Every number here is a request count. It is not an index entry, not an impression and not a citation. A crawler fetching a page has learned the page exists; whether the engine makes it retrievable is a separate decision it does not disclose, and whether anyone ever asks a question it can answer is a third thing again.

These are leading indicators on purpose. The citation gates arrive weeks from now; coverage and first-visit dates move today, so they are what tells us whether anything is happening at all.

Only clients whose User-Agent identifies a known crawler count here. Coverage counts paths that exist as published pages. Reaching a page is not indexing it, and indexing it is not citing it.

This table exists because the panel's first version reported 61.4% coverage from one "unnamed datacenter" client, which read as traction. It was four clients and none was an engine: a scraper identifying as Chrome 42 (released 2015), a WordPress vulnerability scanner asking for software we do not run, a credential scanner asking for .env, and a fleet drawing one URL from rotating IPs. Counting those as reach would have turned automated noise into apparent demand.

Organic excludes our own probes and the operator's own connection. The probe count is traffic this project generated deliberately to prove the instrument works; it is kept visible rather than deleted, because a number that can be inflated by its own author is the most dangerous kind.

blocked means the engine refused to answer (an anonymous rate limit), which is not evidence either way and is never counted as "not cited".