This MCP server reads from up to three public SearXNG servers (one main and two fallbacks) and scrapes the HTML search results, since many public instances do not expose a JSON API. It converts the scraped results into JSON so downstream MCP clients can consume them. Installation is via npm install mcp-searxng-public and configuration is done with environment variables: SEARXNG_BASE_URL supports multiple base URLs separated by semicolons, and DEFAULT_LANGUAGE sets the default language for searches. The server exposes a single endpoint, search, with four parameters: query (required), time_range (optional; day, month, or year per the SearXNG spec), language (optional; falls back to DEFAULT_LANGUAGE if not provided), and detailed (optional; when true, performs a more thorough search by querying up to three servers and fetching pages 1-3 from each, then merging and deduplicating). The response is a JSON array of objects with fields url and summary. The README notes you may need to adjust env vars and that a report task can help identify accessible SearXNG instances. An LM Studio install link is provided for easy setup.