SearXNG MCP Server

@aicrafted

代码仓库

基本信息

分类

其他

传输方式

stdio

作者

aicrafted

提交者

Nik

关于 SearXNG MCP Server

SearXNG MCP Server

#search #searxng #self-hosted

SearXNG MCP Server 配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "searxng": {
      "type": "http",
      "url": "http://localhost:32123/mcp"
    }
  }
}

SearXNG MCP Server 工具

工具列表暂未提供

当前收录暂时没有可展示的工具列表，不代表服务没有工具。请查阅项目文档了解支持的能力。

点击上方按钮,直接连接服务器端点实时获取工具列表。

概览

What is SearXNG MCP?

SearXNG MCP is a Docker‑based MCP (Model Context Protocol) server that wraps SearXNG, a privacy‑focused metasearch engine. It gives LLM clients access to aggregated search results from over 130 engines via tools such as searxng_search and searxng_get_info.

How to use SearXNG MCP?

Run both SearXNG and the MCP server together using the provided Docker Compose file. Configure your MCP client with an HTTP transport pointing to http://localhost:32123/mcp. The server exposes two tools: searxng_search for querying and searxng_get_info for dynamically discovering enabled categories and engines.

Key features of SearXNG MCP

Uses the searxng_search tool with category filtering (general, images, videos, news, map, IT, science, files, social media).
Queries over 130 engines, including Google, Brave, Wikipedia, GitHub, Reddit, and more.
Supports advanced search parameters: categories, engines, language, pageno, time_range, and safesearch.
Provides a searxng_get_info tool for programmatic discovery of available categories and engines.
Runs via Docker Compose with simple environment variable configuration.

Use cases of SearXNG MCP

Retrieve web, image, video, news, or map results for a given query.
Narrow searches to specific domains (e.g., IT or science) or engines.
Filter results by time range (day, month, year) and language.
Control safe search strictness (0=None, 1=Moderate, 2=Strict).
Let the LLM dynamically learn which search categories and engines are enabled.

常见问题

What is SearXNG?

SearXNG is a privacy‑focused metasearch engine that aggregates results from over 130 search engines without tracking users.

How do I configure the MCP transport?

Set the environment variables `MCP_TRANSPORT=http`, `MCP_HOST=0.0.0.0`, and `MCP_PORT=32123`. The client then connects via HTTP at the URL `http://localhost:32123/mcp`.

What tools does the server expose?

It exposes two tools: `searxng_search` for performing searches and `searxng_get_info` for retrieving the list of currently enabled categories and engines.

What parameters are available in `searxng_search`?

The tool supports `categories` (comma‑separated), `engines`, `language`, `pageno`, `time_range` (`day`, `month`, `year`), and `safesearch` (0–2).

Does the server require its own SearXNG instance?

Yes. The MCP server depends on a running SearXNG instance, which it connects to via the `SEARXNG_URL` environment variable.

其他分类下的更多 MCP 服务器

Yuppsy

Yuppsy MCP: build and publish real estate websites from your AI assistant. Yuppsy lets your AI assistant create live websites for real estate agents and individual property listings. You don't need a designer or any code

FastMCP v2 🚀

jlowin

🚀 The fast, Pythonic way to build MCP servers and clients.

26.0K

Glasswarp

See and control a real Windows PC you own — from any MCP client, locally or remotely. Observe (UIA + screenshots), click/type/drag/scroll, launch apps, owner Live View. BYOH: your machine, your key.

Codelf

unbug

A search tool helps dev to solve the naming things problem.

14.2K

Activepieces

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

23.2K