ScaiLabs Search

About ScaiLabs Search

ScaiLabs Search is a privacy-respecting metasearch backend operated by ScaiLabs. It aggregates results from many upstream search engines, normalises them into a single feed, and serves that feed either through this web interface or as a clean HTTP API for AI agents and other downstream applications.

We built it for two reasons:

  1. As the search backbone for our own AI services. Retrieval-augmented generation, agent tooling, and our internal research workflows all need fresh, neutral, attributable web results. Operating our own metasearch layer means we are not at the mercy of any one search vendor’s pricing, rate limits, or terms of service — and our users’ queries never leave the European jurisdiction we host in.

  2. So others can use it too. If you are building anything that needs the live web as input — an AI assistant, a market-monitoring tool, an academic crawler, a fact-checker — and you would rather not stand up your own scraping infrastructure, talk to us. The same backend that powers ScaiLabs internally is available to partners through an authenticated API.

See the API documentation for endpoints, authentication, and example requests.

Light and dark mode

Use the moon/sun toggle in the top-right of the header to flip between light and dark themes. Your choice is stored locally in a small scailabs_theme cookie (path /, max age 1 year, SameSite=Lax) and is applied before the page paints, so there’s no flash of the wrong palette on subsequent visits. Clear the cookie to fall back to your system’s prefers-color-scheme setting.

What this instance does not do

Built on SearXNG

ScaiLabs Search is an adaptation of the open-source SearXNG metasearch project. All the heavy lifting — engine integrations, query parsing, result merging, locale handling — is SearXNG’s work, and credit for that belongs to the SearXNG community and its contributors. Our changes are limited to:

The full source for this adaptation will be published once we have finished cleaning it up — until then, refer to the upstream SearXNG sources for the underlying engine.

If you are interested in SearXNG itself — running your own instance, contributing engines, or improving translations — head straight to the SearXNG project and the SearXNG documentation. Both are excellent.

How does it work?

ScaiLabs Search inherits SearXNG’s design as a metasearch engine: it forwards each query to a curated set of upstream search engines in parallel, collects their responses, deduplicates URLs, scores them, and returns the merged result. The list of active engines and the API configuration of this instance is exposed at /config.

Source code: SearXNG sources