使用代码生成代理创建的软件难以维护？
2 分•作者: nlpnerd•4 个月前
我们已经制定了完善的产品需求文档（PRD）、系统设计和实施计划，并将其提供给代理。
我是一名机器学习工程师（MLE）和软件工程师（SWE），所以我并非无法阅读代码，也了解正确的实践方法。但不同之处在于：
由于我们不像早期编码代理那样每天花费 6-8 小时处理代码库，我们缺乏深入的知识和直觉，这在出现 bug 时有助于我们缩小根本原因的范围，或者知道如何最好地扩展现有代码库以支持新功能。
由于编码代理带来的生产力提升，代码库的规模和应用程序的复杂性相对于开发人员的数量也有所增加。每位开发人员现在都需要在更大的范围和更多的微服务中保持上下文和理解。随着时间的推移，这使得审查应用程序和代码库的扩展在认知上变得越来越困难。
还有其他人也遇到这种情况吗？这仅仅是我们工程流程的问题吗？你们有什么好的方法？
查看原文
We do have in place proper PRDs and system designs and implementation plans that we feed to the agent.<p>I'm an MLE and SWE, so it's not like I can't read code or don't know any proper practices. What is different though are that<p>Since we are not spending 6-8 hours a day working with the code base unlike pre-coding agents, we lack the indepth knowledge and intuition that helps us narrow down root cause when bugs arise, or know how to best extend the current code base to support a new feature<p>Because of the productivity from coding agents, the size of the code base and sophistication of the app relative to number of developers has also increased. Each developer now needs to maintain context and understanding over a greater surface area and number of micro services. Over time this makes reviewing extension to the app and code base increasingly more difficult cognitively.<p>Anyone else dealing with this too? Is it just a problem with our engineering processes? What works for you?