Give Your AI Agents Privacy-First Web Search

Your AI agents need current information, but you don't want to hand over search queries to Big Tech APIs. MCP-searxng connects your agentic systems to SearXNG—a privacy-respecting metasearch engine that aggregates results from multiple sources without tracking.

The API Key Problem

Most developers building AI agents face the same dilemma: you need web search capabilities, but Google and Bing APIs come with usage limits, costs, and privacy concerns. Every search query gets logged and associated with your API key.

MCP-searxng sidesteps this entirely. Connect your agents to SearXNG instances (your own or public ones) and get comprehensive search results without API keys, rate limits, or privacy compromises.

Dead Simple Integration

Add web search to Claude Desktop in under 2 minutes:

{
  "mcpServers": {
    "searxng": {
      "command": "uvx", 
      "args": ["mcp-searxng"]
    }
  }
}

That's it. Your AI agents can now search the web using SearXNG's aggregated results from dozens of search engines.

Real-World Applications

Research Automation: Build agents that gather current information on emerging technologies, market trends, or competitive analysis without hitting API limits.

Self-Hosted AI Workflows: Run your own SearXNG instance and keep all search queries completely private—perfect for sensitive research or proprietary product development.

Multi-Source Intelligence: SearXNG aggregates results from Google, Bing, DuckDuckGo, Startpage, and dozens of other engines, giving your agents broader coverage than single-provider APIs.

Development Without Vendor Lock-in: No API keys to manage, no usage quotas to monitor, no terms of service changes to worry about.

Beyond Basic Search

Set SEARXNG_URL to point at your preferred instance—whether that's a local installation, your company's private instance, or a trusted public server. The flexibility lets you balance privacy, performance, and control based on your specific needs.

The 75 GitHub stars and active development show this isn't just a weekend project—it's become essential infrastructure for developers building privacy-conscious AI systems.

Installation

Install via uvx (recommended):

uvx mcp-searxng

Or clone and run locally:

git clone https://github.com/SecretiveShell/MCP-searxng.git
cd MCP-searxng

Configure your SearXNG endpoint:

export SEARXNG_URL="http://localhost:8080"  # or your preferred instance

Add to your MCP client configuration and start building AI agents that can search without compromise.

Your agents get the web search capabilities they need. You keep control over your data. Everyone wins except the surveillance economy.