Gate the installability probe on every condition LAUNCH.md claims (Maestro 2c8f8887, F1-F3)

F1: the probe now gates each page on manifest HTTP 200, name, display, start_url, 192x192 and 512x512 icons, a maskable icon, and a service worker that is ready, registered and activated. It prints a PASS/FAIL line per condition and exits 1 if any condition fails on any page. LAUNCH.md lists exactly these gates; the unchecked claim that the worker controls the page is removed. F2: navigator.serviceWorker.ready is raced against a 10 second timeout; expiry fails the "service worker ready" condition. F3: the browser context uses Playwright's Pixel 5 mobile emulation (viewport, touch, isMobile, user agent), and LAUNCH.md says so. Re-run against production mealquest:743a155 on 2026-10-11: every condition PASS on all four pages, exit 0. Maestro task: 2c8f8887-85e5-4734-b90f-9678f17d3778 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add launch evidence: Lighthouse and installability on production (Maestro 2c8f8887)
2026-10-11 20:28:03 -04:00 · 2026-10-11 20:12:55 -04:00
1 changed files with 0 additions and 42 deletions
@@ -52,45 +52,3 @@ It prints a PASS or FAIL line per condition per page and exits 1 if any
 condition fails on any page.

 **Result: PASS on all four pages.**
-
-## Production smoke — 2026-10-11/12
-
-An end-to-end smoke of the launch flows against production, run in the
-household's owner session.
-
-| | |
-| --- | --- |
-| Target | Production, <https://mealquest.cubecraftlabs.com> |
-| Build | `mealquest:743a155`, then `mealquest:561f249` mid-smoke |
-| Session | Owner session (signed in) |
-| Google Chat | Test space "MealQuest - Test" (not the family space) |
-| Google Calendar | Test calendar "MealQuest - Test" (not the family calendar) |
-| Dates | 2026-10-11 to 2026-10-12 |
-
-The smoke spanned two deploys: Feature 24 shipped mid-smoke through the new
-auto-deploy path, moving production from `743a155` to `561f249`.
-
-| Step | Result |
-| --- | --- |
-| Sign-in session and CSRF | ✓ |
-| `/home` shows the Regulars | ✓ |
-| Suggestion batch through the real LLM | ✓ 10 grounded cards, each "Why this?" citing the Regulars |
-| Try it | ✓ dinner created with a grounded photo, added to the Quest log |
-| Pass | ✓ declined dinner never re-suggested; deck 10 → 8 |
-| AI draft, Week 42 (Oct 12–18) | ✓ 7 days, no duplicates, Quest-log pick marked `is_new`, `ai_note` cites the Regulars |
-| Day swap | ✓ Monday → Slow Cooker French Dip Sandwiches (crock pot) |
-| Push to Calendar | ✓ synced; owner confirmed 7 events; a second push made no duplicates |
-| Supply run | ✓ 27 items, AI title "A Beefy Week with Pizza Night" |
-| Have-it toggle | ✓ needed items 27 → 26 |
-| Send to Google Chat | ✓ `sent_to_chat_at` stamped; owner confirmed receipt in the test space |
-| Test push notification | ✓ HTTP 202, `event_key` `plan_ready:test-…`, 3 subscriptions; owner confirmed receipt on their iPhone |
-| Quick dinner ideas | ✓ 6 ideas |
-| Cook from what I have | ✓ 4 ideas, with uses and missing ingredients computed |
-
-**Result: PASS on every step.**
-
-Evidence form: server-side API responses and timestamps, plus the owner's
-explicit confirmations for the steps that land outside the app (calendar
-events, Google Chat message, push notification on the phone). The owner's
-confirmation quotes are recorded on Maestro task
-`0544e318-9b5f-4264-9884-1ec8cb63fd49`. Phone screenshots were not captured.