New project, README says npm install and you are good. Half a day later, still debugging the environment. Then I started using Claude Code. Same task, 10 minutes. This article covers exactly how to do it, and when to let AI take over completely.
接手新專案，README 說「npm install 就能跑」，結果花了半天還在 debug 環境。後來我開始用 Claude Code，同樣的事 10 分鐘搞定。這篇講具體怎麼做，以及什麼時候該讓 AI 完全接手。
AI wrote 200 tests for me. All passed. We deployed. Then a bug hit production. The problem wasn’t that AI wrote bad tests—it tested the wrong things. This article shares a simple framework: What stays with humans, How goes to AI.
AI 幫我寫了 200 個測試，全部通過，上線還是出 bug。問題不是 AI 寫錯，是它測的東西根本不是該測的。這篇文章分享一個簡單的框架：What 留給人，How 交給 AI。
Learning to prompt won’t help you master AI. The real bottleneck isn’t AI’s capability—it’s your judgment. This article breaks down the four levels of judgment, five sources it comes from, and concrete methods to develop it.
「學會 Prompt 就能駕馭 AI」這句話誤導了很多人。真正的瓶頸不是 AI 的能力，是你的判斷力。這篇文章拆解判斷力的四個層次、五個來源，以及具體的培養方法。
GitHub Copilot, Cursor, and Claude Code each have different positioning. But even with the right tool, your productivity might still drop—the key isn’t the tool, it’s whether you have the judgment to wield it. This article analyzes real scenarios for AI-assisted development, common problems, and what kind of people can truly use these tools well.
GitHub Copilot、Cursor、Claude Code 各有定位。但工具選對了，效率還是可能變差——關鍵不在工具，在於你有沒有「判斷力」來駕馭它。這篇文章分析 AI 輔助開發的真實場景、常見問題，以及什麼樣的人能真正用好這些工具。
AI excels at executing tests, but struggles with judging ‘what to test’ and ‘whether results are correct.’ QA’s core value is shifting from execution to acceptance—validating AI’s validation. This article analyzes the four most important QA skills in the AI era and the short, medium, and long-term impacts on teams.
AI 擅長執行測試，但不擅長判斷「該測什麼」和「結果對不對」。QA 的核心價值正在從執行轉向驗收——驗證 AI 的驗證。本文分析 AI 時代 QA 最重要的四種能力排序，以及對團隊的短中長期影響。