As spotted by @chrispix initially in issue #1, chatty is not surfacing any errors that occur when making requests to the OpenAI API.

The SDK chatty is currently does not surface any errors, it just prints them to the console.

I want to migrate to a new solution that provides chat completions or roll my own if needed.

Add for example MIT License, BSD License, Apache License or any other open source license.

Having a LICENSE file can help establish trust and openness with regard to the project and its authors, which can be important when collaborating with other developers or when publicly distributing the project.

For more information, visit the GitHub Docs:

• Licensing a repository
• Adding a license to a repository

I've authenticated with my API key, but no matter what I prompt, I get no response. I assume ChatGPT is returning some kind of error, but nothing is reported in the terminal. Would it be possible to relay errors from ChatGPT?