GPT Python Script Documentation

This script interacts with OpenAI's GPT-3 or GPT-4 models to generate text completions based on given prompts. It provides options to use pre-defined prompts, files, or command outputs as input. The generated completion and input can be saved for future reference.

Table of Contents

1. Installation and Configuration
2. Usage
3. Command Line Arguments
4. Adding Custom Prompts
5. Adding Commands

Installation and Configuration

Before running the script, ensure that you have Python and the openai Python package installed.

The script requires a configuration file named .openai_config located in your home directory. This file should contain your OpenAI API key, the default model, and temperature in the following format:

[DEFAULT]
API_KEY = your_openai_api_key
MODEL = model_name
TEMPERATURE = temperature_value

Usage

To execute the script, navigate to the directory containing the script and run:

python gpt.py

Command Line Arguments

The script supports several command line arguments:

• --file: Provide one or multiple filenames to use as input.
• --command-output: Choose a command to run and use its output.
• --prompt: Select a prompt by index.
• --model: Model to use for the chat completion.
• --temperature: Temperature to use for the chat completion.
• --confirm-send: Automatically confirm sending the message without prompt.
• --model-selection: Choose the GPT model to use without prompt.
• --list-saved: List all saved input-output files and exit.
• --output: Provide a custom path for saving the output.

Adding Custom Prompts

You can add custom prompts by editing the prompts.example file in your home directory. The prompts should be in a Python dictionary format, with the keys being the prompt names and the values being the prompt texts.

Adding Commands

You can add commands to be used with the --command-output argument by editing the commands dictionary in the script. The keys should be the command names and the values should be the command to run in a list format.

Functions

The script contains several functions that handle different aspects of the CLI:

1. read_prompts(): Reads prompts from a file named prompts.example located in the user's home directory and returns a dictionary of prompts.
2. get_input_or_file_input(): Parses command-line arguments and returns the input content and argument values.
3. list_saved_files(): Lists saved input-output files and returns the name of the selected file.
4. select_prompt(): Allows the user to select a prompt from a list of predefined prompts and returns the selected prompt.
5. edit_message_in_vim(message): Opens the provided message in Vim for editing and returns the edited message.
6. save_input_output(input_content, output_content): Saves input and output content to timestamped files in a .gpt directory located in the user's home directory.
7. main(): Main function that orchestrates the script's workflow.

Workflow

1. The script reads the prompts from the prompts.example file.
2. It parses the command-line arguments and gets the input content.
3. The user selects a prompt.
4. The message is displayed, and the user can choose to send the message, edit it, or cancel the operation.
5. The user selects the GPT model to use (GPT-3 or GPT-4).
6. The script sends the message to the OpenAI API and receives the chat completion output.
7. The input and output content are saved to timestamped files in the .gpt directory.
8. The user can choose to exit or continue with another input.

Prompts

./gpt
Select a prompt:
0. commit_message_prompt
1. one_shot
2. document
3. debug
4. cover_letter
5. job_classifier
6. job_decider