请问HN：你是如何在编码之外使用编码代理/命令行接口的？
每个人都在创作。打开任何社交平台，比如Twitter、LinkedIn或HN，你会发现每天都有新项目。人工智能让从一个想法变成一个可工作的产品变得如此简单。
如果你不是开发者，或者刚刚起步，或者只了解软件的某个领域，人工智能对你来说是个了不起的工具。你可以构建以前无法实现的东西。如果你已经有经验，现在的速度会快得多。这两种情况都很棒。
我认为我们也可以用同样的热情去学习新事物。不仅仅是用人工智能来写代码，而是用它来理解以前没有时间去理解的内容。多读书，深入新领域，建立实际的知识。
例如，如果你是一名软件工程师，始终想学习硬件，现在是最好的时机。人工智能可以带你了解电路，解释数据表，帮助你调试固件。这些以前需要多年经验才能入门的东西，现在变得更加容易。
大型语言模型可能是有史以来最好的学习工具，但我们大多数人主要用它来跳过学习过程。我相信编程助手非常适合这个工作！
我最近开始这样做。我使用Claude Code的频率减少了，更多的是用它来“帮助我理解这个”，而不是“为我构建这个”。这让我感到非常满意。
我很好奇这里的其他人是如何在编码之外使用助手的。你们是用它们来学习吗？阅读吗？探索新领域吗？什么对你们有效呢？
查看原文
Everyone is shipping. Open any feed, Twitter, LinkedIn, HN, and it's new projects every single day. AI made it so easy to go from an idea to a working product.<p>If you're not a developer, or you're just starting out, or you only knew one area of software, AI is amazing for you. You can build things you could never build before. If you're experienced, you're just way faster now. Both are great.<p>I think we can use that same energy on learning new things too. Not just using AI to write code, but using it to understand things we never had time to understand before. Reading more. Going deep on new domains. Building actual knowledge.<p>For example, if you're a software person who always wanted to learn hardware, this is the best time to do that. AI can walk you through
circuits, explain datasheets, help you debug firmware. Stuff that used to take years of context to even get started with.<p>LLMs might be the best learning tool ever created, but we're mostly using it to skip the learning. I do believe coding agents are perfect for this job!<p>I started doing this lately. I use Claude Code less for "build this for me" (I still do build a lot) and more for "help me understand this." It's been really satisfying.<p>Curious how others here use agents outside of coding. Are you using them to learn? To read? To explore new fields? What's working for you?