Public blog pages (/{username}/{slug}) now render statically and serve from cache, re-validating at most once a day or immediately when a post changes. Previously, every visit triggered fresh server rendering — around 12 database queries on the rich page — because the page read request cookies, which silently disabled caching. A scraper hitting every brand's posts could exhaust the connection pool. Private blogs remain protected, and draft preview in the editor works as before.
A separate scan bug meant clicking "Run scan" on one brand could silently overwrite scan trigger data across every brand in the database with a stale claim older than ten minutes. That's fixed. Scans now correctly scope to the brand you triggered them on.
Three new pages are live in the developer docs: a Content Workflows guide walking through the full in-app content pipeline, an AI Visibility Score guide explaining what the 0–100 score measures and how scans are built and run, and a subdomain hosting guide for serving your agent site on a branded subdomain with a CNAME. The docs site itself is now called Pendium Docs — not "Pendium MCP Docs" — with a reorganized sidebar and a landing page that reflects the full scope of what's documented.
The copy describing full scan mode as using premium models for higher accuracy has been corrected. Both batch and full use the same models. The actual difference is query volume — raise maxQueries for broader coverage.