Summary: Decisions taken and questions still open for the 7Mind React Native migration, newest first. Append, do not rewrite past entries; mark a superseded decision instead. Sources: conversations and code inspection, cited per entry; Google sign-in visibility re-checked in sso_buttons.dart, sso_cubit.dart and authentication modules at app 4.0.9 on 2026-11-09; kickoff meeting 2026-11-09 and Ali Kayhan’s spike report (Slack, 2026-11-10); pulse-app PRs #264 and #280, Gymondo-git/function-sso and Gymondo-git/terraform workspaces/apps (read 2026-11-13) Last updated: 2026-11-19


Decisions

Format: YYYY-MM-DD -- decision. Why. Who. Source.

  • 2026-11-12 — Google sign-in on Android staging uses the web OAuth client of mind-flutter-stg, the staging Firebase project. The staging signing keys (Play app signing, upload key, the Expo template debug key the new app signs with) are registered on its Firebase Android app for de.seven_mind.android.stg; the new app and the Flutter staging flavor read the web client from google-services.json. Not the 7Mind web SSO client in mind-f62c0: that project’s consent screen is internal, so only @7mind.de Google accounts could sign in. Why: Google sign-in on Android staging never worked (no OAuth client in the staging config), and this keeps staging in the staging project for any Google account. Who: Nikolas Haag. Source: pulse-app PR #280; the Gymondo side is an open question below.
  • 2026-11-10 — Direction: 7Mind becomes a sibling app, apps/sevenmind, inside pulse-app, with 7Mind-only package code in each package’s src/sevenmind and a lint rule that keeps the fitness apps out of it. Supersedes the 2026-11-06 separate-repo decision; not formally confirmed yet. Why: Ali Kayhan built the same scope in both layouts; the code looked almost the same, the sibling app is just as isolated, and it keeps one React Native, Expo and dependency set for all apps (“merging, not just migrating”). Who: Ali Kayhan recommended, Matias Olocco acknowledged. Source: Slack C0CFDJ4CM7F thread 1794312214.639879, pulse-app PR #264, rn-migration-goals-and-architecture.
  • 2026-11-09 — Minimum OS for the new app: iOS 17, Android API 28 (Flutter app: iOS 15, minSdk 26). Why: iOS 15 and 16 are about 6% of users; API 28+ is about 97% of the Android market; matches pulse-app. Who: proposed by Ali Kayhan at the kickoff. Source: kickoff notes.
  • 2026-11-09 — Backend unchanged at first: the new app calls the same endpoints as the Flutter app (Gymondo SSO for auth, legacy gateway for 7Mind domains). Why: minimises regression risk during the migration. Who: kickoff. Source: kickoff notes.
  • 2026-11-09 — Analytics and attribution stay as they are (Rudderstack, Braze events, Adjust). Why: core company services depend on them, including practice history in the statistics service. Who: kickoff. Source: kickoff notes.
  • 2026-11-09 — Retire instead of porting: unused feature flags, the legacy DACH token upgrade (UpgradeDachToken), the ZPP flavor. Confirms the 2026-11-06 ZPP decision. Who: kickoff. Source: kickoff notes, rn-migration-not-migrated.
  • 2026-11-09 — Mindfulness minutes go through Thryve after the migration (staging and production Thryve accounts exist), covering Apple Health and Android; the Flutter health integration is not ported. Who: kickoff. Source: kickoff notes.
  • 2026-11-06 — ZPP flavor and the ZPP submission builds are not carried over; ripple-app has one test (staging) and one prod environment. Why: the ZPP flag is not read anywhere on origin/main, so the flavor only changes app name and icon. Who: Nikolas Haag (prevention owner to confirm reviewer access). Source: conversation, rn-migration-not-migrated.
  • 2026-11-06 — ripple-app follows pulse-app’s library choices; where 7Mind uses something else today, it switches to the pulse-app choice (example: Sentry to Firebase Crashlytics). Mapping in rn-migration-dependency-mapping. Who: Nikolas Haag. Source: conversation.
  • 2026-11-06 — (Superseded 2026-11-10 by the sibling-app direction.) The new app lives in a separate repository named ripple-app, not inside pulse-app. It follows pulse-app’s architecture and reuses pulse-app code by import; the import mechanism is still open. Supersedes the earlier recommendation to make 7Mind a pulse-app tenant or a second app in pulse-app. Why: decision by the stakeholders; the spike also showed the tenant model only fits brands of the same product. Who: stakeholders, relayed by Nikolas Haag. Source: conversation, rn-migration-goals-and-architecture.
  • 2026-11-06 — The 7Mind app moves from Flutter to React Native. Why: same stack as the Gymondo app (pulse-app) so more developers can contribute. Who: 7Mind product engineering (Nikolas Haag). Source: conversation.

