Token导航 LogoToken导航

Agent Skill · 研究检索

quiz(quiz 问题管理)

quiz 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

quiz 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 56,417。

总安装 454周安装 17GitHub Stars 33下载 140核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:quiz(quiz 问题管理)
来源仓库:https://github.com/mblode/agent-skills
仓库路径:skills/quiz
安装命令:
npx skills add https://github.com/mblode/agent-skills --skill quiz
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/mblode/agent-skills --skill quiz

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

quiz 用于查找、检索和筛选相关信息,支持关键词定位。适合在 Codex、Claude、Cursor、Gemini CLI 中需要快速定位候选结果时使用。通过 GitHub 安装,使用 npx skills add 命令添加指定仓库的 skill。安装前需确认权限范围、维护状态,以及是否会触发联网或文件读写。

quiz 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/quiz

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/mblode/agent-skills --skill quiz;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

quiz 怎么安装?

在终端执行 npx skills add https://github.com/mblode/agent-skills --skill quiz。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

quiz 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

quiz 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

quiz 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

quiz 用的人多吗?

公开数据是累计安装 454 次,近一周安装 17 次,下载 140 次,GitHub 33 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

quiz 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
quiz

quiz 问题管理 · 本页

45433只读未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Quiz

Generate interactive quizzes about the current codebase to test understanding.

Reference Files

FileRead When
references/question-templates.mdDefault: question types by difficulty with concrete examples
references/difficulty-calibration.mdApplying difficulty and quantity settings, exploration depth, question stems
references/scoring-and-output.mdStep 5: scoring rubric, result formatting, final output template

Configuration

Collect from the user (ask only what was not provided):

Number of Cards

OptionCardsBest for
Fewer5Quick check, focused review
Standard10Balanced coverage
More15Deep dive, comprehensive review

Default: Standard (10)

Difficulty

LevelFormatQuestion style
EasyMultiple choiceWhat does X do? Where is Y defined?
MediumMultiple choice + explanationHow does X interact with Y? What pattern is used for Z?
HardOpen-ended + explanationWhat would break if X changed? How would you refactor Y?

Default: Medium

Workflow

Copy this checklist to track progress:

Quiz progress:
- [ ] Step 1: Gather configuration
- [ ] Step 2: Explore the codebase
- [ ] Step 3: Generate questions
- [ ] Step 4: Present quiz interactively
- [ ] Step 5: Score and summarize

Step 1: Gather configuration

Ask the user for card count and difficulty. Use defaults if not specified. Accept informal phrasing ("give me a hard quiz", "just 5 questions").

If the user provides both values upfront, skip straight to Step 2.

Step 2: Explore the codebase

Systematically scan the project to build a question pool:

  • Read the project structure (directories, key files, entry points)
  • Identify the primary language(s), framework(s), and tooling
  • Read key files: configuration, entry points, core modules, tests
  • Map dependencies and module boundaries
  • Note patterns: naming conventions, architecture style, error handling
  • Find non-obvious details: edge cases, workarounds, gotchas

Load references/difficulty-calibration.md for exploration depth and question strategy per level.

Step 3: Generate questions

Using references/question-templates.md, generate the configured number of questions at the selected difficulty. Apply calibration from references/difficulty-calibration.md.

Each question must:

  • Reference real code from the current codebase (file paths, function names, patterns)
  • Have one clear correct answer
  • Be answerable from the codebase alone (no external knowledge required)
  • Cover different parts of the codebase (avoid clustering)

For multiple-choice questions, provide 4 options (A, B, C, D) with plausible distractors drawn from real names in the project.

For medium and hard questions, prepare a brief explanation to show after the user answers.

Check every question against the quality checklist below. Remove or rewrite any that fail.

Step 4: Present quiz interactively

Present one question at a time. For each question:

  1. Show the question number and total as a header (e.g., "Question 3/10")
  2. For multiple choice (easy/medium): use AskUserQuestion with the question text and 4 options so the user can select an answer interactively
  3. For open-ended (hard): show the question text and wait for a free-text response
  4. After the user responds, show whether the answer was correct
  5. Show the explanation with file:line references (medium and hard only)
  6. Proceed to the next question

Step 5: Score and summarize

Load references/scoring-and-output.md for the result template.

After all questions are answered, present the final scorecard with score, rating, breakdown by area, and recommended next steps.

Quality Checklist

Every question must pass all of these:

  • Codebase-grounded — References a real file path, function, or pattern from the current project
  • Single focus — Tests one concept. If you need "and" to describe it, split it.
  • Unambiguous — Has exactly one correct answer. No trick questions.
  • Self-contained — Answerable from the codebase alone, no external knowledge required
  • Plausible distractors — Multiple-choice options use real names from the project, not random strings
  • Spread coverage — Questions cover different files and areas, no clustering
  • Difficulty-matched — Matches the cognitive level of the selected difficulty
  • No generated files — Never asks about lock files, node_modules, or build output
  • Explanation ready — Medium and hard questions have a prepared explanation with file:line references

Anti-patterns

  • Do not ask questions about external libraries or frameworks (only the project's own code)
  • Do not ask questions that require running the code to answer
  • Do not reveal all questions at once (present one at a time, wait for response)
  • Do not skip the explanation on wrong answers (the quiz is for learning)
  • Do not ask trick questions or ambiguous questions with multiple valid answers
  • Do not repeat questions about the same file or function
  • Do not generate questions before exploring the codebase (Step 2 must complete first)

Related Skills

  • plan-feature for understanding codebase architecture before quizzing
  • review-pr for code review exercises on recent changes

平台分布

Codex

31.36%
按下载量换算44

Claude

31.19%
按下载量换算44

Cursor

20%
按下载量换算28

Gemini CLI

9.54%
按下载量换算13

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills