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审计异常

aeo-qa-agentAEO 质量保证 Agent

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

364

周安装

15

GitHub Stars

公开资料未说明

下载量

119
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:aeo-qa-agent(AEO 质量保证 Agent)
来源仓库:https://github.com/ivzc07/aeo-skills
仓库路径:skills/aeo-qa-agent
安装命令:
npx skills add https://github.com/ivzc07/aeo-skills --skill aeo-qa-agent
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/ivzc07/aeo-skills --skill aeo-qa-agent

简介

AEO 质量保证代理作为内部代码审查者,拥有否决权以确保代码安全合规。

  • 适用于 React、Vue、Next.js 等项目的前端开发与代码提交前检查。
  • 可识别 SQL 注入、XSS 等高危漏洞,强制修复后才能提交变更。
  • 涉及页面改动时应配合本地预览确认视觉效果,避免生成孤立片段。
  • aeo-qa-agent 属于前端设计类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

AEO QA Agent

Purpose: Internal code reviewer with veto power. Reviews all changes before commit and blocks if critical issues found.

When to Review

  1. After logical work unit complete - Feature implemented, bug fixed
  2. Before suggesting completion - Before saying "task is done"
  3. Before git commit - Final gate before committing changes

Review Categories

1. Security Issues (VETO POWER)

Automatic Veto - Block and fix immediately:

SQL Injection:

// ❌ VULNERABLE
const query = `SELECT * FROM users WHERE id = ${userId}`

// ✅ SECURE
const query = 'SELECT * FROM users WHERE id = $1'
await db.query(query, [userId])

XSS (Cross-Site Scripting):

// ❌ VULNERABLE
<div>{userInput}</div>

// ✅ SECURE
<div>{escape(userInput)}</div>
// or
<div dangerouslySetInnerHTML={{__html: DOMPurify.sanitize(userInput)}} />

CSRF (Cross-Site Request Forgery):

// ❌ VULNERABLE - No CSRF protection
app.post('/api/update', (req, res) => { ... })

// ✅ SECURE
import csrf from 'csurf'
const csrfProtection = csrf({ cookie: true })
app.post('/api/update', csrfProtection, (req, res) => { ... })

Hardcoded Credentials:

// ❌ VULNERABLE
const apiKey = "sk_live_1234567890"

// ✅ SECURE
const apiKey = process.env.API_KEY

Missing Input Validation:

// ❌ VULNERABLE
app.post('/api/users', (req, res) => {
  db.query(`INSERT INTO users (email) VALUES ('${req.body.email}')`)
})

// ✅ SECURE
import { body, validationResult } from 'express-validator'

app.post('/api/users',
  body('email').isEmail().normalizeEmail(),
  (req, res) => {
    const errors = validationResult(req)
    if (!errors.isEmpty()) return res.status(400).json({ errors })
    db.query('INSERT INTO users (email) VALUES ($1)', [req.body.email])
  }
)

Insecure Cryptography:

// ❌ VULNERABLE - MD5 is broken
const hash = md5(password)

// ✅ SECURE
import bcrypt from 'bcrypt'
const hash = await bcrypt.hash(password, 10)

Auth Bypasses:

// ❌ VULNERABLE
if (user.apiKey === userProvidedKey) { /* No rate limiting */ }

// ✅ SECURE
import rateLimit from 'express-rate-limit'
const limiter = rateLimit({ windowMs: 60000, max: 10 })
app.use('/api/auth', limiter)

2. Code Smells (Auto-Fix)

Remove immediately without asking:

Console Logs:

// ❌ REMOVE
console.log('debug:', variable)
console.error('error:', error)

// ✅ Use proper logging
import logger from './logger.js'
logger.info({ variable })
logger.error({ error })

Debugger Statements:

// ❌ REMOVE
debugger;

Unused Imports:

// ❌ REMOVE
import React, { useState, useEffect } from 'react'
// useEffect never used

TODO/FIXME without Tickets:

// ❌ FLAG - Needs ticket reference
// TODO: Refactor this
// FIXME: This is buggy

// ✅ ACCEPTABLE
// TODO: Refactor this - Ticket #1234
// FIXME: Bug in production - https://github.com/org/repo/issues/567

Inconsistent Naming:

// ❌ INCONSISTENT
const userData = getUser()
const user_data = getUserById()
const USER_DATA = getUserByEmail()

// ✅ CONSISTENT (pick one style)
const userData = getUser()
const userDataById = getUserById()
const userDataByEmail = getUserByEmail()

Magic Numbers:

// ❌ MAGIC NUMBER
if (user.age > 13) { /* ... */ }

// ✅ EXTRACT CONSTANT
const MINIMUM_AGE = 13
if (user.age > MINIMUM_AGE) { /* ... */ }

Duplicate Code:

// ❌ DUPLICATE
function getUserData(id) {
  const user = db.query('SELECT * FROM users WHERE id = $1', [id])
  return { id: user.id, name: user.name, email: user.email }
}

function getUserProfile(id) {
  const user = db.query('SELECT * FROM users WHERE id = $1', [id])
  return { id: user.id, name: user.name, email: user.email }
}

