Internal acceptance evidence

These are local fixture measurements. No synthetic production events, provider calls, messages, paid actions or live widget activation were used.

The setup clocks cover prepared automation: baseline extraction, applying an approved patch, local package install, real MCP handler calls, generated helpers, isolated host SQL and actual Crust receipts. They exclude prior development and external client/credential setup. No three-external-developer or unassisted setup timing claim is made.

The optional engagement helper measured 2,274 bytes gzip. One paired 1280×720 local browser observation recorded LCP 104/48 ms, CLS 0/0, and provisional INP candidate 24/24 ms before/after. Event Timing threshold was 16 ms; this is not field INP or evidence of a causal improvement. A separate 359×780 mobile check observed LCP40 ms, CLS0.004995 and INP candidate32 ms. Actual expiry renewal, idempotent response retry, opt-out and keyboard dismissal were verified through the production helper and SQL services with fixture host authentication.

Live Shopify storefront Web Vitals, natural Shopify Dashboard observations, 95% eligible-install coverage, mature health outcomes, commercial gates and genuine Mantle imports remain separate acceptance work. Hosted backend and Shopify extension releases remain distinct.