Searcharvester MCP Server
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Searcharvester MCP Server
这是一个自托管网页搜索与页面提取服务适配器,基于 Model Context Protocol (MCP) 运行。 你可以将自建的 Searcharvester 桥接到 Claude Code、Claude Desktop 等大模型客户端,作为它们的原生网络搜索及提取工具。
✨ 特性
多端同步配置:无需在多台电脑上配置复杂的绝对路径,支持通过
npx远程拉起,免安装且配置天然同步。内网优先与网络容灾:支持“内网优先 -> 外网备份”的双重备灾降级。若配置了内网并畅通,搜索秒级返回;若内网离线,无缝切换到外网代理并自动伪装 User-Agent 进行最多 3 次重试。
协同回退原生工具:当内外网服务全部挂掉时,工具会自动输出友好引导警示,引导 Claude 临时降级调用自带的
WebSearch和WebFetch兜底。自动级联分页拼接:提取超长网页时(
size: "f"),MCP 服务会在后台自动请求所有切片页并拼接整合成完整的 Markdown 输出给模型,免除大模型多次调用工具拉取分页的消耗。
Related MCP server: web-mcp
🛠️ 环境变量配置
MCP Server 在拉起时将读取以下环境变量进行寻址和鉴权:
环境变量 | 必须 | 说明 | 示例 |
| 是 | 你的 Searcharvester 服务校验 Key |
|
| 否 | 自建局域网内部部署的地址 |
|
| 否 | 自建外部公网的代理或反代节点 |
|
💡 提示:
SEARCHARVESTER_INTERNAL_URL与SEARCHARVESTER_EXTERNAL_URL至少配置一项。
🚀 多端一键配置使用 (Claude Code / Claude Desktop)
由于你的代码已打包为独立仓库,在所有的设备中,你的全局配置文件(例如 ~/.claude/settings.json)只需配入以下参数。npx 会在两台电脑的后台全自动获取最新代码并跑起来,保证随时同步:
{
"mcpServers": {
"searcharvester": {
"command": "npx",
"args": ["-y", "github:MaYunFei/searcharvester-mcp"],
"env": {
"SEARCHARVESTER_API_KEY": "你的验证KEY",
"SEARCHARVESTER_INTERNAL_URL": "http://内网地址:8001",
"SEARCHARVESTER_EXTERNAL_URL": "https://你的外部公网地址"
}
}
}
}🧰 提供的 Tool List
一经连接,客户端将可以使用以下两个工具:
1. searcharvester_search
利用自托管的聚合搜索平台查询网络。
参数:
query(string, 必须): 查询词。max_results(number): 返回条数(支持 1 - 20,默认 10)。engines(string): 选填,覆盖默认的聚合引擎(如google,duckduckgo)。categories(string): 选填,分类类型(如news|images|science)。
2. searcharvester_extract
对提取的目标链接进行 trafilatura 无噪精研提取。
参数:
url(string, 必须): 需要抓取的页面地址。size(string): 限制大小。's' (约5k字符), 'm' (10k), 'l' (25k), 'f' (全部内容,自动后台拼接多页文件)。
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Web search and any page as LLM-ready Markdown for AI agents. Free keyless tier, no signup.
Hosted Web Crawler MCP server for Claude, ChatGPT and Cursor, with each action switched per account.
Web search and content extraction for AI: Google results, any page as Markdown or JSON
Web search and page extraction for AI agents, paid per call with x402 (USDC on Base).
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceProvides web search and page fetching tools for third-party models (e.g., DeepSeek, Qwen, Kimi) in Claude Code, with configurable search backends, Markdown output, and safety boundaries.69 PyPI63MIT
- AlicenseAqualityBmaintenanceGives AI agents local web search and page reading with no API keys, layering a SearXNG-backed search cache with browser-based fallbacks and returning clean page text from HTML, PDF, JSON, and archived snapshots. Fetches run in Docker alongside the search engine with SSRF protection, egress filtering, and untrusted-content marking so agents can safely retrieve live web data.2MIT
- AlicenseAqualityBmaintenanceSelf-hosted web search and page extraction for AI agents. Stdio MCP server, HTTP API and Python/npm CLI.852 npm2MIT
- AlicenseNot gradedqualityBmaintenanceGives AI assistants keyless web search, news search, and full-page reading tools by querying multiple engines in parallel and returning merged results. It can run locally over stdio or remotely over HTTP for Claude, Cursor, and other agents.4MIT