SearXNG MCP Server

Overview

SearXNG MCP Server is a lightweight, open-source Model Context Protocol (MCP) server that bridges AI assistants (Claude, Cursor, LM Studio, local LLMs, etc.) with SearXNG — the popular self-hosted, privacy-respecting metasearch engine.

It allows agents to perform web searches without relying on commercial APIs like Google or Bing, keeping queries private and free from tracking, profiling, or paywalls. Multiple community implementations exist, with the most popular being the actively maintained Node.js version by ihor-sokoliuk.

Key Features

• Privacy-First Search: Aggregates results from dozens of search engines (Google, Bing, DuckDuckGo, etc.) without user tracking.
• Configurable SearXNG Instance: Works with any public or self-hosted SearXNG instance via its JSON API.
• Pagination Support: Fetch additional pages of results for comprehensive research.
• Safe Search Control: Filter explicit content with configurable levels.
• Structured Results: Returns clean, parsed results optimized for LLM consumption.
• Easy Deployment: Run via npx, Docker, or locally with Node.js/Python/Go variants available.
• Multi-Client Compatibility: Works seamlessly with Claude Desktop, Cursor, Windsurf, and other MCP-compatible tools.

Use Cases

• Grounded Research: Let Claude fetch up-to-date information for accurate answers instead of relying on outdated training data.
• Local/Private AI Workflows: Combine with Ollama or LM Studio for completely offline-capable (except search) agent setups.
• Coding Assistance: Search documentation, libraries, or error messages in real time during development.
• News & Current Events: Get the latest articles and summaries without ads or sponsored content.
• Multi-Tool Orchestration: Pair with other MCP servers (browser, databases, code execution) for powerful autonomous agents.

How It Works

1. Run the MCP server (local process or Docker container).
2. Configure your MCP client (e.g., Claude Desktop) to connect to it.
3. The AI discovers the searxng_web_search tool automatically.
4. When the agent needs fresh web data, it calls the tool via the MCP protocol.
5. The server queries your chosen SearXNG instance and returns structured results.

Quick Installation & Setup (Most Popular Node.js Version)

Using npx (easiest):

Add the following to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "searxng": {
      "command": "npx",
      "args": ["-y", "mcp-searxng"],
      "env": {
        "SEARXNG_URL": "https://your-searxng-instance.example.com"
      }
    }
  }
}

Self-hosted SearXNG recommended for maximum privacy (run via Docker from official SearXNG repo).

Docker option:

docker pull mcpcommunity/ihor-sokoliuk-mcp-searxng:latest

Other notable implementations:

• Python/uv variants (e.g., The-AI-Workshops)
• Go version for high performance
• Enhanced forks with scraping or category support

Configuration Options

• SEARXNG_URL: URL of your SearXNG instance (public or private).
• Proxy and authentication support in several forks.
• Safe search level and result categories (general, news, images, etc.).

Restart your MCP client after adding the server. The tool should appear automatically in the agent's available tools.

Common Tips

• Use a self-hosted SearXNG instance for best privacy and reliability.
• Public instances work for quick testing but may have rate limits or downtime.
• Combine with page-fetching tools for deeper research (many forks include content extraction).

SearXNG MCP is one of the most widely used privacy-focused web search tools in the MCP ecosystem, making it essential for anyone building or using local-first or privacy-conscious AI agents.

For the latest version and full documentation, visit the primary repository.