[sushilm20]

Docs only. This records what changed in production on 2026-10-28.

Catalogs. goBILDA, ServoCity, AndyMark, and Pitsco were re-crawled locally at a slow pace (500 ms between pages, 1 s between Shopify feed pages, honouring robots.txt and Retry-After) and published through catalog-load.ts with --expect HORS SKUs and --spot 20. Every spot check matched 20/20.

Vendor	SKUs	Published (UTC)
goBILDA	1,991	16:13
ServoCity	4,318 (4,287 verified packs)	21:33
AndyMark	5,279 (was 5,313)	16:00
Pitsco	1,625	15:59

Each previous generation is kept for a day of rollback. REV was not crawled.

REV cron (PR 55). The cron ran two full cycles in production. Cloudflare's CPU per run:

• download: 5 ms;
• row steps: 8, 7, 9, 23, 11, 7, and 8 ms;
• final staging step: 14 ms;
• idle run: 2 ms.

Every traced run ended ok, but three went above 10 ms. Smaller steps barely help, because most of a step's cost is fixed per invocation, so the docs record this margin instead of claiming the steps fit.

[greptile-apps]

Confidence Score: 5/5

[Low impact] This documentation-only PR appears safe to merge.

Summary

Records the 2026-10-28 catalog refresh and REV's production CPU measurements.

• The catalog guide now shows the October refresh results.
• REV’s guide now reports the cron’s measured production CPU.

No actionable issues were found.

_{Reviews (1) · Last reviewed commit: "Record the 2026-10-28 catalog refresh an..." · Reviewed by Greptile}