Make setup work all the way through

Monday’s work closed several gaps between having a feature and being able to use it: executable documentation, a database with vector support, blog connection instructions and working billing configuration. Some changes reached production; others remain local or still need a complete customer journey to prove them.

• db3 · examples that can be executed. HTML and Markdown documentation now share instructional content across 90 pages, and eight previously hidden examples appear on the local website. The feature-app guide covers definition, installation, startup, cleanup, authentication and testing. Checks run its published examples in a clean generated app. Seventy-three documentation tests, 830 framework tests, consumer checks and a production Docker build passed; eight Redis tests were skipped. This batch was not deployed, and the wider editorial rewrite remains unfinished.
• db3 · a verified local database path. The starter Docker image moved to MariaDB 11.8. Eight creator tests and real vector storage/search checks passed, including persistence across restart. The product plan separates account-free local Docker setup from an optional account-backed Cloud database. The guided interface and Cloud provisioning remain proposals. A broader packed-app check stopped during dependency installation with an npm internal error, before runtime verification.
• GrowthScout · hosted blogs and clearer DNS targets. Basic blog hosting was deployed with saved brand colours, fonts and logo, manual DNS instructions and a public setup guide. A later deployed fix moved new connection targets to the separate host.growthscout.io namespace, allowing GrowthScout’s own blog address through the normal setup flow. The pending setup was migrated before the old wildcard was removed. Focused tests and production health checks passed; a real customer-domain certificate and first publication remained the end-to-end proof still to complete.
• GrowthScout · live billing configuration. Production switched to the live Stripe account with its live prices, webhook and billing portal configuration. Live Checkout and portal sessions, webhook delivery and production health were verified without taking a payment. Existing application records were preserved, but sandbox-backed subscriptions still require reconnection before live billing operations work. Tax calculation and a completed real-payment journey remained unverified; this was a configuration cutover, not an application-code release.
• GrowthScout · public legal-page corrections. Updated legal contact details and deletion disclosures were committed, pushed and deployed. Fourteen tests, type checks, the production build and live health checks passed. The data-processing agreement and liability amount remained unfinished, so the deployment does not establish that all legal work is complete.
• DOM Studio · repair existing licence databases. A missing-column repair was deployed for existing billing databases. An authorised reconciliation then restored licence and source access for a verified purchase while preserving its original purchase date and receipt. Production access was checked. This was a concrete schema and entitlement repair; it does not imply that every possible billing failure was tested.