After eating this, I have to give a PR for adding llms.txt to tailwindcss. tailwindcss's response makes one sigh deeply, again and again. I have no answers either.

I don't know if you all feel this way, but today's AI writes code quite well, especially for small projects or when just starting out; the file organization is still acceptable. However, when things start to grow larger, it can still write code decently, but the stability and reliability of its file organization rapidly collapse. I have no doubt that AI can write good code, but so far, it still doesn't do very well in organizing project structures.

More and more people do not look at the code, but they will establish very strict tests and various boundary conditions to ensure that this black box of code written by AI works properly—at least within the specified range. But I am different; I still look at the code, even if it seems a bit less efficient because of it. The functionality of the module is a product of my code; whatever code I have, I will have that functionality. I belong to the school of thought that emphasizes the primacy of code.