In short, MD files. After stuff works, I canonize the documentation about how it works by having Cursor write all of that in a special folder. Then, I can @refe

In short, MD files. After stuff works, I canonize the documentation about how it works by having Cursor write all of that in a special folder. Then, I can @reference that folder or a doc in it at the start of a prompt that will need that context. It's part of a larger process for working with LLMs that I call "Plans on Plans." I wrote about it on Medium.[0] [0] https://levelup.gitconnected.com/you-are-bugs-improving-your-code-with-ai-agents-d6052de6633d https://levelup.gitconnected.com/you-are-bugs-improving-your...