Token导航 LogoToken导航

Agent Skill · 前端设计

iblai-test(伊布莱测试)

用于辅助测试设计、自动化测试、用例整理和回归验证。它适合让 Agent 编写单元测试、端到端测试、测试计划或根据失败日志定位问题。使用时需要确认项目测试框架、运行命令和夹具数据,避免为了通过测试而改坏真实逻辑;涉及浏览器或外部服务时,应区分本地模拟、测试环境和生产环境。

iblai-test 是前端设计类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为接触敏感数据;许可证信息还不明确,第三方扫描有未通过项,按安装量在前端设计类的 10,216 个 Skill 里排第 4,410。

总安装 894周安装 38GitHub Stars 14下载 313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:iblai-test(伊布莱测试)
来源仓库:https://github.com/iblai/vibe
仓库路径:skills/iblai-test
安装命令:
npx skills add https://github.com/iblai/vibe --skill iblai-test
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/iblai/vibe --skill iblai-test

这个 Skill 是做什么的

前端设计类 Skill 关注页面结构、视觉规范和组件写法,输出通常是可运行的界面代码。

PNPM 构建 -- must pass with zero errors pnpm test -- vitest must pass 询问用户测试凭据(电子邮件+密码) Touch test every page -- pnpm test:e2e for agent apps, or the 剧作家登录 + 适用于普通应用程序的屏幕截图脚本(每页等待 10 秒) 阅读每个屏幕截图并修复任何设计问题 Fix any failures before showing work to the user Full E2E Reference 用于编写自定义的经过身份验证的旅程测试、身份验证设置和 multi-browser patterns, see: https://github.com/iblai/iblai-app-cli/blob/main/skills/testing/iblai-add-test.md Brand guidelines : BRAND.md 每周安装量 38 存储库 iblai/vibe GitHub 之星 14 第一次看到 2026 年 3 月 30 日 安全审计 Gen Agent Trust Hub 警告 插座警告 斯尼克失败

来源仓库内路径:skills/iblai-test

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/iblai/vibe --skill iblai-test;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

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

iblai-test 怎么安装?

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

iblai-test 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。

iblai-test 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

iblai-test 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。

iblai-test 用的人多吗?

公开数据是累计安装 894 次,近一周安装 38 次,下载 313 次,GitHub 14 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。

iblai-test 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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。
  • 需要处理前端设计类任务、又不想每次都手写提示词的人。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 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。

前端设计类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
frontend-design

前端设计 · Codex、Claude

8,823,360126,072只读MIT通过
vercel-composition-patterns

Vercel 组合模式 · Codex、Claude

3,858,81625,893需要联网MIT通过
vercel-react-native-skills

Vercel React native skills 前端 · Codex、Claude

2,468,66425,878只读MIT通过
adapt

adapt 搜索 · Codex、Claude

2,027,04023,125操作浏览器未明确通过
bolder

bolder 分析 · Codex、Claude

2,004,28823,130只读未明确通过
iblai-test

伊布莱测试 · 本页

89414接触敏感数据未明确未通过

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

/iblai-test

You MUST run these checks before presenting any work to the user. Do NOT show untested code.

Use pnpm as the default package manager. Fall back to npm if pnpm is not installed.

Step 1: Build

Run the production build to catch TypeScript errors, missing imports, and config issues:

pnpm build

Fix all errors. A failed build means broken code -- do not proceed until the build passes cleanly.

Step 2: Run Unit Tests

pnpm test

This runs vitest to verify:

  • All @source paths in CSS files resolve to existing directories
  • The lib/iblai/sdk symlink is valid and targets the SDK dist
  • lib/iblai/sdk/web-containers/source contains compiled JS for Tailwind class generation

If tests fail, the SDK symlink may be broken. Run ls -la lib/iblai/sdk to check. If it's missing, run iblai add auth again to recreate it.

Step 3: Get Test Credentials

Before running E2E tests or authenticated screenshots, ask the user for their ibl.ai login credentials:

  1. Email -- the ibl.ai account email
  2. Password -- the account password

Store them in e2e/.env.development (create the file if it doesn't exist):

PLAYWRIGHT_USERNAME=user@example.com
PLAYWRIGHT_PASSWORD=their-password

If the user declines to provide credentials, skip authenticated page screenshots and only test unauthenticated pages (home, SSO callback).

Step 4: Touch Test Pages

If this is an agent app (from iblai startapp agent)

The app already has a full Playwright E2E suite in e2e/. Use it:

pnpm test:e2e

This runs auth setup + journey tests across Chromium, Firefox, and WebKit.

