SearXNG — Privacy-First Metasearch Engine

What it is

SearXNG is a free, self-hosted metasearch engine that aggregates results from over 70 search engines and sources. It sends queries to multiple backends (Google, Bing, DuckDuckGo, Wikipedia, and others), combines the results, and returns them without tracking, profiling, or advertising. SearXNG is a maintained fork of the original SearX project.

SearXNG targets privacy-conscious users, self-hosters, and developers building AI search applications. As a self-hosted backend, it provides a clean search API that AI agents can query without API key limits or tracking.

How it saves time or tokens

SearXNG provides a unified search API across 70+ sources. Instead of integrating with Google Custom Search API, Bing API, and others individually, you query one SearXNG instance. The JSON API returns structured results that are ready for LLM consumption. No rate limits beyond what your server can handle, and no per-query API costs.

How to use

1. Deploy SearXNG with Docker: docker run -d --name searxng -p 8080:8080 searxng/searxng.
2. Open http://localhost:8080 for the web interface.
3. Use the JSON API at http://localhost:8080/search?q=query&format=json for programmatic access.

Example

# Deploy with Docker
docker run -d --name searxng \
  -p 8080:8080 \
  -v ./searxng:/etc/searxng \
  -e SEARXNG_BASE_URL=http://localhost:8080/ \
  searxng/searxng

# Web search via browser
# Open http://localhost:8080

# JSON API for AI applications
curl 'http://localhost:8080/search?q=python+web+frameworks&format=json'

# Use SearXNG as AI search backend
import requests

def search(query: str) -> list:
    resp = requests.get(
        'http://localhost:8080/search',
        params={'q': query, 'format': 'json'}
    )
    return resp.json()['results']

Related on TokRepo

• Research Tools — AI-powered research and search tools
• Self-Hosted Tools — Self-hosted infrastructure

Common pitfalls

• SearXNG depends on upstream search engines. If Google or Bing blocks your server's IP (common with high query volumes), result quality degrades. Rotate IPs or use proxy configurations.
• Default configuration may not enable all search engines. Edit settings.yml to activate the sources you need.
• Public SearXNG instances get rate-limited by upstream engines. For reliable AI search, run your own private instance.