Matrix GPT is a Matrix chatbot that uses OpenAI for real-time chatting.

Run the Docker container:

docker run -d --name matrix-gpt \
  -e MATRIX_PASSWORD="matrix password" \
  -e MATRIX_ID="matrix id" \
  -e MATRIX_URL="matrix server url" \
  -e OPENAI_TOKEN="openai token" \
  -e SQLITE_PATH="persistent path for sqlite database"
  -e USER_IDS="allowed user ids"
  ghcr.io/mazzz1y/matrix-gpt:latest

You can configure GPT Matrix using the following environment variables:

• SERVER_URL: The URL to the Matrix homeserver.
• USER_ID: Your Matrix user ID for the bot.
• PASSWORD: The password for your Matrix bot's account.
• SQLITE_PATH: Path to SQLite database for end-to-end encryption.
• HISTORY_EXPIRE: Duration after which chat history expires.
• GPT_MODEL: The OpenAI GPT model being used.
• GPT_HISTORY_LIMIT: Limit for number of chat messages retained in history.
• GPT_TIMEOUT: Duration for OpenAI API timeout.
• GPT_MAX_ATTEMPTS: Maximum number of attempts for GPT API retries.
• GPT_USER_IDS: List of authorized user IDs for the bot.

Alternatively, you can set these options using command-line flags. Run ./matrix-gpt --help for more information.

This bot supports the following commands:

• !image[-natural/-vivid]: This command will create and return an image based on the text you provide. The default style is "Natural".
• !reset [text]: This command will reset the user's history. If you provide text after the !reset command, the bot generates a response using GPT, based on this input text.
• [text]: If you simply input text without any specific command, the bot will automatically generate a GPT-based response related to the text provided.

• You can use short aliases for a command; for example, !i for !image, or !iv for !image-vivid.
• If you need to stop any ongoing processing, you can just delete your message from the chat`.
• In case of errors, the bot reacts with a ❌. If you notice this, please check logs.