Simple GPT CLI

April 13, 2023

Using my friend ChatGPT I made a little CLI to talk to… GPT. It’s basically a bash script that you can place in one of you bin folders and execute from wherever in your terminal:

gpt "I need a new cactus, what is a nice type that needs little maintenance?"

After which GPT will answer you with (hopefully) a nice answer:

GPT-3 answers:

One type of cactus that is low-maintenance is the Peruvian apple cactus. This cactus is slow-growing and does not need to be watered often. It is also tolerant to a range of temperatures, making it a

Script

#!/bin/bash

function print_help() {
    GREEN="\033[1;32m"
    BLUE="\033[1;34m"
    RESET="\033[0m"

    echo -e "${GREEN}Usage:${RESET}"
    echo -e "  ${BLUE}gpt [options] [question]${RESET}"
    echo ""
    echo -e "${GREEN}Options:${RESET}"
    echo -e "  ${BLUE}-h, --help${RESET}        Show this help message and exit"
    echo -e "  ${BLUE}--t=NUM${RESET}           Set the maximum number of tokens in the response (default: 50)"
    echo ""
    echo -e "${GREEN}Examples:${RESET}"
    echo -e "  ${BLUE}gpt --t=200 Tell me a story about a programmer who became a potato${RESET}"
    echo -e "  ${BLUE}gpt --t=100 \"How can I make my computer run faster with a rubber chicken?\"${RESET}"
    echo -e "  ${BLUE}gpt What would a conversation between a cat and a toaster sound like --t=20${RESET}"
}

if [[ -z "$OPENAI_API_KEY" ]]; then
    echo "Please set the OPENAI_API_KEY environment variable."
    exit 1
fi

# Set default max_tokens
max_tokens=50

# Parse command-line arguments
for i in "$@"
do
case $i in
    -h|--help)
    print_help
    exit 0
    ;;
    --t=*)
    max_tokens="${i#*=}"
    shift
    ;;
    *)
    if [[ -z "$QUESTION" ]]; then
        QUESTION="$i"
    else
        QUESTION="$QUESTION $i"
    fi
    shift
    ;;
esac
done

# Prompt for question and max_tokens if not provided
if [[ -z "$QUESTION" ]]; then
    echo "Please enter your question:"
    read -r QUESTION
fi

if [[ -z "$max_tokens" ]]; then
    echo "Please enter the max tokens (default: 50):"
    read -r max_tokens_input
    if [[ ! -z "$max_tokens_input" ]]; then
        max_tokens="$max_tokens_input"
    fi
fi

response=$(curl -s -X POST "https://api.openai.com/v1/engines/text-davinci-002/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d "{\"prompt\": \"${QUESTION}\", \"max_tokens\": ${max_tokens}, \"n\": 1, \"stop\": null, \"temperature\": 0.7}")

error_message=$(echo "$response" | jq -r '.error.message' 2>/dev/null)

if [[ "$error_message" != "null" ]] && [[ ! -z "$error_message" ]]; then
    echo "Error: $error_message"
else
    answer=$(echo "$response" | jq -r '.choices[0].text' | sed 's/^[[:space:]]*//')
    echo "GPT-3 answers: $answer"
fi

How to use

First, we need to make sure jq is installed. jq is a light-weight command-line JSON processor used by this script to parse the response from the OpenAI API endpoint.

# For Ubuntu/Debian
sudo apt-get install jq

# For macOS (using Homebrew)
brew install jq

Then, let’s create the folder in which we will put the script later on.

Open a terminal window.
Create a bin directory in the user’s home directory, if it doesn’t already exist:

mkdir -p ~/bin

Change to the newly created bin directory:

cd ~/bin

We will now create the script.

Create a new file called gpt using a text editor (e.g., nano, vim, or code). Copy and paste the provided script into the file:

vi gpt

To able to execute the script later on, we need to give it executive permissions.

Save the file and exit the text editor.
Set the script to be executable:

chmod +x gpt

We will now add the bin directory to the PATH environment, allowing us to call the script from any location.

Add the ~/bin directory to the PATH environment variable. For bash, add the following line to the end of the ~/.bashrc or ~/.bash_profile file:

export PATH="$HOME/bin:$PATH"

For zsh, add the following line to the end of the ~/.zshrc file:

export PATH="$HOME/bin:$PATH"

Source the appropriate shell configuration file to apply the changes:

For bash:

source ~/.bashrc
# Or, if using macOS:
source ~/.bash_profile

For zsh:

source ~/.zshrc

Set the OpenAI API key as an environment variable by adding the following line to the same shell configuration file used in step 7:

export OPENAI_API_KEY="your-api-key"

Replace your-api-key with the actual API key.

Source the shell configuration file again (as shown in step 9) to apply the new environment variable.

Now the gpt script should be ready to use from anywhere in the terminal. Type gpt followed by a question or use the -h flag to display the help message.

Script

How to use

See also