Skip to main content
Glama

AI QA Agent MCP

一个基于 Playwright 和 Node.js 构建的、支持 MCP 的 QA 自动化框架,用于通过模型上下文协议(MCP)服务器执行 UI 和 API 自动化测试。

该项目展示了如何将传统测试自动化暴露为可复用的 MCP 工具,使兼容 MCP 的客户端能够触发测试套件并接收结构化的测试结果。

🚀 功能特性

  • Playwright UI 自动化

  • 使用 Playwright 进行 API 测试

  • 页面对象模型(POM)

  • MCP 服务器集成

  • 基于 MCP 工具的测试执行

  • 结构化 JSON 测试结果

  • 登录、库存和结账测试覆盖

  • 正向和负向测试场景

  • 回归测试执行

  • 基于环境的配置

  • HTML 测试报告

  • 失败时截图

  • GitHub Actions CI/CD

  • 在推送和拉取请求时自动执行

Related MCP server: mcp-playwright-test

🧰 技术栈

  • JavaScript

  • Node.js

  • Playwright

  • 模型上下文协议(MCP)

  • MCP Inspector

  • Git

  • GitHub

  • GitHub Actions

🏗️ 架构

                MCP Client / Inspector
                         |
                         v
                    MCP Server
                  (mcp/server.js)
                         |
                         v
                   run_tests Tool
                         |
             +-----------+-----------+
             |                       |
             v                       v
       Test Tool Layer          Suite Selection
       (tools/*.js)        login / inventory /
                           checkout / api /
                           regression
             |
             v
        testRunner.js
             |
             v
          Playwright
         /          \
        v            v
    UI Tests      API Tests
       |
       v
  Page Objects

📁 项目结构

my-ai-qa-agent/
│
├── .github/
│   └── workflows/
│       └── playwright.yml
│
├── data/
│   ├── customer.js
│   └── users.js
│
├── mcp/
│   └── server.js
│
├── pages/
│   ├── LoginPage.js
│   ├── InventoryPage.js
│   ├── CartPage.js
│   └── CheckoutPage.js
│
├── tests/
│   ├── api/
│   │   └── users.api.spec.js
│   ├── login.spec.js
│   ├── inventory.spec.js
│   └── checkout.spec.js
│
├── tools/
│   ├── testRunner.js
│   ├── runLoginTests.js
│   ├── runInventoryTests.js
│   ├── runCheckoutTests.js
│   ├── runApiTests.js
│   └── runRegressionTests.js
│
├── .env.example
├── .gitignore
├── package.json
├── playwright.config.js
└── README.md

pages/

包含可复用的页面对象模型类,封装了 UI 定位器和用户操作。

tests/

包含 Playwright UI 和 API 测试规范。

tools/

充当 MCP 请求与 Playwright 测试执行之间的桥梁。各个工具选择测试套件,而 testRunner.js 执行 Playwright 并汇总结果。

mcp/

包含将 QA 自动化能力暴露为 MCP 工具的 MCP 服务器。

.github/workflows/

包含用于在 CI 中执行自动化测试套件的 GitHub Actions 工作流。

⚙️ 安装

克隆仓库:

git clone https://github.com/bisminizzar84/ai-qa-agent-mcp
cd ai-qa-agent-mcp

安装依赖:

npm install

安装 Playwright 浏览器:

npx playwright install

根据 .env.example 创建本地 .env 文件:

BASE_URL=https://www.saucedemo.com

🧪 运行测试

运行完整测试套件:

npm test

使用可见浏览器运行测试:

npx playwright test --headed

仅运行登录测试:

npx playwright test tests/login.spec.js

运行 API 测试:

npx playwright test tests/api/users.api.spec.js

打开 Playwright HTML 报告:

npx playwright show-report

🤖 MCP 集成

该项目通过 MCP 服务器暴露 QA 自动化能力。

通过 MCP Inspector 启动服务器:

npx -y @modelcontextprotocol/inspector@latest node mcp/server.js

run_tests MCP 工具支持多个套件:

  • login

  • inventory

  • checkout

  • api

  • regression

示例请求:

{
  "suite": "api"
}

示例响应:

{
  "suite": "api",
  "status": "passed",
  "total": 1,
  "passed": 1,
  "failed": 0,
  "skipped": 0,
  "durationMs": 1841
}

MCP 层将测试执行转换为结构化结果,可供兼容 MCP 的客户端使用。

🔄 CI/CD

当代码推送到 main 分支或拉取请求指向 main 分支时,GitHub Actions 会自动执行 Playwright 测试套件。

该流水线执行以下操作:

  1. 仓库检出

  2. Node.js 环境设置

  3. 依赖安装

  4. Playwright 浏览器安装

  5. 自动化测试执行

  6. Playwright 报告上传

这为每次代码变更提供自动化的回归反馈。

🔮 未来增强

  • 将 LLM 连接到 MCP 服务器,实现自然语言测试执行

  • AI 辅助失败分析

  • 自动缺陷摘要

  • 从自然语言需求生成测试

  • 更多 API 覆盖

  • 并行和跨浏览器执行

  • Docker 化测试执行

  • 测试结果通知

Maintenance

ActivitySlowing
ResponsivenessNo issues

Related MCP Connectors

  • Discover Playwright workflows, start runs, and inspect results in Playrunner Cloud.

  • Run, debug, and triage tests from your IDE using natural language, no dashboard switching, no manual data transfers. The TestMu AI (formerly LambdaTest) MCP Server is a single remote server exposing four tool suites: HyperExecute — analyze your project, generate YAML configs and test runner commands, then monitor jobs and sessions. Automation — pull a TestID's details plus command, network, and console logs into one chat for instant root-cause analysis. Includes mobile app upload. SmartUI — explain pixel, layout, DOM, and perceptual changes in a visual regression run, with context-aware React/HTML/CSS fixes. Accessibility — audit any public URL or a local React app against WCAG and get ready-to-apply remediation steps. Connects over https://mcp.lambdatest.com/mcp using OAuth 2.1 — no API keys in your config. One-click install in Cursor; works with Claude, GitHub Copilot, Cline, and any MCP client. Tests execute on the TestMu AI cloud: 3,000+ browsers and 10,000+ real devices.

  • Katalon True Platform MCP: design tests, run with AI, upload reports, assess release readiness.

  • MEOK MCP Test MCP — golden-file + schema-drift + tool-failure tests for any MCP server. Drop-in

Related MCP Servers

Appeared in Searches