A simple script to chat with OpenAI's ChatGPT

Index

• Introduction
• Getting Started
• Screenshots & Demo
• Requirements
• Installation
• Usage
• Project Structure
• Documentation
• OpenAI's Data Policy
• Credit

Introduction

Inspired by the infamous HAL9000 from the cult classic "2001: A Space Odyssey", this simple CLI enables you to interact with OpenAI's ChatGPT and generate images.

Getting Started

You would need an API key from OpenAI to use this CLI. Open this link and generate an API key for yourself. Upon installation, this CLI stores your personal API key in a file called .chat-api-key in your Home directory.

This CLI is using the gpt-3.5-turbo model by default. You can read more about it here.

To interact with text-davinci-003 model, simply start your query with -davinci followed by a space and the rest of your query after that.

gpt-4 Model is also available. Start a prompt with -gpt4 followed by the rest of the query.

Screenshots

Requirements

• curl
• jq
• wget

Optional

• glow helps with better output formatting.

Installation

curl -sS https://github.com/Brutuski/hal2023-cli/blob/main/build.sh | sudo -E bash

Manual Installation

• Download the latest release from here.
• Unzip it and copy hal2023 to /usr/local/bin.
• You can copy it into any directory of your choice, but remember to add the path to your $PATH.
• Store your OpenAI API key in your home directory as such ~/.chat-gpt-api. Paste only your API key in said file with no spaces or newlines.

Usage

• Run the command hal2023

• To exit simply type exit or quit or :q

• Multiline prompts are supported as. Press enter once to continue your prompt in a newline. Press enter twice to send the prompt.

• Default model is gpt-3.5-turbo. If you would like to interact with the text-davinci-003 model simply start your prompt with -davinci

• To interact with gpt-4 model, start the prompt with -gpt4 followed by the query.

• Image generation has been added. Simple start your input with -image followed by the image prompt. Generated image URL is shortened using the tny.im API and displayed.

• Type -email or --e and press enter for quick email generation process.

• Type -article or --a and press enter for quick article generation process.

• Type -cmd followed by the query to generate single line shell commands ready to be executed in he terminal.

• More options:

  make check
  make about
  make help

• A history feature has been added as of v1.4.1. The history file is created and maintained at ~/.hal2023_history.txt.

Project Structure

├── hal2023                     # Main script
├── README.md                   # Documentation file
├── CONTRIBUTING.md             # Contributing document
├── CODE_OF_CONDUCT.md          # Code of Conduct document
├── build.sh                    # To install, check requirements and take API key input
├── image-assets                # Directory for all documentation related image assets
├── .gitignore                  # File to primarily exclude .idea and .DS_Store
├── .github                  
  ├── CHANGELOG.md              # Changelog
  ├── ISSUES_TEMPLATE.md        # Issues reporting template
  └── PULL_REQUEST_TEMPLATE.md  # Pull request submission template
└── LICENSE                     # MIT License file

Documentation

Issues Template. Please report any issues or feature requests here.

Changes can be tracked in the Changelog

To contribute, please check out Contributing and Pull Request Template

Code of Conduct

OpenAI's Data Policy

Please refer to OpenAI's data usage policy to stay informed about how your queries and prompts are being handled by them.

Credit

The installation script was inspired by bash-script-template. It replaced the previously used method of Makefile and a separate check_requirements.sh script.