Open questions

QuestionOwnerStatus
Target optionstakeholdersdirection 2026-11-10: sibling app in pulse-app (was: separate repo ripple-app, 2026-11-06); formal confirmation pending
Contribution and review rules for 7Mind work in pulse-app (Ali’s ground rules; can 7Mind PRs merge without Gymondo approval?)7Mind + pulse-app ownersopen. PR #264 itself also changes existing shared and Gymondo code (ui Typography, the api client and interceptors, config types, the Gymondo AccountScreen, the TV theme), so it needs Gymondo review either way
How code is imported from a separate repo7Mind + pulse-app ownersmoot while the sibling-app direction holds
Which 7Mind API domains move to Gymondo APIs, which stay on the legacy gateway for now?backend ownersdecided for the migration 2026-11-09: backend unchanged, same endpoints as the Flutter app; moving domains is later work
Audio library for meditationsspikeopen; spike 2026-11-16 recommends expo-audio, fallback react-native-video, see rn-migration-audio-player
Sentry or Crashlytics?7Minddecided 2026-11-06: Crashlytics, like pulse-app (recorded as an example of the 2026-11-06 rule, not yet confirmed with the 7Mind mobile lead); tracking spike 2026-11-16 recommends Crashlytics via @pulseapp/observability, see rn-migration-crash-reporting
Rudderstack: keep a sink, or record practice sessions through an API outbox like pulse-app?7Mind + data + backenddecided 2026-11-09: analytics and attribution stay as they are, so Rudderstack stays
Feature flags: Firebase Remote Config like pulse-app, or Braze flags as in MIND-880?7Minddecided 2026-11-06: Firebase Remote Config; Braze for audience targeting only
Search: Algolia or the Gymondo search API?7Mind + backendopen
Is Facebook login still needed? If yes, SDK or pulse-app’s web social login?productopen; PR #264 already has Facebook login built with the Facebook SDK (react-native-fbsdk-next)
Does current Google sign-in have a storefront-country gate?mobile engineeringresolved 2026-11-09 by source inspection at 4.0.9: no gate in SSO UI; storefront country in auth is analytics-only. See rn-migration-not-migrated
Apple Health mindful minutes: still wanted?productdecided 2026-11-09: via Thryve after the migration, Apple and Android
Which flags are unused, so their features need not be migrated?7Mindretiring them agreed 2026-11-09; the per-flag list is still open
Does the rewrite affect the ZPP certification, and how does a ZPP reviewer get unlocked prevention content without a ZPP build?prevention owneropen
Feature freeze policy for the Flutter app during the migrationproductopen
What do users on iOS 15 and 16 (about 6%) get: stay on the last Flutter version, any notice?productopen
Who adds the staging Android Google client to the Gymondo SSO? Its Google sign-in Lambda (function-sso) accepts per vendor only the client ids in siwg_configuration of the AWS secret sso-cognito-service-test; until sevenmind → android_client_id is the mind-flutter-stg web client, the token exchange answers 500 “audience provided does not match aud claim in the JWT”Gymondo platformopen (2026-11-13)