npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

ui-compare-mcp

v1.2.3

Published

UI图片精准对比MCP服务 - 基于算法的像素级差异检测

Downloads

584

Readme

UI Compare MCP

基于算法的 UI 图片精准对比 MCP 服务 - 像素级、结构级、感知级多维度差异检测

npm version License: MIT

特性

  • 精确量化: 像素级、色值级、结构级多维度对比
  • 智能过滤: 自动过滤抗锯齿、渲染差异等误报
  • 可视化输出: 差异热力图、标注图(Base64)
  • 纯内存处理: 默认不保存文件,直接返回 JSON 数据
  • MCP 集成: 无缝集成到 Cursor IDE
  • 批量对比: 支持多组图片批量处理

核心算法

| 算法 | 用途 | 输出 | |------|------|------| | SSIM | 结构相似性检测 | 0-1 分数 | | Pixel Diff | 像素级差异(LAB 色差) | 差异率 + 像素数 | | Color Histogram | 颜色分布对比 | Chi-square 距离 | | Edge Detection | 边缘结构对比 (Canny) | 边缘差异率 | | Anti-aliasing Filter | 抗锯齿过滤 | 过滤误报 |

快速开始

1. 在 Cursor 中配置

编辑 Cursor 配置文件(~/Library/Application Support/Cursor/User/globalStorage/@cursor-server/settings.json):

{
  "mcpServers": {
    "ui-compare": {
      "command": "npx",
      "args": ["-y", "ui-compare-mcp"]
    }
  }
}

2. 重启 Cursor

配置后重启 Cursor IDE,MCP 服务将自动启动。

3. 使用示例

在 Cursor 中,Agent 可以直接调用 MCP 工具:

// Agent 自动调用
const result = await ui_compare({
  design_image: "/path/to/design.png",
  screenshot_image: "/path/to/screenshot.png",
  threshold: 0.1,
  return_heatmap: true
});

// 查看结果
console.log(`相似度: ${result.metrics.overall_similarity * 100}%`);
console.log(`检测到 ${result.diff_regions.length} 个差异区域`);

// Claude 可以看到 Base64 图片
// result.images.heatmap - 热力图
// result.images.annotated - 标注图

本地开发

1. 安装依赖

npm install
pip install -r requirements.txt

3. 本地测试

node index.js

4. Cursor 本地配置

{
  "mcpServers": {
    "ui-compare": {
      "command": "node",
      "args": ["/path/to/ui-compare-mcp/index.js"]
    }
  }
}

MCP 工具

ui_compare - 单次对比

对比两张图片并返回详细差异分析。

参数:

| 参数 | 类型 | 必填 | 默认值 | 说明 | |------|------|------|--------|------| | design_image | string | ✅ | - | 设计图路径 | | screenshot_image | string | ✅ | - | 截图路径 | | threshold | number | ❌ | 0.1 | 像素差异阈值 (0-1) | | anti_aliasing | boolean | ❌ | true | 是否启用抗锯齿过滤 | | return_heatmap | boolean | ❌ | true | 是否返回热力图 | | return_annotated | boolean | ❌ | true | 是否返回标注图 | | color_tolerance | number | ❌ | 10 | 颜色容差 (Delta E) |

返回:

{
  "success": true,
  "metrics": {
    "overall_similarity": 0.8745,
    "ssim_score": 0.8923,
    "pixel_diff_rate": 0.1234,
    "color_distance": 23.45,
    "edge_diff_rate": 0.0876,
    "total_diff_pixels": 45678,
    "image_size": [1920, 1080]
  },
  "diff_regions": [
    {
      "region": "region_1",
      "bbox": [100, 50, 400, 100],
      "issues": [
        {
          "type": "color",
          "metric": "color_distance",
          "value": 15.3,
          "severity": "medium",
          "description": "颜色差异 Delta E: 15.30"
        }
      ]
    }
  ],
  "images": {
    "heatmap": "data:image/png;base64,...",
    "annotated": "data:image/png;base64,..."
  }
}

ui_compare_batch - 批量对比

批量对比多组图片。

参数:

