📄 中文摘要
在软件开发中，传统的测试驱动开发（TDD）强调代码的局限性，而现在大语言模型（LLMs）能够在前提错误的情况下仍然生成连贯的代码。这种现象使得开发者在面对看似正确的代码时，容易产生误解。LLMs通常不会直接反驳用户的请求，反而会填补空白，这在项目管理中可能导致预算超支和决策失误。因此，提出了一种新的开发流程，即以提示驱动的开发（PDD），旨在帮助开发者更好地理解和验证生成的代码，从而避免仅依赖表面上的解决方案。
以提示驱动的开发（PDD）：反对舒适猜测的宣言
出处: Prompt Driven Development (PDD) A Manifesto Against Comfortable Guessing
发布: 2026年2月26日
📄 English Summary
Prompt Driven Development (PDD) A Manifesto Against Comfortable Guessing
The traditional practice of Test-Driven Development (TDD) emphasizes the limitations of code, while the advent of Large Language Models (LLMs) allows for coherent code generation even when the initial premise is flawed. This phenomenon can mislead developers into trusting seemingly correct code. LLMs are typically polite and do not contradict user prompts, instead filling in gaps, which can lead to budget overruns and poor decision-making in project management. To address this issue, a new development process called Prompt Driven Development (PDD) is proposed, aimed at helping developers better understand and validate the generated code, thus avoiding reliance on superficial solutions.