Route Android uploads to an internal draft and bind iOS submissions to the verified App Store Connect app. Document explicit-path local artifact submission and preserve separate release authorization.
Add a conditional Home cue for partner-answered questions and route both the card and catch-up reminders into the pending answer flow. Track prompt engagement and cover hidden, visible, CTA, and notification behavior.
Add a guarded mobile release skill that synchronizes public versions, writes on-brand notes, verifies signed iOS and Android artifacts, and tags the verified release commit.
Measure untouched mobile source, add regression coverage for native media paths and account operations, and enforce ratcheting API and mobile thresholds in required CI checks.
Share one configured PostgreSQL pool across worker repositories and make API and worker connection budgets explicit.
Add privacy-safe aggregate request, query, pool, sync, and outbox telemetry, a loopback-only representative load fixture, bounded worker shutdown, and focused regression and integration coverage.
Document the database-wide budget formula and defer pg_stat_statements activation to a reviewed migration and planned restart.
Share bounded database pools across API repositories, consolidate sync and chat queries, and make mobile refreshes cache-aware and foreground-aware.
Add focused regression coverage, client freshness documentation, and repository agent role configuration.
Add a repository-owned Codex skill and helper for bounded Claude Code execution, model and effort selection, resumable follow-ups, and independent acceptance checks.
Combine the Android API 36 release, dares loop, RevenueCat documentation correction, and daily engagement features. Resolve migration ordering and preserve both dares expiry and streak reminder operations.
Use reported user timezones for daily response limits, expose couple streak state and reminders, and improve swipe and live-drawing synchronization with focused tests and documentation.
Completes the half-built dares feature: send a dare, respond, report done,
confirm, and get a push notification on every transition.
Backend (apps/api):
- New `dares` domain with couple-shared premium entitlement, a free weekly
send allowance, and server-enforced actor/state rules on every transition.
- `submitted` status so the recipient reports completion and the sender
confirms, rather than the sender guessing.
- Notifications ride the existing operations outbox via a trigger on
`sent_dares`. Push copy is generic; store compliance forbids dare text in a
preview and a test asserts no payload leaks it.
- Worker expiry sweep now covers `pending` as well as `active`/`submitted`;
an invitation nobody opened previously stayed open forever.
Fixes two latent defects:
- Deleting any dare that had been sent aborted with a check violation, because
`dare_id` is ON DELETE SET NULL while `sent_dares_dare_or_custom_check`
required it. Dare text is now snapshotted at send time, which also keeps a
couple's history stable across catalogue edits.
- The free send allowance was raceable: concurrent requests all read the same
count and every one succeeded. The count and insert are now serialized on
the sender's profile row.
Also ports `content_status` review metadata to the API catalogue, which had no
store-safety gating at all, and mirrors the review decision already enforced on
the legacy catalogue.
Mobile: dares screen (active/send/history), send sheet with duration presets,
home tile badging incoming dares, paywall on both proactive and 402 paths.
Verified: lint, typecheck, 379 unit tests, build, 71 integration tests against
local PostgreSQL.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Upgrade the mobile runtime and tracked Android project to compile and target API 36. Align Expo modules, React Native, Node, Gradle, and Android build tooling while preserving the custom widget sources.\n\nAdapt file-system, notification, animation, test, and Metro compatibility for the SDK 54 runtime.
The Platform Support section claimed Android was "not yet implemented".
Android has full native SDK support: getApiKey() selects the Android key,
com.android.vending.BILLING is declared in app.json, the paywall has no
platform gating, and the webhook passes event.store through unchanged so
PLAY_STORE events need no separate handling.
Also flag the entitlement ID mismatch between the "pro" fallback in
revenuecat.ts and the "Sauci Pro" value set in eas.json. A build profile
missing that variable yields isProUser: false with no error, which
presents as churn rather than as misconfiguration.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Promote the reproduced release lessons into harness enforcement: keep tracked native regeneration separate from artifact builds, separate Android build and upload authorization, use stable login readiness in Maestro, and retain the iOS 43 and Android 46 version increments.
Replace the Supabase application data plane with the Node/PostgreSQL API, migrate mobile, admin, web, and MCP callers, add controlled cutover tooling, and define hardened Dokploy infrastructure.
Remove client-side adult-content preferences and badges so the centrally reviewed catalogue is the single visibility boundary.\n\nUse neutral relationship-focused copy across onboarding, discovery, paywall, and login surfaces, and retain biometric controls under App Lock for both native platforms.
Classify every environment through the immutable allowlist and archive unknown concurrent additions. Remove the remote-project dependency while retaining fail-closed postconditions.
Keep exact audited-count assertions for the production project while allowing local and non-production fixture catalogues to be classified by the same immutable allowlist.
Hide archived and unreviewed catalogue content from existing clients through RLS and guarded recommendation RPCs. Reject stale writes, preserve historical rows, and seed reviewed defaults when the cutover removes a couple's full active selection.
Add reversible review states, audit history, and super-admin controls for catalogue entities without changing current client visibility.
Stage 30 high-risk production packs for archival review, cap generated content at non-explicit language, and preserve a checksummed pre-change production catalogue snapshot.
jjbrunton
