gpt-shell-4o-mini 1.2.0

A simple, lightweight CLI to use OpenAI's ChatGPT and DALL-E from the terminal

original work

https://github.com/0xacx/chatGPT-shell-cli

I translated it to python, and added some features, Few things was not working for me, also updated to gpt 4o-mini

** greate work from the original authour **

GPT-shell-4o-mini

A simple, lightweight CLI to use OpenAI's ChatGPT and DALL-E from the terminal.
This project now uses a Python installer (install.py) for setup and a Python uninstaller (remove.py) for clean removal.
The main shell script is chatgpt.sh, which is installed to your system path for easy access.

Features ✨

🌍 Cross-Platform Support

• Windows 10/11 - Full support with automatic environment variable setup
• macOS - Works seamlessly on all versions
• Linux - Supports all major distributions (Ubuntu, Fedora, Debian, Arch, etc.)

🧠 Advanced Terminal Context

• Static User Profile - Remembers your OS, username, and distribution
• Dynamic Session Capture - Sends recent terminal commands AND outputs to ChatGPT
  • tmux support - Captures last 30 lines automatically
  • screen support - GNU screen buffer capture
  • History fallback - Works even without tmux/screen
• Intelligent Responses - ChatGPT sees your errors, current directory, and shell type

🚀 Seamless Setup

• First-run wizard - Automatic API key setup and verification
• No external dependencies - Uses Python's requests library (no curl needed)
• Platform-aware - Automatically configures for your OS

---

Getting Started

Prerequisites

• Python 3.7+ (usually pre-installed)
• An OpenAI API key - Get one free at OpenAI

Optional (for best terminal context):

• tmux - Highly recommended for full terminal session capture

  sudo apt install tmux   # Ubuntu/Debian
  sudo dnf install tmux   # Fedora
  brew install tmux       # macOS
  

---

Installation

Windows

Pip Installation (Only method for Windows):

pip install gpt-shell-4o-mini

After installation, run gpt to start the setup wizard:

gpt

The wizard will:

• Prompt for your OpenAI API key
• Verify the key with OpenAI
• Save it to Windows environment variables automatically
• Create your user profile

Note: Manual installation via install.py is not supported on Windows.

macOS / Linux

Option 1: Pip Installation (Recommended)

pip install gpt-shell-4o-mini

Run gpt to start the setup wizard.

Option 2: Manual Installation

Quick install without cloning:

curl -O https://raw.githubusercontent.com/wkdkavishka/GPT-shell-4o-mini/origin/install.py
sudo python3 install.py

Or clone and install:

git clone https://github.com/wkdkavishka/GPT-shell-4o-mini.git
cd GPT-shell-4o-mini
sudo python3 install.py

You will be prompted for your OpenAI API key during the installation. Alternatively, you can provide your key directly:

sudo python3 install.py --key <YOUR_OPENAI_API_KEY>

This will:

• Download and install chatgpt.sh to /usr/local/bin/gpt
• Optionally create a symlink /usr/local/bin/chatgpt
• Add your API key and /usr/local/bin to your shell profile

Uninstallation

The remove.py script provides comprehensive uninstallation that works whether you installed via pip or manually.

For a quick uninstall without cloning:

curl -O https://raw.githubusercontent.com/wkdkavishka/GPT-shell-4o-mini/origin/remove.py
python3 remove.py
# Or with sudo if you used manual installation:
sudo python3 remove.py

What it removes:

• Package installed via pip (if applicable)
• Manual installation from /usr/local/bin/gpt and /usr/local/bin/chatgpt (if applicable)
• OPENAI_KEY and /usr/local/bin PATH modifications from your shell profiles
• Chat history file (~/.chatgpt_py_history)

The script automatically detects how the package was installed and removes it accordingly.

Manual Installation

If you want to install manually:

• Download the chatgpt.sh file to a directory in your $PATH
• Add your OpenAI API key to your shell profile:
  export OPENAI_KEY=your_key_here
• Make sure /usr/local/bin is in your $PATH
• (Optional) Install imgcat for iTerm2 image support

Usage

Start

Chat Mode

• Run the script by using the gpt or chatgpt command anywhere. By default the script uses the gpt-4o-mini model.

Pipe Mode

• You can also use it in pipe mode:
  echo "What is the command to get all pdf files created yesterday?" | gpt

Script Parameters

• You can also pass the prompt as a command line argument:
  gpt -p "What is the regex to match an email address?"

---

🧠 Terminal Context Feature

How It Works

GPT-shell-4o-mini automatically sends contextual information with every prompt:

Static Profile (~/.chatgpt_py_info):

• Your username
• Operating system and version
• Linux distribution (if applicable)

Dynamic Terminal Session (captured in real-time):

• Current working directory
• Shell type (bash, zsh, PowerShell, etc.)
• Recent terminal commands AND their outputs!

Example

When you run gpt in tmux, ChatGPT sees:

[Static Profile: User: john | OS: Linux | Distro: Ubuntu 22.04]
[Terminal Session (Shell: bash | CWD: /home/john/project | Source: tmux):
john@ubuntu:~/project$ npm run dev
Error: Missing script: "dev"
]

You: How do I fix this?

ChatGPT response will be context-aware:

"I see you're missing the 'dev' script in package.json. Since you're on Ubuntu..."

Best Experience

For full terminal session capture (commands + outputs):

# Install tmux
sudo apt install tmux   # Ubuntu/Debian
sudo dnf install tmux   # Fedora  
brew install tmux       # macOS

# Start tmux
tmux

# Now run gpt - it captures your entire session!
gpt

Without tmux/screen, it falls back to command history only.

---

Support

For issues, questions, or contributions, please visit the
GitHub Repository

Commands

• image: To generate images, start a prompt with image:
• history To view your chat history, type history
• models To get a list of the models available at OpenAI API, type models
• model: To view all the information on a specific model, start a prompt with model: and the model id
• command: To get a command with the specified functionality and run it, just type command: and explain what you want to achieve

Chat context

• Enable chat context mode for the model to remember your previous chat questions and answers. Start the script with -c or --chat-context.

Set chat initial prompt

• Set your own initial chat prompt with -i or --init-prompt
  Example: gpt -i "You are Rick from Rick and Morty, reply with references to episodes."

Use the official ChatGPT model

• The default model used is gpt-4o-mini.

Use GPT-4

• If you have access to the GPT-4 model you can use it by setting the model to gpt-4, i.e. gpt --model gpt-4

Set request parameters

• To set request parameters:
  gpt --temperature 0.9 --model text-babbage:001 --max-tokens 100 --size 1024x1024
  • temperature, -t or --temperature
  • model, -m or --model
  • max number of tokens, --max-tokens
  • image size, -s or --size
  • prompt, -p or --prompt
  • prompt from a file, --prompt-from-file