Thank you for creating such a wonderful project! I planned to create a simular application, that includes the generation of tests.
Observations
- After interacting with ChatGPT for a while, I noticed it would often require some encouragements of me to create a complete function/response. (E.g. it would only return half of the expected function, and after asking it 5 times, "please finish it", it would (eventually)).
- I noticed its solutions were often not compliant with (strict) pre-commit requirements (e.g.). However, after it was given the feedback of the
pre-commit
violations, e.g. line-length violations by python black, or missing type-hints, it would often correct itself.
Question
Hence, I would like to ask:
would you be willing to enhance this repository to facilitate automatic feedback into chatgpt based on pre-commit output?
(Example output:)
For completeness, a detailed suggestion on how such a feedback could be realised can be found in here, and perhaps you have different ideas :)
Ps. If you have any questions on pre-commit or otherwise, feel free to reach out!