SearXNG Public Scraper

The SearXNG Public MCP server solves a common pain point for developers who want to give their AI assistants access to the vast array of public SearXNG instances without dealing with inconsistent output formats. Traditional MCP implementations for SearXNG rely on the engine’s native JSON API, which is fast but unavailable on many public deployments that only expose a web interface. This server bridges that gap by scraping the HTML results from up to three public instances, normalising them into a clean JSON payload that AI assistants can consume directly.

At its core, the server offers a single endpoint that accepts a query string along with optional parameters for time range, language, and a detailed flag. When is enabled, the server performs a more exhaustive crawl: it queries three distinct SearXNG servers, fetches the first three pages of each, and merges the results while eliminating duplicates. The output is a concise array of objects, each containing a and a short . This format is immediately usable by tools such as Claude, allowing the assistant to present search results or follow links without additional parsing logic.

Developers benefit from several key capabilities. First, the server abstracts away the need to maintain a list of working public instances; it automatically falls back to secondary servers if one is unreachable. Second, the language and time‑range options let assistants tailor searches to user preferences or context (e.g., recent news in Spanish). Third, the mode is ideal for scenarios that require deeper coverage—think investigative research or compliance checks where breadth of results matters. Finally, because the server outputs plain JSON, it integrates seamlessly into any MCP‑compatible workflow without extra adapters or custom parsers.

Real‑world use cases abound. A customer support chatbot could query a public SearXNG instance to fetch up‑to‑date troubleshooting articles, then summarize them for the user. A research assistant could gather recent academic papers across multiple domains by setting the to “day” and enabling detailed search. Privacy‑conscious applications can rely on this server to avoid sending queries to proprietary engines, keeping user data off commercial platforms while still delivering relevant results.

In summary, the SearXNG Public MCP server turns a fragmented web‑based search engine into a unified, AI‑ready service. By handling scraping, fallback logic, and result normalisation internally, it empowers developers to embed robust, privacy‑respecting search capabilities into their AI assistants with minimal effort.