请问HN:你是如何高效地使用Claude代码的?

2作者: nocobot7 个月前原帖
我对基于大型语言模型的编码工具持怀疑态度,但许多人似乎对其带来的巨大生产力提升赞不绝口,我对此也不介意。不过,当我尝试使用 cc 时,我感到非常失望。为了提供一些背景,我正在进行一个相对全新的 Rust 项目,并给它布置了一些我认为适合初级同事的任务,比如: - 更改一个 trait 及其所有实现的返回类型 - 将重复的代码重构为一个辅助函数 - 用一个外部 crate 替换我们的一些代码 它没有正确完成任何一项任务,而且花费了很长时间。我是不是用错了这个工具? 你们是如何使用 cc 或其他自主工具的?
查看原文
I tend to be sceptical when it comes to LLM based coding tools but many people seem to be raving about huge productivity gains which I wouldn’t mind as well.<p>However when trying cc it left me vey disappointed. For context I’m working on a relatively greenfield rust project and gave it tasks that I would consider appropriate for a junior level colleague like:<p>- change the return type of a trait and all it’s impls<p>- refactor duplicate code into a helper function<p>- replace some of our code with an external crate<p>it didn’t get any of them correct and took a very long time. Am I using the tool wrong?<p>How are you using cc or other agentic tools?