README style GPT conversations for any IDE
This script is a command-line utility that works as an interface for the OpenAI GPT-3 and GPT-4 models. The script listens for changes made in markdown files within the ~/gptme/gpt/ directory, and accordingly generates textual content using the GPT-3 or GPT-4 model.
Create a .md readme file in your favorite IDE inside gpt/ dir and start typing and press save to talk to ChatGPT, see sample.md.
| Prefix | Function |
|---|---|
4 |
Use GPT-4 for generating content. |
[space] or [new line] |
Only send the current prompt in the thread to reduce the number of tokens used (cost). |
mm |
Generate image from the prompt. |
= |
Request GPT to fix any grammar in the user prompt using excellent English, but retaining any foul language. |
. |
Ask GPT to provide short and concise answers. |
/ |
Instruct GPT to walk through its thinking process step by step. |
- Sending file content to GPT-3 or GPT-4 based on the first character(s) in the file.
- Creating an image from the prompt if the first two characters are 'mm'.
- Tailoring responses from GPT based on certain prefixes.
- Install the utility and set it up in your IDE.
- Save a markdown file with your content in the
~/gptme/gpt/directory. - Customize the conversation with GPT-3 or GPT-4 by starting your conversation with special prefixes, where the first characters of the conversation have a special meaning.
For API limitations or cost considerations, you may modify the contents of the files to manage the number of tokens used by the models.
The script also allows you to cancel a running task and revert all the changes made by the last command.
Dependencies:
- Python3
- openai
- threading and queue for Python
You need to have the OpenAI API key to use this utility. You can either put your OpenAI API key in an environment variable named OPENAI_API_KEY or directly assign it to the openai_api_key variable in the script.
Note: The implementation shown above uses an environment variable to assign the API key.
A detailed usage guide can be obtained by running .<path_to_run_file>/run --usage in the terminal.