gpt-cli

Simple CLI for Chat GPT

Usage

Chatting in a text file

Pass in a file with the text you want to generate a response to.

./gpt file.txt

The contents of the file will be the complete prompt for the GPT model.

# file.txt

Hello, how are you?

The Chat GPT response will be appended to the file:

# file.txt

Hello, how are you?

**GPT Response (model: gpt-4-turbo-preview)**
Hello! I'm here and ready to help you. How can I assist you today?

**User Response**

You can continue the conversation by adding your response under **User Response** and running the command again with the same file. The GPT response will be appended to the file.

Scripting with GPT

You can also use the CLI to script with GPT. Pass in the prompt as a string and the response will be printed to the console.

echo "Hello, how are you?" | ./gpt
# Hello! I'm here and ready to help you. How can I assist you today?

Useful for more elaborate scripting use cases or for integrating with other tools.

Using different models

By default, the CLI uses the gpt-4-turbo-preview model. You can specify a different model by passing in the --model flag.

./gpt --model gpt-3.5-turbo file.txt

You can also list the available models by running the CLI with the --list-models flag.

./gpt --list
# babbage-002
# dall-e-2
# dall-e-3
# davinci-002
# gpt-3.5-turbo
# gpt-3.5-turbo-0125
# ...

Installation

• Clone the repository
• Make sure jq is installed in your environment.
• Have an active OpenAI API key set to the OPENAI_API_KEY environment variable. Thats it!

TIP: Add an alias to your .bashrc or .zshrc for easy access.

alias gpt="/path/to/gpt-cli/gpt"

Now you can run the CLI from anywhere in your terminal.