Room workers use sealed, versioned selection catalogues. The larger Standard research working database is separate; background collection does not directly edit a room's catalogue.

The operator-approved Standard refresher processes records older than 30 days in oldest-first order, at five HTTP starts per minute including authentication and retries. It refreshes profiles, complete top-score snapshots and map metadata in the isolated research database. The older expansion collector remains disabled. This is background maintenance, not a player command or a promise that all records become fresh together.

Replacing a live catalogue requires analysis, freezing the Standard/DT pair, checking calculator/source identity, exclusions and room eligibility, then a validated deployment. Room workers stay pinned to their own catalogue until updated; research progress alone does not change available maps.

From v6.7.30, a DT catalogue older than 30 days is due for refresh, not automatically disabled solely because of that age. Invalid calculation dates, mismatched source identity/calculator and uncertain per-chart checksums or revisions still reject unsafe DT variants. A changed chart must not retain an outdated DT rating merely because the catalogue stays available.

This rule is distinct from configured map-age filters, ranked-date filters and the research freshness threshold. It does not broaden stars, playable duration or source/style eligibility. NoMod fallback remains available.

The operator exclusion list still blocks listed exact charts even if a rebuilt catalogue accidentally contains them. Export-time exclusion removal/validation is a separate obligation; the legacy research export is not yet automatically wired to this runtime list. Removing a runtime exclusion cannot restore a row absent from the live catalogue.

Mixed DT · Map selection · Catalogue deployment · Release history

Independent operators download the complete prepared exports and companions from the owner's verified HTTPS origin. They do not run their own research collector or recreate the large analysis databases. Earlier capped inspection counts were reporting limits, not missing catalog rows. See full selection catalogs and research databases for verified example counts and size differences. Catalog activation/trust changes remain distinct from a compatible worker code update.