searxng_skill

Overview

This skill connects to a local SearXNG instance to perform privacy-respecting metasearch across web, images, news, videos and more. It runs searches locally against your SearXNG JSON API, producing human-readable or machine-friendly JSON output. The skill avoids external API dependencies and preserves search privacy by using your own instance.

How this skill works

The skill sends queries to the SearXNG JSON endpoint specified by the SEARXNG_URL environment variable (defaults to http://localhost:8080). You can request different categories (images, news, videos), adjust result count, set language and time-range, and output results as formatted text or as JSON for programmatic consumption. No authentication is required by default unless your instance enforces it.

When to use it

• When you want private web searches routed through your self-hosted SearXNG instance.
• When aggregating results across multiple engines without using external search APIs.
• When you need image, news, or video search categories in addition to web results.
• When integrating search into scripts or tools that prefer JSON output.
• When benchmarking or archiving search results from a local search endpoint.

Best practices

• Set SEARXNG_URL to the full JSON API endpoint of your instance and verify connectivity before running queries.
• Use JSON output (--format json) for downstream automation or logging to ensure consistent parsing.
• Limit result counts for large batch jobs to avoid overloading your SearXNG instance.
• Specify language and time-range filters for more relevant and repeatable results.
• Protect your SearXNG instance behind access controls if exposed to untrusted networks.

Example use cases

• Quickly fetch top 10 web results for a research query while keeping searches private.
• Retrieve the latest news items for a topic using the news category and a day time-range.
• Automate image discovery by querying the images category and piping JSON to an image processor.
• Collect search snapshots for archival or analysis by running repeated queries and storing JSON outputs.
• Integrate local metasearch into chatbots or personal assistants without external search APIs.