Twitch ChatGPT Bot

This repository provides a Twitch chatbot that leverages the OpenAI API.

Current version: gpt-3.5-turbo

Prerequisites:

• OpenAI API Token: Sign up on their platform here
• Twitch Token: Login with your Twitch account and generate a token here

Execution Requirements:

• Python 3.10 (includes switch-case instruction)

Environment Variables:

You need to create an .env file at the same level as the .main file that contains these variables:

Variable	Value	Description
GPT_ENDPOINT	https://api.openai.com/v1/chat/completions	OpenAI API endpoint
GPT_TOKEN	Bearer xx	Token for your registered OpenAI account
TWITCH_IRC_SERVER	irc.chat.twitch.tv	Twitch IRC URL
TWITCH_IRC_PORT	6667	Twitch IRC port
TWITCH_IRC_NICKNAME	xx	Nickname of the account used as a bot
TWITCH_TOKEN	oauth:xx	Token for the account used as a bot
CHANNEL_MONITORING	#xx	Name of the channel to monitor, preceded by '#'

Environment variables are visible by root users, consider avoid this usage on production or encode them by using a library like cryptography.

Execution:

cd src/  
python3 main.py

How to prompt on Twitch chat:

To be able to prompt on Twitch chat, the program needs to be running and the connection with IRC must be established.

The format to make prompts is the following:

!chat inputPrompt

If there are no errors, the bot will answer the prompt quoting the transmitter.

If you want to change the number of characters the GPT will answer, you can do it in the gpt.py file in the send_request function, where a limit of 150 characters is specified.

def send_request(self, prompt, limit = 150):

Future Optimizations:

• Implementation of prompt queue to be resolved by the GPT.
• Multithreading to monitor multiple Twitch channels simultaneously.