📄 中文摘要
大型语言模型（LLMs）显著加快了开发速度，但如果开发者对系统理解不深，可能会加大所有权差距。在缺乏自律的情况下，智能体往往会重复逻辑、膨胀控制文件，并产生微妙的技术债务。虽然像持久“技能”这样的工具可以提供帮助，但保持清晰的架构仍然是开发者的责任。回想起没有LLMs的日子，开发者手动编写代码，虽然耗时，却能深入了解整个系统，快速定位故障。AI的出现使得代码生成的速度大幅提升，但也带来了新的挑战。
技能问题
出处: A skill issue
发布: 2026年3月4日
📄 English Summary
A skill issue
Large Language Models (LLMs) can significantly accelerate development, but they also widen the ownership gap if developers do not have a deep understanding of their systems. Without discipline, agents tend to duplicate logic, bloat control files, and create subtle technical debt. While tools like persistent 'skills' can help, maintaining clean architecture remains the developer's responsibility. Reflecting on the days without LLMs, developers wrote code by hand, which took time but allowed them to know the entire system intimately and troubleshoot effectively. The advent of AI has drastically increased the speed of code generation, but it also introduces new challenges.