{
  "comparisons": [
    {
      "design_image": "/path/to/design1.png",
      "screenshot_image": "/path/to/screenshot1.png",
      "name": "homepage"
    },
    {
      "design_image": "/path/to/design2.png",
      "screenshot_image": "/path/to/screenshot2.png",
      "name": "about"
    }
  ],
  "threshold": 0.1,
  "return_images": false
}

返回:

{
  "success": true,
  "results": [
    {
      "name": "homepage",
      "metrics": {...},
      "diff_regions": [...]
    }
  ],
  "summary": {
    "total": 2,
    "passed": 2,
    "failed": 0,
    "avg_similarity": 0.89
  }
}

指标解读

整体相似度 (overall_similarity)

综合评分 (0-1):

  • > 0.95: 高度相似
  • 0.85-0.95: 基本相似
  • 0.70-0.85: 中等差异
  • < 0.70: 明显差异

SSIM 分数 (ssim_score)

结构相似性指数 (0-1):

  • 衡量布局和结构的相似程度
  • 对光照、对比度变化不敏感

像素差异率 (pixel_diff_rate)

差异像素占比 (0-1):

  • < 0.05: 几乎无差异
  • 0.05-0.15: 轻微差异
  • > 0.15: 明显差异

颜色距离 (color_distance)

颜色分布差异(Chi-square 距离):

  • < 10: 颜色分布一致
  • 10-50: 颜色有差异
  • > 50: 颜色分布明显不同

边缘差异率 (edge_diff_rate)

结构轮廓差异 (0-1):

  • 反映布局和形状的差异

使用场景

1. UI 还原验收

// 截取页面
const screenshot = await browser_take_screenshot({
  filename: "current_page.png"
});

// 对比设计稿
const result = await ui_compare({
  design_image: "designs/homepage.png",
  screenshot_image: screenshot.path
});

// Claude 分析差异并生成修复建议

2. 视觉回归测试

// 批量对比多个页面
const result = await ui_compare_batch({
  comparisons: [
    { design_image: "baseline/home.png", screenshot_image: "current/home.png", name: "home" },
    { design_image: "baseline/about.png", screenshot_image: "current/about.png", name: "about" }
  ]
});

// 检查是否有回归
if (result.summary.failed > 0) {
  console.log("发现视觉回归!");
}

3. 设计稿走查

// 精确对比细节
const result = await ui_compare({
  design_image: "design.png",
  screenshot_image: "impl.png",
  threshold: 0.05,  // 更严格
  color_tolerance: 5
});

// 输出差异报告
for (const region of result.diff_regions) {
  console.log(`区域 ${region.region}:`);
  for (const issue of region.issues) {
    console.log(`- ${issue.description}`);
  }
}

阈值调优

| 场景 | threshold | color_tolerance | 说明 | |------|-----------|-----------------|------| | 图标/Logo 对比 | 0.05 | 5 | 要求严格 | | 文本页面对比 | 0.15 | 12 | 字体渲染容差 | | 色彩设计对比 | 0.08 | 8 | 色彩准确性 | | 布局结构对比 | 0.12 | 15 | 关注结构而非色彩 | | 整页对比 | 0.10 | 10 | 平衡准确性和效率 |

系统要求

  • Node.js: >= 14.0.0
  • Python: >= 3.8
  • 操作系统: macOS, Linux, Windows

Python 依赖

自动安装以下依赖:

  • mcp>=0.9.0
  • opencv-python>=4.8.0
  • numpy>=1.24.0
  • scikit-image>=0.21.0
  • Pillow>=10.0.0

故障排查

Q: 提示 "未找到 Python 3"?

A: 确保已安装 Python 3.8+:

python3 --version

Q: 依赖安装失败?

A: 手动安装依赖:

pip3 install -r requirements.txt

Q: 图片读取失败?

A: 检查文件路径是否正确,支持绝对路径和相对路径。

Q: MCP 服务启动失败?

A: 查看 Cursor 的 MCP 日志:

cat ~/Library/Application\ Support/Cursor/logs/mcp-*.log

开发

运行测试

npm test

构建

npm run build

发布

npm publish

许可证

MIT License - 详见 LICENSE 文件

贡献

欢迎提交 Issue 和 Pull Request!

致谢

算法实现参考了以下开源项目: