Tetrachrome is a small, self-hosted search project. This page describes, plainly, what happens to your data when you use it — not a template pulled from elsewhere, but a direct account of what the code in this repository actually does.
When you run a search, the query text, the timestamp, the number of results returned, and how long it took to answer are recorded. If you click a result, the URL you clicked and its position in the results are also recorded — this is what powers the admin dashboard's click-through analytics. Specifically, Tetrachrome does not:
Search results link out to pages Tetrachrome has crawled from the public web. Once you click through, you're on someone else's site, subject to their privacy policy — Tetrachrome has no visibility into what happens after you leave. Result snippets, headlines, and preview images shown on Tetrachrome are drawn directly from those source pages; see Content & Copyright for how that content is used and how a site owner can opt out of being indexed at all.
The public search site you're using runs on Google Cloud Run. Crawled pages, extracted text, image metadata, and the query/click logs above are stored in a Postgres database hosted by Supabase; the built search index itself is synced through Google Cloud Storage between the crawler (which runs on a private machine, not publicly reachable) and this public server. Supabase and Google Cloud are infrastructure providers the operator of this project pays to host it — not analytics companies or ad networks, and they receive your query only as a side effect of running the database and server this site is built on, the same way any website's hosting provider does. The crawler and the admin dashboard that manages it are never themselves exposed to the public internet.
If what the code actually collects changes, this page will be updated to match — the intent is for this document to always describe real behavior, not aspirational behavior.