I've made a GitHub bot for automatic issue triaging that uses OpenAI

URL (because apparently I can’t post links): github[dot]com/pierotofy/issuewhiz/

In short, you can define yes/no questions using natural language about newly opened issues and then define rules using boolean logic to act on the results of such questions.

For example:

* A: Does this text look like a software bug report?
* B: Is this about a frontend problem?

Actions:

* if A and B --> Add label bug frontend, then stop
* if A --> Add label bug

Thought I’d share and invite a discussion on possible feedback, ideas for improvements, etc.

One annoyance I’ve found is that the results from the API are different than the ones I get from the web browser chat interface, which is something that has been observed in this forum before.

1 Like