Comment by Supermancho
2 days ago
Claude has been miserable for anything Java related, in my experience. Use it maybe as a cross check for ChatGPT, which is far superior. They both make mistakes, but Claude repeatedly changes course, contradicts itself, or makes suggestions that are basically unusable. I don't understand why conversations here are always focusing on a bad LLM.
You likely need to provide it with more direction in your prompts.
I use the same prompts I used for others. Then I rephrased and expanded. This did not improve the quality of responses.
Have you tried prompting the loop? Something like "keep trying until <insert format command here> is happy?"
have you used claude code?
yes. specifically.
I couldn’t quite tell from your comment. Anyway it’s not my experience with CC but everyone seems to have a slightly different take on it.