SearXNG
otsumami.searxng
A Docker Compose file for SearXNG alone, generated from the catalog and checked by the safety inspector. No account needed.
Metasearch engine that queries many upstream search engines on the user's behalf and merges the results, forwarding no identifying information and storing no search history.
SearXNG sits between a person and the search engines it queries, issuing requests itself so that the upstream engine sees the instance rather than the individual, and returning a merged, deduplicated result set. Which engines are consulted is configurable per category and per query, and results can be filtered by time range, language and safe-search level. Because it exposes a JSON API as well as an HTML interface, it is commonly used as the retrieval layer behind local research and agent workflows rather than only as a browser search page. A first deployment sets the instance secret and base URL, then confirms a query returns results from several engines.
You know it worked when
- The search page loads at the configured base URL.
- A query returns results attributed to more than one upstream engine.
- The engine preferences page lists engines as enabled and responding.
- A JSON API request returns structured results once that format is enabled.
- No search history is retained between sessions.
Known sharp edges
- Upstream engines rate-limit and block instances that query them heavily, so engines drop out of results over time and the failure looks like missing results rather than an error.
- The JSON API format is disabled by default and must be enabled explicitly before any programmatic client can use the instance.
- A secret key must be set before first start, and the shipped placeholder leaves session handling predictable.
- A publicly reachable instance attracts automated query traffic that will get its address blocked by upstream engines, degrading it for everyone using it.