quiz

分类: 开发与编程 | 上传者: akshan-main | 下载: 0 | 版本: v1.0（最新）

On-demand product comprehension quiz about your current code changes

更新日志: Source: GitHub https://github.com/akshan-main/vibe-check

目录结构

当前层级: templates/project/.claude/skills/quiz/

📄 SKILL.md 3.5 KB

SKILL.md

---
name: quiz
description: On-demand product comprehension quiz about your current code changes
---

Run a VibeCheck quiz right now based on the current git diff.

STEP 1: Run these git commands to gather context:
- `git diff --name-only` and `git diff --staged --name-only` to get changed files
- `git diff --unified=3` and `git diff --staged --unified=3` to get the diff (truncate to 2000 chars if very long)

If there are no changes, say "No code changes to quiz on. Make some changes first!" and stop.

STEP 2: Analyze the diff and create ONE multiple-choice question.

IMPORTANT: Focus on the MOST IMPORTANT change, not the largest one. A 2-line behavior change can matter more than 50 lines of boilerplate. Read the entire diff, use your understanding of the code and the product, and pick the single change that has the most meaningful impact on what users experience. Ignore formatting, imports, renaming, and refactors that don't change behavior.

Classify what happened:
- Was a feature ADDED? (new capability that didn't exist before)
- Was a feature CHANGED? (existing behavior now works differently)
- Was something REMOVED? (capability or safeguard that's now gone)
- Was it a FIX? (broken thing that now works)

Then ask a question that tests whether the developer understands the REAL-WORLD IMPACT of this specific change on their product. Vibe coders build products - they need to understand what their product does, not how to read code.

QUESTION FORMULA - pick one:
  * WHAT CHANGED: "Before this change, [X happened]. What happens now instead?"
  * WHAT'S NEW: "A user tries [action] for the first time. What do they experience?"
  * WHAT'S GONE: "You removed [feature/check/step]. What can users do now that they couldn't before - or what protection is no longer there?"
  * SIDE EFFECTS: "This change also affects [related area]. What's different there now?"
  * EDGE CASE: "A user does [unusual but realistic action]. How does your app handle it after this change?"
  * LIMITS: "What's the maximum/minimum [value/count/size] this feature now supports? What happens at the boundary?"
  * DATA: "After this change, what new data is being stored/sent/exposed? Who can see it?"

NEVER ASK:
  * About code syntax, language features, or programming concepts
  * About which library or framework is used
  * Anything a developer would need to read code to answer - the question should be answerable by someone who understands the PRODUCT but not the code
  * Generic questions unrelated to this specific diff

WRONG ANSWERS: Each should be a plausible misunderstanding of what the product change does. Things a developer might assume if they didn't pay attention to what Claude actually built.

Format: exactly 4 options (labels "A", "B", "C", "D"), one correct. Ask via AskUserQuestion with header: "VibeCheck", multiSelect: false.

STEP 3: After the user answers:
1. Explain the correct answer in plain language - what the product does now and why
2. If wrong: explain what they misunderstood about the change and what their answer would have meant for users
3. PROMPTING TIP: You have the full conversation context - you know what the user asked for and what you built. Compare those. If their prompt was vague and the implementation has gaps or surprises they might not expect, suggest a more specific prompt that would have covered those gaps. If their prompt was already detailed and the implementation matches well, say so. Don't fabricate issues.

End your message with: [vibecheck:done]

IMPORTANT: Do NOT use Edit, Write, or any code-modifying tools. This is learning-only.

登录后下载/点赞/收藏 ❤ 14 | ★ 0

quiz

目录结构

SKILL.md

举报内容

提示