Copilot, for your terminal

A CLI tool that generates shell scripts from a human readable description using Ollama.

Installation

Nix

Add plz-cli to your flake inputs:

{
  description = "Your nixos configuration";

  inputs = {
    nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable";
    
    plz-cli = {
      url = "github:rukh-debug/plz-cli";
    };
  };
}

Then create a wrapper script in your home-manager configuration to set up the required environment variables:

{ inputs, pkgs, ... }:
let
  plz-wrapper = pkgs.writeShellScriptBin "plz" ''
    export OLLAMA_MODEL="qwen2.5-coder:7b"
    export OLLAMA_URL="http://localhost:11434"
    exec ${inputs.plz-cli.packages."${pkgs.system}".default}/bin/plz "$@"
  '';
in
{
  home.packages = with pkgs; [
    plz-wrapper
  ];
}

This wrapper automatically sets the required environment variables, so you don't need to configure them manually.

You may need to close and reopen your terminal after installation. Alternatively, you can download the binary corresponding to your OS from the latest release.

Usage

plz uses Ollama to generate shell commands. To use it, you'll need to have Ollama running and specify which model to use.

Required Environment Variables

OLLAMA_MODEL (required): The Ollama model to use for generating commands

export OLLAMA_MODEL='qwen2.5-coder:7b'

OLLAMA_URL (optional): The Ollama server URL

export OLLAMA_URL='http://localhost:11434'

If not set, defaults to https://ollama.rukh.me. For local Ollama installations, use http://localhost:11434.

Setup

1. First, make sure you have Ollama installed and running. Visit ollama.com for installation instructions.

2. Pull the qwen2.5-coder:7b model:

   ollama pull qwen2.5-coder:7b

3. List available models on your system:

   ollama list

4. Set your environment variables (skip this step if using the Nix wrapper above):

   export OLLAMA_MODEL='qwen2.5-coder:7b'
   export OLLAMA_URL='http://localhost:11434'

   You can add these to your bash/zsh profile for persistence between sessions.

Once you have configured your environment, run plz followed by whatever it is that you want to do.

Examples

plz "list all files in current directory"
plz "find all rust files and count lines of code"
plz "compress this folder into a tar.gz archive"
plz "show me disk usage for each directory"

To get a full overview of all available options, run plz --help

$ plz --help
Generates bash scripts from the command line

Usage: plz [OPTIONS] <PROMPT>

Arguments:
  <PROMPT>  Description of the command to execute

Options:
  -y, --force    Run the generated program without asking for confirmation
  -h, --help     Print help information
  -V, --version  Print version information

Error Handling

If you encounter errors, here are some common issues and solutions:

• "This program requires an Ollama model to be specified": You need to set the OLLAMA_MODEL environment variable
• "model not found, try pulling it first": The specified model isn't available on your Ollama server. Run ollama pull qwen2.5-coder:7b to download it
• Connection errors: Check that your OLLAMA_URL is correct and the Ollama server is running

Develop

Make sure you have the latest version of rust installed (use rustup). Then, you can build the project by running cargo build, and run it with cargo run.

License

This project is open-sourced under the MIT license. See the License file for more information.