For a single browser:

npx playwright test --config e2e/playwright.config.ts --project=chromium

If this is a vanilla Next.js app

Install Playwright and a browser:

pnpm add -D @playwright/test
npx playwright install --with-deps chromium

--with-deps installs the browser binary AND system libraries (libatk, libcups, etc.). Without it the browser may fail to launch on Linux/CI.

Start the dev server:

pnpm dev &

Wait for http://localhost:3000 to be ready.

Login with Playwright before taking screenshots

Most pages require authentication. Write and run a Playwright script that logs in through the ibl.ai Auth SPA and saves the browser state, then takes screenshots of every page with a 10-second wait:

// e2e/visual-test.ts
import { chromium } from '@playwright/test';

const APP = 'http://localhost:3000';
const AUTH = 'https://login.iblai.app';

(async () => {
  const browser = await chromium.launch();
  const context = await browser.newContext();
  const page = await context.newPage();

  // --- Login ---
  await page.goto(APP, { waitUntil: 'domcontentloaded', timeout: 30000 });

  // AuthProvider redirects to the auth SPA
  await page.waitForURL(url => url.href.includes('/login'), { timeout: 60000 });

  // Click "Continue with Password"
  const pwdBtn = page.getByRole('button', { name: /continue with password/i });
  await pwdBtn.waitFor({ timeout: 30000 });
  await pwdBtn.click();

  // Fill credentials from env
  await page.getByRole('textbox', { name: /email/i }).fill(process.env.PLAYWRIGHT_USERNAME!);
  await page.getByLabel(/password/i).fill(process.env.PLAYWRIGHT_PASSWORD!);
  await page.getByRole('button', { name: /^continue$/i }).click();

  // Wait for redirect back to the app with auth tokens
  await page.waitForURL(url => url.href.startsWith(APP), { timeout: 60000 });
  await page.waitForFunction(() => !!localStorage.getItem('axd_token'), { timeout: 30000 });

  // --- Screenshot every page (10s wait each) ---
  const pages = [
    ['/', '/tmp/home.png'],
    ['/profile', '/tmp/profile.png'],
    ['/account', '/tmp/account.png'],
    ['/analytics', '/tmp/analytics.png'],
    ['/notifications', '/tmp/notifications.png'],
  ];

  for (const [path, file] of pages) {
    await page.goto(`${APP}${path}`, { waitUntil: 'domcontentloaded' });
    await page.waitForTimeout(10000);
    await page.screenshot({ path: file, fullPage: true });
  }

  await browser.close();
})();

Run it:

npx dotenv -e e2e/.env.development -- npx tsx e2e/visual-test.ts

If dotenv-cli is not installed, run:

pnpm add -D dotenv-cli tsx

Add every page you created to the pages array. If any screenshot shows an error page, fix it and re-run.

Kill the dev server when done.

Step 5: Review Screenshots for Design Issues

After taking screenshots, read every screenshot image and inspect for:

  • Broken layouts -- elements overlapping, overflowing containers, misaligned columns, collapsed sections with zero height
  • Missing content -- blank pages, empty cards, "Loading..." stuck forever, placeholder text not replaced
  • Style problems -- wrong colors (brand blue is #0058cc), unstyled raw HTML, missing icons, broken images, invisible text on same-color backgrounds
  • Responsive issues -- horizontal scrollbar, content cut off, elements too small or too large
  • SDK component issues -- components rendering outside their white wrapper, no border/rounding on SDK cards, SDK styles breaking page layout
  • Navigation problems -- navbar not visible, active tab not highlighted
  • Typography -- text too small, headings same size as body, missing spacing between sections

For each issue found, fix it in the code and re-screenshot to verify. Do NOT present work to the user with visible design problems.

How to Tell Which App Type

  • Agent app: has e2e/ directory with playwright.config.ts and auth.setup.ts, and package.json has test:e2e script
  • Vanilla Next.js app: no e2e/ directory, no test:e2e script

Summary

  1. pnpm build -- must pass with zero errors
  2. pnpm test -- vitest must pass
  3. Ask the user for test credentials (email + password)
  4. Touch test every page -- pnpm test:e2e for agent apps, or the Playwright login + screenshot script for vanilla apps (10s wait per page)
  5. Read every screenshot and fix any design issues
  6. Fix any failures before showing work to the user

Full E2E Reference

For writing custom authenticated journey tests, auth setup, and multi-browser patterns, see: https://github.com/iblai/iblai-app-cli/blob/main/skills/testing/iblai-add-test.md

Brand guidelines: BRAND.md

平台分布

Codex

35.31%
按下载量换算111

Claude

30.23%
按下载量换算95

Cursor

21.87%
按下载量换算68

Gemini CLI

9.59%
按下载量换算30

安全审计

Gen Agent Trust Hub

可疑

Socket

可疑

Snyk

未通过

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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