ChatGPT-bot

Interact with ChatGPT

Go CLI to fuels a Telegram bot that lets you interact with ChatGPT, a large language model trained by OpenAI.

Installation

Download the file corresponding to your OS in the releases page.

• chatgpt-telegram-Darwin-amd64: macOS (Intel)
• chatgpt-telegram-Darwin-arm64: macOS (M1)
• chatgpt-telegram-Linux-amd64: Linux
• chatgpt-telegram-Linux-arm64: Linux (ARM)
• chatgpt-telegram-Win-amd64: Windows

After you download the file, extract it into a folder and open the env.example file with a text editor and fill in your credentials.

• TELEGRAM_TOKEN: Your Telegram Bot token
  • Follow this guide to create a bot and get the token.
• TELEGRAM_ID (Optional): Your Telegram User ID
  • If you set this, only you will be able to interact with the bot.
  • To get your ID, message @userinfobot on Telegram.
  • Multiple IDs can be provided, separated by commas.
• EDIT_WAIT_SECONDS (Optional): Amount of seconds to wait between edits
  • This is set to 1 by default, but you can increase if you start getting a lot of Too Many Requests errors.
• MANUAL_AUTH (Optional): Setting to true will disable the browser authentication
  • Requires setting OpenAI Session Token manually, use /setToken <token> bot command
  • See auth section for info on how to get your session token
• Save the file, and rename it to .env.

Note Make sure you rename the file to exactly .env! The program won't work otherwise.

Finally, open the terminal in your computer (if you're on windows, look for PowerShell), navigate to the path you extracted the above file (you can use cd dirname to navigate to a directory, ask ChatGPT if you need more assistance 😉) and run ./chatgpt-telegram.

Running with Docker

If you're trying to run this on a server with an existing Docker setup, you might want to use our Docker image instead.

docker pull ghcr.io/m1guelpf/chatgpt-telegram

Here's how you'd set things up with docker-compose:

services:
  chatgpt-telegram:
    image: ghcr.io/m1guelpf/chatgpt-telegram
    container_name: chatgpt-telegram
    volumes:
      # your ".config" local folder must include a "chatgpt.json" file
      - .config/:/root/.config
    environment:
      - TELEGRAM_ID=
      - TELEGRAM_TOKEN=

Note The docker setup is optimized for the Browserless authentication mechanism, described below. Make sure you update the .config/chatgpt.json file in this repo with your session token before running.

Authentication

By default, the program will launch a browser for you to sign into your account, and close it once you're signed in. If this setup doesn't work for you (there are issues with the browser starting, you want to run this in a computer with no screen, etc.), you can manually extract your session from your browser instead.

To do this, first sign in to ChatGPT on your browser, then open the Developer Tools (right click anywhere in the page, then click "Inspect"), click on the Application tab and then on the Cookies section, and copy the value of the __Secure-next-auth.session-token cookie.

You will then have to create a config file in the following location depending on your OS (replace YOUR_USERNAME_HERE with your username:

• ~/.config/chatgpt.json: Linux
• C:\Users\YOUR_USERNAME_HERE\AppData\Roaming\chatgpt.json: Windows
• /Users/YOUR_USERNAME_HERE/Library/Application Support/chatgpt.json: macOS

Note If you have already run the program, the file should exist but be empty. If it doesn't exist yet, you can either run the program or manually create it.

Finally, add your cookie to the file and save it. It should look like this: { "openaisession": "YOUR_COOKIE_HERE" }.

License

This repository is licensed under the MIT License.