Serper Search MCP Server
Serper 搜索 MCP 服务器
一个模型上下文协议 (MCP) 服务器,通过 Serper API 提供 Google 搜索功能,以及一个由 AI 驱动的深度研究工具。该服务器可轻松将搜索和研究功能集成到您支持 MCP 的应用程序中。
✨ 特点
🌐 通过 Serper API 实现强大的 Google 搜索集成
🔄 丰富的搜索响应数据包括:
知识图谱信息
自然搜索结果
“人们还问”问题
相关搜索
🧠 人工智能深度研究工具:
执行多步骤、迭代研究
生成子查询以彻底探索主题
综合来自多个来源的信息
提供所有信息的引文
可调节的研究深度水平
内置质量指标,持续改进
🛠可配置的搜索参数:
国家定位
语言选择
结果计数定制
自动更正选项
🔒 安全 API 密钥处理
⚡️ 速率限制和缓存支持
📝 TypeScript 支持完整的类型定义
📊 集成性能指标以优化结果
Related MCP server: google-search-mcp
🚀 安装
克隆存储库:
git clone https://github.com/yourusername/serper-search-server.git
cd serper-search-server安装依赖项:
pnpm install构建服务器:
pnpm run build⚙️ 配置
从 [Serper.dev]( https://serper.de获取您的 Serper API 密钥
在根目录中创建一个
.env文件:
# Required
SERPER_API_KEY=your_api_key_here
# Optional - Advanced Quality Metrics Configuration (pre-configured by default)
USAGE_METRICS_KEY=your-custom-metrics-key # Optional
USAGE_PROJECT_ID=your-custom-project-id # Optional
METRICS_ENDPOINT=https://your-custom-host.com # Optional
DISABLE_METRICS=false # Not recommended有关以下方面的详细信息,请参阅TELEMETRY.md :
质量指标收集
性能监控
使用情况分析
仪表板设置
持续改进
🔌 集成
克劳德桌面
将服务器配置添加到您的 Claude Desktop 配置中:
MacOS : ~/Library/Application Support/Claude/claude_desktop_config.json Windows : %APPDATA%/Claude/claude_desktop_config.json
{
"mcpServers": {
"serper-search-server": {
"command": "/path/to/serper-search-server/build/index.js",
"env": {
"SERPER_API_KEY": "your_api_key_here"
}
}
}
}🛠 使用方法
搜索工具
服务器提供了一个强大的搜索工具,其参数如下:
{
"query": string, // Search query
"numResults"?: number, // Number of results (default: 10, max: 100)
"gl"?: string, // Country code (e.g., "us", "uk")
"hl"?: string, // Language code (e.g., "en", "es")
"autocorrect"?: boolean, // Enable autocorrect (default: true)
"type"?: "search" // Search type (more types coming soon)
}深度研究工具
为了满足更全面的研究需求,服务器提供了一个深度研究工具,可以使用以下参数进行多步骤研究:
{
"query": string, // Research query or question
"depth"?: "basic" | "standard" | "deep", // Research depth (default: "standard")
"maxSources"?: number // Maximum sources to include (default: 10)
}深度研究工具:
将复杂查询分解为重点子查询
执行多次搜索以收集全面的信息
使用人工智能综合来自多个来源的信息
使用适当的引用和参考来格式化结果
根据中间结果调整研究策略
收集匿名质量指标以改善搜索结果
深度级别:
基础:快速概览(3-5 个来源,约 5 分钟)适合:简单的事实、快速的定义、直接的问题
标准:综合分析(5-10 个来源,约 10 分钟)适合:大多数研究需求,深度和速度均衡
深度:详尽的研究(10+ 个来源,约 15-20 分钟)适合:复杂主题、学术研究、透彻分析
搜索工具示例响应
搜索结果包含丰富的数据:
{
"searchParameters": {
"q": "apple inc",
"gl": "us",
"hl": "en",
"autocorrect": true,
"type": "search"
},
"knowledgeGraph": {
"title": "Apple",
"type": "Technology company",
"website": "http://www.apple.com/",
"description": "Apple Inc. is an American multinational technology company...",
"attributes": {
"Headquarters": "Cupertino, CA",
"CEO": "Tim Cook (Aug 24, 2011–)",
"Founded": "April 1, 1976, Los Altos, CA"
}
},
"organic": [
{
"title": "Apple",
"link": "https://www.apple.com/",
"snippet": "Discover the innovative world of Apple...",
"position": 1
}
],
"peopleAlsoAsk": [
{
"question": "What does Apple Inc mean?",
"snippet": "Apple Inc., formerly Apple Computer, Inc....",
"link": "https://www.britannica.com/topic/Apple-Inc"
}
],
"relatedSearches": [
{
"query": "Who invented the iPhone"
}
]
}🔍 响应类型
知识图谱
包含可用的实体信息:
标题和类型
网站网址
描述
关键属性
自然搜索结果
搜索结果列表包括:
标题和 URL
片段(描述)
在结果中的位置
附加链接(如有)
人们还问
与搜索相关的常见问题:
问题文本
答案片段
来源链接
相关搜索
用户经常进行的相关搜索查询列表。
📊 质量指标
深度研究工具包括综合质量指标:
研究过程指标
性能监控
问题跟踪
使用模式
结果质量指标
有关为提高搜索质量而收集的指标的详细信息,请参阅TELEMETRY.md 。
🤝 贡献
欢迎贡献代码!欢迎提交 Pull 请求。
📝 许可证
该项目根据 MIT 许可证获得许可 - 有关详细信息,请参阅LICENSE文件。
🙏 致谢
Serper API提供 Google 搜索功能
MCP 框架的模型上下文协议
PostHog的分析功能
Available Tools
1 toolserper-google-searchB
Perform a Google search using the SERPER API. Returns rich search results including knowledge graph, organic results, related questions, and more.
Inputs: query (required), gl, hl, numResults, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 1 tool
There is only one tool, so there is no possibility of selecting the wrong one. Its purpose (run a Google search via Serper) is clearly stated.
The single name 'serper-google-search' is descriptive and uses a consistent kebab-case style, but it is a compound noun rather than a verb_noun action pattern. With only one tool there is no broader convention to judge against.
One tool is thin for a server whose stated purpose is exposing the Serper search API, which supports multiple result types. It is functional but leaves the surface minimal.
A general web search covers the core use case and returns rich payloads, but the Serper API's other endpoints (news, images, places/maps, scholar, webpage scraping) are absent. Agents needing those result types have no path forward with this server.
Maintenance
Related MCP Connectors
Unofficial Serper MCP — wraps the Serper Google Search API (serper.dev)
MCP server for Google search results via SERP API
Unofficial SerpApi MCP — wraps SerpApi (serpapi.com) search engines
Google SERP API MCP: live Google, Bing, AI Overview and Google Maps results as JSON. Pay per call.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables Google Search, News, and Places queries via the Serper API, integrated with the Pipeworx MCP gateway.324 npmMIT
- AlicenseNot gradedqualityDmaintenanceEnables LLMs to perform real-time Google searches and retrieve web results via the MCP protocol.-
- FlicenseNot gradedqualityBmaintenanceProvides Google search and web page fetching tools to MCP clients via the Serper API, enabling models to search and read pages during conversations.-
- AlicenseAqualityCmaintenanceEnables AI agents to run live Google searches and retrieve organic results, snippets, sitelinks, People Also Ask, related searches, and Google's AI Overview with citations via a zero-dependency MCP server.2MIT