<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: schwann2402</title>
    <description>The latest articles on DEV Community by schwann2402 (@schwann2402).</description>
    <link>https://dev.to/schwann2402</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F1654920%2F606d0dc2-3241-44e6-adf2-bb62a06868eb.png</url>
      <title>DEV Community: schwann2402</title>
      <link>https://dev.to/schwann2402</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/schwann2402"/>
    <language>en</language>
    <item>
      <title>I Made Claude Code Quiz Me Before It Lets Me Ship</title>
      <dc:creator>schwann2402</dc:creator>
      <pubDate>Mon, 10 Aug 2026 22:57:24 +0000</pubDate>
      <link>https://dev.to/schwann2402/avoid-vibe-coding-with-claude-code-407o</link>
      <guid>https://dev.to/schwann2402/avoid-vibe-coding-with-claude-code-407o</guid>
      <description>&lt;p&gt;I’ve watched engineers accept a whole PR’s worth of AI-generated code without being able to explain what half of it does. I’ve been that engineer. &lt;/p&gt;

&lt;p&gt;It’s an easy trap: the suggestion looks reasonable, the tests pass, and you tab-complete your way to a merge you couldn’t defend in a code review.&lt;/p&gt;

&lt;p&gt;So instead of trying to will myself into better habits, I built a gate. quiz-me is a Claude Code plugin that stands between you and your commit. &lt;/p&gt;

&lt;p&gt;Before Claude will help you implement a fix, it quizzes you on the code you’re about to ship — and only lets you proceed once you can show you actually understand it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;How it works&lt;/strong&gt;&lt;br&gt;
    1.  You ask Claude Code to implement a fix or feature as usual.&lt;br&gt;
    2.  Before writing the change, quiz-me interrupts with a short multiple-choice quiz about the relevant code — what it does, why it’s structured that way, what could break.&lt;br&gt;
    3.  You pick an answer.&lt;br&gt;
    4.  If you get it right, Claude proceeds with the implementation. If you don’t, it explains the gap and quizzes you again — no shortcut to “just give me the code.”&lt;/p&gt;

&lt;p&gt;The point isn’t to slow you down for its own sake. It’s to make sure that when something breaks in prod six weeks from now, you’re not staring at a diff you don’t recognize.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Why I built it&lt;/strong&gt;&lt;br&gt;
I kept seeing the same failure mode on my own team: someone ships an AI-authored fix, it works in the demo, and then it quietly breaks something adjacent because nobody — including the person who “wrote” it — understood the full blast radius. Linters and tests catch a lot. They don’t catch “I don’t actually know why this works.”&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Try it&lt;/strong&gt;&lt;br&gt;
It’s MIT-licensed and takes two commands to install:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;/plugin marketplace add schwann2402/quiz-me-skill
/plugin &lt;span class="nb"&gt;install &lt;/span&gt;quiz-me@quiz-me
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Source: github.com/schwann2402/quiz-me-skill&lt;/p&gt;

&lt;p&gt;I’d love feedback — especially from anyone who tries it on a real codebase and finds it too strict, too lenient, or just annoying. Open an issue or drop a comment below.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>productivity</category>
      <category>claude</category>
      <category>career</category>
    </item>
  </channel>
</rss>
