Created by pasting the entire Swift GitHub repo into Gemini 2.0 and asking it to port it to a web page: https://gist.github.com/simonw/b4aec4e879e50ac74f6f9cc6e1cdc...
With such variance though, it now becomes much easier for me to see why the question of if LLMs are any good at coding is so contentious every time it comes up on HN. If, even for such a small, well defined task, there's such variance in behavior from seemingly small prompt changes, it's now easier for me to see why some people see it as the second coming and others think LLM-assisted program is all hot air.
I agree, I have noticed some prompts which work perfectly fine on Claude when used in WindSurf IDE which uses Claude the same prompt did not work.
LLM models work fine for small scripts but when it comes to large Codebase I just cannot trust them.