Anonymous-user diagnostic — funnel, retention, conversion

Five views on anonymous-user behaviour, bundled into one response.

Designed to answer the standing product question: "what do free users actually do, and do any of them sign up?" Each view is one query; none of them mutate; all of them tolerate a partly-populated audit table so the page renders on a fresh instance.

Views: • funnel — 30-day check-packages calls per anon IP, bucketed • retention — weekly cohort × week-offset heatmap (last 90d) • conversion — IPs that hit ≥9 anon calls (near the 10 wall), with whether the IP later got bound to an account (= signed up) • by_ecosystem — anon vs auth distribution of queried ecosystems • ci_suspect — anon IPs making ≥5 check-packages calls in a single hour (CI/automation signal)