A simple, and efficient, CLI program to find a Linux command with chat GPT.

GPT-CLI

Setup

You need to have the var OPENAI_KEY exported from your environment. I recommend exporting it on .zshrc for zsh shells (or .bashrc for bash). Or you can also create a local .env at the root of the project.

Create a .env file containing your key at the root of the project

echo "export OPENAI_KEY=<your key>" > .env

Usage

gpt-cli [OPTIONS] <PROMPT>

Arguments:
  <PROMPT>  Your prompt

Options:
  -t, --tokens           The max number of tokens generated per message
  -m, --model            The model to be used [default: text-davinci-003]
  -h, --help             Print help
  -V, --version          Print version

Example

   gpt-cli "syncronize two folders" -t 100

Installation

The binary name is called gpt. In the future, the binaries will be added to the Arch Linux package manager, pacman. For now, you can download the compiled binaries on the releases page, or build it from source.

Building from source

gpt-cli is written in Rust, so you'll need to have a Rust installation in order to compile it.

To build gpt-cli:

$ git clone [email protected]:gabrieldemian/gpt-cli.git
$ cd gpt-cli
$ cargo build --release
$ ./target/release/gpt --help

Copy the binary to a PATH folder:

$ sudo cp ./target/release/gpt /usr/bin/gpt