Searpa Documentation#
Searpa is a privacy-first metasearch engine: one query blends the best results from several independent sources into a single clean, fast page, with no tracking, no ads and no profiling. Its search engine is open source, so anyone can audit how it works or run their own copy.
These docs cover three audiences:
User guide#
Everything you need to get the most out of Searpa: searching the web, the Images / News / Videos / Maps / Translate tabs, instant answers, knowledge cards, bangs and per-account settings. → Start with Getting started.
Self-hosting guide#
Run your own Searpa instance. Requirements, a Docker quick start, every configuration option, wiring up search providers, creating users and issuing API keys, plus production and maintenance notes. → Start with Quick start.
API guide#
Use every Searpa search feature programmatically over a JSON API:
authentication, rate limits and a reference for each endpoint, with copy-paste
curl and Python examples.
→ Start with the API overview.
What Searpa does#
- Web search across three engines, Brave, Mojeek and Marginalia. Enable any subset; when more than one is active their result lists are merged with Reciprocal Rank Fusion and deduplicated, so a result several engines agree on rises to the top.
- Images, News, Videos, Maps and Translate tabs, each backed by independent providers.
- Instant answers above the results, a calculator, unit and currency conversion, weather, world clock, QR codes, hashes and more, most computed locally on the server.
- Knowledge cards beside the results, from Wikipedia, TMDB, TripAdvisor and Stack Exchange.
- Per-account settings that sync across devices, with export and import.
- A public JSON API covering every one of those features.
Privacy in one line#
Searpa does not track you, show ads or build a profile. API keys for upstream providers stay on the server and are never exposed to your browser, and queries are sent to providers without identifying you.