (but keyword replies might still work).
  • Usage: {COMMAND_PREFIX}unsetchannel

• {COMMAND_PREFIX}ignore:

  • Disables keyword-triggered replies in the current channel. The bot will not respond to keywords in this channel.
  • Usage: {COMMAND_PREFIX}ignore

• {COMMAND_PREFIX}unignore:

  • Enables keyword-triggered replies in the current channel (if they were previously disabled).
  • Usage: {COMMAND_PREFIX}unignore

• {COMMAND_PREFIX}setcontext <context_name>:

  • Sets a specific AI context (persona/topic) for the current channel. The context_name must match the filename (without .txt) of a file in your context/ directory. This context will apply to all AI interactions in this channel.
  • Usage: {COMMAND_PREFIX}setcontext tech_support

• {COMMAND_PREFIX}unsetcontext:

  • Removes any custom AI context set for the current channel, reverting AI responses in this channel to use the SYSTEM_PROMPT defined in your .env file.
  • Usage: {COMMAND_PREFIX}unsetcontext

• {COMMAND_PREFIX}time:

  • Shows the system uptime of the server where the bot is hosted.
  • Usage: {COMMAND_PREFIX}time

Data Storage

• The bot stores its configuration (set channels, ignored channels, active contexts) and conversation histories in a file named bot_data.json.
• This file is created automatically if it doesn't exist.
• User-specific conversation history for keyword replies is stored under user_specific_context.
• Channel-specific conversation history for "set channels" is stored under main_chat_history.

Troubleshooting

• "DISCORD_TOKEN not found" or "Invalid Discord token": Ensure your .env file is correctly named, in the root project directory, and contains the correct DISCORD_TOKEN.
• "GEMINI_API_KEY not found": Ensure your .env file has the GEMINI_API_KEY set correctly.
• "Failed to configure Gemini API or model": Double-check your GEMINI_API_KEY and GEMINI_MODEL_NAME in the .env file. Ensure the model name is valid for the Gemini API.
• Bot not responding:
  • Check the console logs for any errors.
  • Ensure the bot has the necessary permissions in your Discord server (Read Messages, Send Messages, View Channels).
  • Verify the COMMAND_PREFIX and BOT_KEYWORDS in your .env file.
• Permissions Error for uptime command: The system running the bot might not have the uptime command available or accessible.
• Context not loading/found: Ensure your CONTEXT_DIR in .env is correct and that .txt files are placed directly inside it. Check bot logs for "Error loading context file" messages.

Contributing

Feel free to fork the repository, make improvements, and submit pull requests.