I'm using GPT4 to power language agnostic mutation testing

Hey so I’m using ChatGPT4 to power our open-source mutation testing tool here- github. com/codeintegrity-ai/mutahunter

To get good mutations I’m pageranking the AST nodes of all the covered sources files similar to aider chat and using that as context for GPT4 to allow it to generate good code mutations.

I’m not too experienced with using Gpt4 for Codegen like this and wanted to know if I was missing some best practices or if there is some additional context I can pass to Gpt4 to further increase the quality of code mutations it creates.

Appreciate the help, thanks!

1 Like

The author is using ChatGPT4 (presumably a large language model) to generate mutations for their open-source mutation testing tool (Mutahunter) available on [CODEINTEGRITY AI mutahunter ON GitHub github.com].
They’re leveraging PageRank on AST nodes (representing code structure) to provide context to GPT4, aiming to improve the quality of the generated code mutations.
The author seeks advice on best practices or additional context to enhance the quality of these mutations.