How the index works.

Any public website we can reach. A site is never removed because enrichment failed — if we can confirm the domain exists, it stays in the index with whatever data we have, even if that's just a favicon and a hostname.

Domains are discovered from open, public sources only: the Tranco top-1M list, Hacker News, GitHub trending and curated awesome-lists, public RSS feeds, public web directories, and the open web index (Common Crawl).

We never scrape private platforms, bypass paywalls, or use data that requires authentication.

Each new domain is fetched with a plain HTTP request to read its public metadata (title, description, favicon, OpenGraph image). Sites that require JavaScript to render fall back to a headless renderer. Once we have text content, an AI pass infers a category, a region, and a brand color.

None of this changes whether a site is listed. Enrichment failure only changes how the card looks — never whether it appears.

Each listing accumulates signals over time and rolls up into four scores: Hot (current activity), New (first-detection recency), Rising (acceleration vs. baseline), and Overall. Tranco rank is a heavy input on Overall (~50%), with outbound clicks (~25%) and freshness signals (~15%) adding to it. All weights decay with time.

The default feed mixes Hot, Rising, and New entries and de-duplicates by domain so a single site never dominates the page.

The visit number on each card is an estimate, not measured traffic. It's derived from the site's Tranco rank using a power-law approximation calibrated to publicly known traffic data. Treat it as an order-of-magnitude signal, not a precise figure.

A blocklist removes adult, gambling, and link-shortener domains. A quality and spam score filters out parked pages and obvious low-content junk. A health checker only marks a site offline after ten consecutive failed probes — short outages are never enough to hide a listing.

No user accounts, profiles, votes, comments, follows, or social features. No fake engagement counts. No paid placement. No AI-generated reviews or editorial summaries — descriptions come from each site's own public metadata.

If you operate a site listed here and would like it removed or corrected, see the removal page.