Queries a self-hosted SearxNG instance to return JSON or URL search results.
npx skills add https://github.com/ythx-101/ask-search --skill ask-search
Please help me install this Agent Skill. Skill: ask-search Source: https://github.com/ythx-101/ask-search/tree/main Command: npx skills add https://github.com/ythx-101/ask-search --skill ask-search
AI agents often rely on external web search APIs, which can leak queries and incur costs. ask-search provides a private, self-hosted web search solution powered by SearxNG, returning title, URL, and snippet in a privacy-preserving way.
Run ask-search against a locally running SearxNG instance to fetch the initial results.
High-intent search queries and answers about installing and using this skill.
To perform private web searches for AI agents without leaking queries, use a self-hosted SearxNG instance. This setup aggregates multiple engines locally, returning titles, URLs, and snippets while preserving query privacy and requiring zero external API keys.
You do not need API keys to search multiple engines locally with SearxNG. The setup aggregates results from Google, Bing, DuckDuckGo, and Brave directly. Optional Tavily integration is available if you want to expand your search capabilities further.
To integrate local SearxNG search into an MCP-compatible workflow, configure the environment with your local SEARXNG_URL. The tool then operates within OpenClaw or Claude Code environments, outputting raw JSON or URL-only lists for further agent processing.
You can filter SearxNG search results by language or category for agent workflows using the flexible CLI tool. It fetches top results and outputs raw JSON or URLs, allowing agents to easily deep-dive selected links with web_fetch for detailed analysis.
Limitations of using a self-hosted SearxNG instance for agent search include relying entirely on your local infrastructure's uptime and configuration. The tool handles errors gracefully but requires a properly running local SEARXNG_URL to function and return results.
The best way to fetch web content for analysis after retrieving search URLs is to use a web_fetch tool. The ask-search workflow retrieves top result URLs, allowing your agent to select and deep-dive into specific links for full content extraction.
devskaleQuery a SearXNG instance via Python HTTP requests with filters.
abdullah1854Crawl websites and extract content into Markdown, JSON, or text.
sweeden-ttuBuild, test, and debug MCP servers with FastMCP and MCP Inspector.
skaile-aiQuery a local SearXNG instance and return web results as structured markdown.