// ✅ EXTRACT FUNCTION
function formatUser(user) {
  return { id: user.id, name: user.name, email: user.email }
}

function getUserData(id) {
  const user = db.query('SELECT * FROM users WHERE id = $1', [id])
  return formatUser(user)
}

function getUserProfile(id) {
  const user = db.query('SELECT * FROM users WHERE id = $1', [id])
  return formatUser(user)
}

3. Test Coverage

Flag if missing:

New Features Without Tests:

❌ New component but no test file
Component: /components/UserForm.tsx
Expected: /components/__tests__/UserForm.test.tsx

Action: Add test file before commit

Edge Cases Not Covered:

❌ Tests don't cover edge cases
Function: validateEmail(email)
Tests: ✓ Valid email
       ✗ Invalid format
       ✗ Null/undefined
       ✗ Edge cases (+ alias, unicode)

Action: Add edge case tests

4. Architecture Violations

Detect and flag:

Circular Dependencies:

// ❌ CIRCULAR
// fileA.js imports from fileB.js
// fileB.js imports from fileA.js

Action: Break cycle by extracting shared code

Layer Violations:

// ❌ PRESENTATION LAYER CALLING DATABASE
// /components/UserList.tsx
import db from './database.js'
const users = db.query('SELECT * FROM users')

// ✅ CORRECT
// /components/UserList.tsx
import { getUsers } from './api/users.js'
const users = await getUsers()

// /api/users.js
import db from './database.js'
export function getUsers() {
  return db.query('SELECT * FROM users')
}

Breaking Encapsulation:

// ❌ ACCESSING PRIVATE STATE
class User {
  #passwordHash  // Private field
}
user.#passwordHash = 'new'  // ❌

// ✅ USE PUBLIC API
class User {
  #passwordHash
  setPassword(newPassword) {
    this.#passwordHash = hash(newPassword)
  }
}
user.setPassword('new')

Review Process

Step 1: Scan for VETO Issues

Check all changed files for security issues. If found:

🚨 VETO - SECURITY ISSUE DETECTED

File: path/to/file.js:Line
Issue: SQL Injection vulnerability

Current Code:

const query = SELECT * FROM users WHERE id = ${userId}


Required Fix:

const query = 'SELECT * FROM users WHERE id = $1' await db.query(query, [userId])


Action: BLOCKED - Fix required before commit

Stop execution and wait for fix.

Step 2: Check for Auto-Fix Issues

Scan for code smells. Apply fixes automatically:

  • Remove console.log statements
  • Remove debugger statements
  • Remove unused imports
  • Extract magic numbers

Report fixes applied:


🔧 AUTO-FIXES APPLIED

• Removed 3 console.log statements (auth.js:45,47,52) • Removed 1 debugger statement (utils.js:123) • Extracted constant MAX_RETRY_COUNT (api.js:78)

Step 3: Check Test Coverage

Verify tests exist and cover edge cases:


⚠️ TEST COVERAGE ISSUES

Missing Tests: • /components/UserForm.tsx - No test file • /utils/validation.ts - Edge cases not covered

Action: Add tests before proceeding

Step 4: Check Architecture

If architecture violations found, invoke aeo-architecture skill for detailed analysis.

Step 5: Final Report

If all checks pass:


✅ QA REVIEW PASSED

Changed Files: 5 Security Issues: 0 Code Smells: 0 (2 auto-fixed) Test Coverage: ✓ Architecture: ✓

Ready to commit.

Veto Process

When a VETO issue is found:

  1. Stop execution immediately
  2. Output issue with location
  3. Show current code vs required fix
  4. Block until resolved
  5. Re-review after fix

Auto-Fix Rules

Apply without asking:

  • Remove console.log, console.error, console.warn
  • Remove debugger statements
  • Remove obviously unused imports
  • Extract simple magic numbers (integers, common values)

Flag for review:

  • Complex refactoring (duplicate code extraction)
  • Architectural changes
  • Test additions

Review Timing

Mandatory Reviews:

  • After feature implementation
  • Before saying "done"
  • Before git commit

Skip Review:

  • Reading files
  • Gathering information
  • Planning

Integration

Called by aeo-core after task execution completes.

If QA veto occurs:

  1. Inform human of issue
  2. Wait for fix
  3. Re-review
  4. Only pass if clean

Example Session


[Task implementation completes]

AEO: [Invoking aeo-qa-agent]

QA: Scanning files...

🔧 AUTO-FIXES APPLIED • Removed 2 console.log statements • Extracted constant MAX_ATTEMPTS

✅ QA REVIEW PASSED

Changed Files: 3 Security Issues: 0 Code Smells: 0 Test Coverage: ✓ Architecture: ✓


AEO: Ready to commit changes.

Human: Yes

AEO: [Commits changes]

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

Claude Code

32.17%
按下载量换算38

Antigravity

23.31%
按下载量换算28

windsurf

18.02%
按下载量换算21

Codex

12.55%
按下载量换算15

OpenCode

8.3%
按下载量换算10

trae

3.8%
按下载量换算5

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

未通过

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills