While looking through the inspect-ai repo, I saw that they have a CONTRIBUTING.md file which might also make sense for us. From how I understand it, the name is automatically recognized and can be selected easily when looking at the repo, just like README.md
To me, things we might want to put in there are:
- add relevant unit tests whenever fitting (e.g. fixing a typo in a comment without it is fine)
- using AI is fine, but issues/PR messages should remain human-written, or have a dedicated block like "AI summary" at the end. The point is that nobody should be forced to read through walls of AI slop.
While looking through the inspect-ai repo, I saw that they have a
CONTRIBUTING.mdfile which might also make sense for us. From how I understand it, the name is automatically recognized and can be selected easily when looking at the repo, just likeREADME.mdTo me, things we might want to put in there are: