searxng-mcp

🇷🇺 Русская версия

MCP server that exposes SearxNG web search and a handful of fetch tools (web page, GitHub README, article) over the Model Context Protocol.

Features

• web_search_xng — search via one or two SearxNG instances, with routing controlled by the instance argument

• fetch_web_content — download a page and strip HTML

• fetch_github_readme — fetch a repo README from GitHub

• fetch_article — download an article and extract the main text

• Two transports: SSE (/sse + /messages) and direct HTTP (/mcp)

• Configured entirely via environment variables

Related MCP server: websharp-mcp

Instance routing

If only one instance is configured, the instance argument is ignored and every call goes to that instance.

Configuration

At least one of the two instance URLs must be set, otherwise the server exits on startup.

Installation

Docker (recommended)

docker pull ghcr.io/dolphin2702/searxng-mcp:latest

Run

docker run -d
--name searxng-mcp
-p 3006:3006
-e SEARXNG_INSTANCE_RU=https://search.example.ru
-e SEARXNG_INSTANCE_VPS=https://search.example.com
ghcr.io/dolphin2702/searxng-mcp:latest

Deployment examples

Two instances (RU + VPS): examples/docker-compose.two-instances.yml One instance: examples/docker-compose.one-instance.yml

MCP client configuration

SSE transport

{ "mcpServers": { "searxng": { "url": "http://localhost:3006/sse" } } }

Streamable HTTP

{ "mcpServers": { "searxng": { "type": "streamable", "url": "http://localhost:3006/mcp" } } }

Tools

web_search_xng

Search the web via SearxNG.

Arguments:

• q (string, required) — search query

• count (number, default 5) — number of results

• instance (string, default default) — default / ru / vps / all

Returns: markdown-formatted list of results.

fetch_web_content

Fetch a page and return its text content with HTML stripped.

Arguments:

• url (string, required)

• max_chars (number, default 30000)

fetch_github_readme

Fetch README.md from a GitHub repository (tries main, then master).

Arguments:

• url (string, required) — repository URL

fetch_article

Fetch an article and extract the main text (strips header, footer, nav, scripts).

Arguments:

• url (string, required)

HTTP API

Development

git clone git@github.com:dolphin2702/searxng-mcp.git cd searxng-mcp npm install SEARXNG_INSTANCE_VPS=https://search.example.com node server.js

License

MIT