Token导航 LogoToken导航TokenDH.com
前端设计需要联网github未标认证来源可访问clear审计通过

qa-discussion质量保证讨论

Agent Skill

用于辅助前端页面、组件、样式和交互逻辑的开发与维护。它适合让 Agent 生成或审查 React、Next.js、Vue、Tailwind、CSS 等相关代码,整理组件结构,或定位布局和性能问题。使用时需要结合项目现有设计系统、路由和构建方式,避免只生成孤立片段;涉及页面改动时,应配合本地预览和构建检查确认视觉效果。

总安装

894

周安装

38

GitHub Stars

52

下载量

313
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

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:qa-discussion(质量保证讨论)
来源仓库:https://github.com/zenobi-us/dotfiles
仓库路径:skills/qa-discussion
安装命令:
npx skills add https://github.com/zenobi-us/dotfiles --skill qa-discussion
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/zenobi-us/dotfiles --skill qa-discussion

简介

qa-discussion 用于辅助前端页面、组件和样式开发,适合生成或审查 React、Vue 等代码。

  • 适用于组件结构整理、布局问题定位和性能优化支持。
  • 通过 npx skills add 命令从指定 GitHub 仓库安装使用。
  • 需结合项目路由和构建方式,避免生成孤立片段。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Q&A Discussion Conductor

Overview

Systematically conduct structured Q&A discussions with users. Handles multiple question types, tracks progress, validates responses, and generates comprehensive summary reports. Perfect for assessments, onboarding, feedback collection, or educational interactions.

Core Capabilities

Question Types:

  • Multiple Choice: Select one or more correct answers from options
  • Single Choice: Select exactly one correct answer from options
  • Open-Ended: Free-form text responses with optional validation

Session Features:

  • Progress tracking (Question X of Y)
  • One question at a time for clarity
  • Smart re-prompting on invalid answers with helpful feedback
  • Complete response logging with validation status
  • Summary report after discussion completion

When to Use

Use this skill when you need to:

  • Conduct structured assessments or quizzes
  • Gather feedback through systematic questioning
  • Onboard users with validated knowledge checks
  • Create educational interactions with tracking
  • Run audits or compliance checks requiring documented responses

Don't use when:

  • Questions are casual or conversational (just chat naturally)
  • You don't need to validate or track responses
  • The discussion is exploratory rather than structured

The Process

Step 1: Define Questions

Create a structured question array with all necessary metadata:

const questions: Question[] = [
  {
    id: 'q1',
    text: 'What is the capital of France?',
    type: 'single',
    options: ['London', 'Paris', 'Berlin', 'Madrid'],
    correctAnswer: 'Paris'
  },
  {
    id: 'q2',
    text: 'Select all that apply: Which are JavaScript frameworks?',
    type: 'multiple',
    options: ['React', 'Django', 'Vue', 'Laravel'],
    correctAnswers: ['React', 'Vue']
  },
  {
    id: 'q3',
    text: 'Explain why you chose your answer above',
    type: 'open',
    validation: (answer) => answer.length >= 10
  }
];

Step 2: Conduct the Discussion

Call the main function to start:

const session = await conductDiscussion(questions);

The skill will:

  1. Display each question clearly with progress indicator
  2. Wait for user response
  3. Validate against question criteria
  4. Re-prompt if invalid with helpful feedback
  5. Store validated response
  6. Move to next question
  7. Generate summary when complete

Step 3: Review Results

Access the session data for reporting:

// Session structure:
{
  totalQuestions: number;
  completedQuestions: number;
  responses: Array<{
    questionId: string;
    text: string;
    answer: string;
    isValid: boolean;
    attempts: number;
  }>;
  summary: {
    totalAttempts: number;
    validResponses: number;
    invalidResponses: number;
    completionTime: number;
  };
}

Display Conventions

Question Presentation:

  • Clear title showing question number and total (e.g., "Question 1 of 5")
  • Question text prominently displayed
  • Options numbered for easy reference (1, 2, 3...)
  • Instructions vary by type:

- Single choice: "Select one option (enter number)" - Multiple choice: "Select all that apply (enter numbers separated by commas)" - Open-ended: "Enter your response (minimum 10 characters)"

Validation Feedback:

  • Invalid input: "Invalid selection. Please enter 1, 2, 3, or 4"
  • Multiple choice format error: "Invalid format. Use: 1,2,3"
  • Open-ended validation failed: "Response too short. Please provide more detail"
  • Empty input: "Response cannot be empty"

Progress Display:

  • Each question shows: "Question 2 of 7"
  • Visual spacing between questions for clarity
  • Summary displayed after final question

Error Handling

Input Validation:

  • Out of range selections → re-prompt with valid range
  • Wrong format (single choice given multiple) → explain expected format
  • Empty responses → ask to provide an answer
  • Custom validation failures → explain criteria with helpful message

Session Errors:

  • No questions provided → throw clear error
  • Invalid question type → validate during setup
  • Timeout/interruption → save progress and allow resume

Key Principles

  • One question at a time - Never show multiple questions together
  • Clear feedback - Always explain why an answer is invalid
  • Complete tracking - Log every response with validation status
  • Flexible validation - Support exact match, list validation, and custom functions
  • User-friendly - Numbered options, clear instructions, helpful messages
  • Resumable - Design allows pausing and resuming if needed

Example Usage

See @index.ts for complete implementation with:

  • Full TypeScript types for all question types
  • Comprehensive error handling
  • Real-world usage examples
  • Integration patterns

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenCode

29.32%
按下载量换算92

Antigravity

21.91%
按下载量换算69

Claude Code

15.97%
按下载量换算50

Gemini CLI

10.68%
按下载量换算33

windsurf

7.57%
按下载量换算24

Codex

3.15%
按下载量换算10

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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。

来源信息

继续浏览同类 Skills