It does. But it requires much longer to write the first prompt, and much much longer to write the follow up prompts to fix the code from the previous prompt.
There are process issues, though. Sometimes, you can put the error message into the prompt and tell the LLM to avoid it, for example. Given that the LLM starts over from scratch every time you change the prompt, however, it's probably easier to fix the code than fix the prompts—at least that's usually what I do.
It does. But it requires much longer to write the first prompt, and much much longer to write the follow up prompts to fix the code from the previous prompt.
There are process issues, though. Sometimes, you can put the error message into the prompt and tell the LLM to avoid it, for example. Given that the LLM starts over from scratch every time you change the prompt, however, it's probably easier to fix the code than fix the prompts—at least that's usually what I do.