chatgpt-cli

User Guide - ChatGPT Command-Line Interface (CLI)

The ChatGPT CLI allows you to interact with ChatGPT directly from your command line, offering an efficient platform for real-time communication. This user guide will provide you with simple, straightforward instructions on how to install, set up, and effectively use the CLI.

Examples

Simple Interactive Chat

Ask a Question, non-Interactively

{ echo "What is the largest file in this directory?"; ls -l } | chatgpt-cli chat

Improving README, non-Interactively

echo "Rewrite this README file as a user guide. Make it easy to read and informative. Use a helpful and clear style" \
  | chatgpt-cli chat < README.md > README-new.md
mv README-new.md README.md

Editing a directory full of notes, non-Interactively

for file in notes/*.md; do
  printf "\nEditing '%s'\n============================\n" "$file"
  {
    printf "Revise my notes. "
    printf "Use Markdown format. "
    printf "Revise the text of the note below to use a clear and informative style. "
    printf "Use newlines to keep line length less than 120 characters.\n\n"
    cat "${file}"
  } | chatgpt-cli chat > "${file}.new"

  if [ $? -ne 0 ]; then
    printf "  Request Failed: '%s'\n" "${file}"
    rm "${file}.new"
  else
    mv "${file}.new" "${file}"
  fi
done

Simple Image Generation

Which gives us the resulting image.

Generate an Image, non-Interactively

echo "Monkey in a banana costume" | chatgpt-cli image -o monkey

In non-Interactive mode only the name of the output files are sent to stdout. In this case, monkey-01.png, shown here.

Installation

To install ChatGPT CLI using Homebrew:

brew tap duanemay/tap
brew install chatgpt-cli

Generating Cover Images for a directory full of notes, non-Interactively

for file in notes/*.md; do
  dir=$(dirname "${file}")
  filename=$(basename "${file}")
  base_filename=${filename%.*}

  printf "\nCreating Images '%s'\n============================\n" "$file"
  {
    printf "Describe the contents of an image that would make a good cover image for the blog post below.\n\n"
    cat "${file}"
  } | chatgpt-cli chat --system-message "As an expert creator of blog posts" > "${dir}/${base_filename}-img-description.txt"

  if [ $? -ne 0 ]; then
    printf "  Request Failed: '%s'\n" "${file}"
    rm "${dir}/${base_filename}-img-description.txt"
    continue
  fi

  chatgpt_file=$( ls chatgpt-cli* | tail -1 )
  grep -A 1 user "${chatgpt_file}" | grep '"content":' | cut -d':' -f 2 | sed 's/"//g'
  cat "${dir}/${base_filename}-img-description.txt" | chatgpt-cli image -o "${dir}/${base_filename}-img"
done

Initial Set Up

To use the ChatGPT CLI, you'll need a ChatGPT API key. You can generate this key by signing up on the OpenAI platform.

It is recommended to create a .chatgpt-cli file in your home directory or the directory where you'll run the CLI. Inside this file, include your API key alongside additional default settings you wish to use.

Example:

API_KEY=sk-mysupersecretAPIkey

Configuration Settings

The priority order for the configuration settings is as follows:

Command line flags > Environment Variables > Configuration files > Defaults

Each flag can be set in a configuration file, by changing it to uppercase and replacing - with _. Environment variables are the same as the configuration file flag, with the addition of the prefix CHATGPT_.

As an example, the --api-key sk command line flag can be set in a configuration file as API_KEY=sk, and as an environment variable as CHATGPT_API_KEY=sk.

The full list of available flags and corresponding config file variables:

Common Flags:

Chat Flags:

Image Flags:

For instance, if you want to change the end of the message and session markers, modify them in your configuration file.

You can select a different configuration file using --config flag. Each config file should specify settings as KEY=VALUE pairs, with each pair on a separate line. Lines commencing with # are considered comments and ignored.

Usage

The basic command to use chatgpt-cli is as follows:

chatgpt-cli [command]

The available commands are as follows:

1. chat: Start a chat session with ChatGPT.
2. image: Generate an image using DALL-E
3. completion: Generate the autocomplete script for your chosen shell.
4. help: Seek help regarding any command.
5. list-models: Retrieve a list of all models available to your account.
6. replay-session: Replay a chat session from a previously saved file. 7version: Get version information.

Chatting

Initiate a chat session with ChatGPT using the chat command:

chatgpt-cli chat

You'll be prompted to input your message, which can span multiple lines. Send your message with TAB or CTRL+C.

Continue your conversation with ChatGPT by inputting a new message once you receive a response.

Exiting the chat is made possible by inputting CTRL+C or TAB with no message.

All chat sessions are saved in a session file, for which the --session-file flag can specify the file of your choice:

chatgpt-cli chat --session-file session.json

The --model, --role, --temperature, --max-tokens, and --top-p flags allow for individual configuration of the Model, Role, Temperature, Max Tokens, and Top P respectively.

A system prompt can be set by using the --system-message flag:

chatgpt-cli chat --system-message "You are a captivating storyteller who brings history to life by narrating the events, people, and cultures of the past."

Replaying a Session

Replaying a chat session lets you revisit a previous chat in a more readable format than the raw JSON. Use the replay-session command:

chatgpt-cli replay-session --session-file session.json

Generating Images

Generate an image with DALL-E using the image command:

chatgpt-cli image

You'll be prompted to input your description of an image, which can span multiple lines. Send your description with TAB or CTRL+C.

Exiting the chat is made possible by inputting CTRL+C or TAB with no description.

All images are saved with a prefix in the form dall-e-DATE-TIME-nn.png where DATE-TIME is the timestamp when the session started, and nn for the image number from the session. You can override the --output-prefix or -o flags.

You can control how many variants of the requested images to generate with the --number or -n flag. The Number of Images must be between 1 and 10, inclusive.

You can control the size of the requested images with the --size or -s flag. The Size must be one of 256x256, 512x512, or 1024x1024.

Listing Models

list-models retrieves a list of all available models related to your account:

chatgpt-cli list-models

Checking the Version

Fetch information about the CLI's version using version:

chatgpt-cli version

Build and Release

For a dry-run of the deployment process, you can run the following:

goreleaser --snapshot --skip-publish --clean

For a full deployment, you'll need to add a GitHub token to the ./.github_token file, then run the following:

goreleaser --clean
brew upgrade