ai_plz

A simple command-line helper powered by LLM.

Introduction

ai_plz is a minimalist command-line helper that sends your need to LLM, asks for a shell command, and executes it if permitted. This helper is designed to be a one-time command generator, not a full AI agent. The command-line entrypoint of this package is "plz" (short for "please").

Usage

# Install
$ uv tool install ai_plz  # or: pipx install ai_plz

# Get help
$ plz --help
Usage: plz [OPTIONS] [USER_INPUT]...

  A simple command-line helper powered by LLM.

Options:
  -e, --env FILE            Path to optional env file
  -y, --run / -n, --no-run  Enable/Disable automatic execution
  -p, --prompt [en|zh]      Select prompt
  -z, --zh                  Shortcut for --prompt=zh
  --show-vars               Show environment vars
  --show-prompt             Show prompt
  --debug                   Enable debug logging
  -h, --help                Show this message and exit

# Example (User inputs will be concatenated by spaces)
$ plz Clear conda cache
Clear the conda cache to free up disk space and remove outdated packages.
>>> conda clean --all
Run the above command? [y/n]:

Configuration

Model settings should be provided through environment variables, which can be listed by plz --show-vars. Usually, you need to specify your own LLM provider through AI_PLZ_API_KEY, AI_PLZ_BASE_URL, and AI_PLZ_MODEL.

Caveats

When permitted, the command provided by LLM will be executed in a shell subprocess. It is recommended to check the command before execution, so a confirmation is required by default, as shown in the example usage above. You can pass the -y option to skip confirmation and take your own risk. You can also pass -n to skip execution without ask.