Is there anybody else maintaining a “why” folder next to AI-generated code?

That began for me after the third time I opened a file I created two weeks prior and had absolutely no idea why some condition had been there. Everything worked fine but I could not determine whether this particular thing was critical or left over from previous iterations.

Right now I create a simple .txt file next to each project called decisions.md and each time I implement some piece of AI code that is not completely self-explanatory, I leave just one line of explanation. No frills whatsoever: "Implemented the retry loop because the API times out in case of high load and not because of user error."

This has made me a bit slower. However, this way I do not have to spend an entire day explaining my previous choices to myself before sending the project to the contractor.

Interesting to see how other people organize themselves or whether this approach is overkill for smaller projects.