Summary: Why 7Mind moves its app to React Native, the current direction (a sibling app apps/sevenmind inside pulse-app, after Ali Kayhan’s spikes of both repo layouts), the ground rules for 7Mind work in that repo, and why the tenant model of pulse-app does not fit 7Mind.
Sources: kickoff meeting 2026-11-09 (Ali Kayhan, Matias Olocco, Ahmed Bayoumy, Nikolas Haag); Ali Kayhan’s spike report in Slack C0CFDJ4CM7F, 2026-11-10 (thread 1794312214.639879); Gymondo-git/pulse-app PR #264 “[WIP] Add the 7Mind app as a sibling app” and its apps/sevenmind/README.md; sevenmind/ripple-app branch spike/rn-migration-auth-onboarding-home-profile; code inspection of both apps (2026-11-06); local tenant spike (2026-11-06); PR #264 re-read on 2026-11-13
Last updated: 2026-11-13
Goal
Put the 7Mind app on the same stack as the Gymondo app, pulse-app, so that more developers across 7NXT can contribute to it. Since the 2026-11-10 spike report the goal is phrased as merging 7Mind into the 7NXT apps rather than only migrating it: one React Native, Expo and dependency set for all apps, consistency forced by sharing a repo (not shared features or UI, at this point).
Current direction (2026-11-10): sibling app in pulse-app
Not formally confirmed yet. At the kickoff (2026-11-09) the architecture was deferred. Ali Kayhan then built the same scope twice and recommended the sibling app; Matias Olocco acknowledged it (“very good, thanks for the clarity on this”). The 2026-11-06 decision for a separate ripple-app repo is superseded by this direction, see rn-migration-decision-log.
The two spikes, each covering launch, welcome, sign-up and login (email, Google, Facebook, Apple, magic login link), password-reset request, session restore, onboarding, a “For you” tab and a Profile tab with stats and weekly streak:
| Spike | Where | Ali’s verdict |
|---|---|---|
| a) Sibling app | Gymondo-git/pulse-app, PR #264 (draft, branch feat/add-sevenmind-sibling-app, stacked on #253, the Expo 58 upgrade) | Chosen. The 7Mind code is so isolated that it behaves like a standalone repo, and it keeps one version of React Native, Expo and every dependency for all apps |
| b) Standalone repo | sevenmind/ripple-app (internal, created 2026-11-06), branch spike/rn-migration-auth-onboarding-home-profile | Easier for humans to read and more independent, but shared code would need separate versions of React Native and every dependency |
Kickoff arguments the spikes confirmed: keeping the app in pulse-app simplifies maintaining shared packages and dependencies; a separate repo adds overhead for sharing core packages and API configuration; the monorepo already supports several apps on shared packages (Ahmed Bayoumy).
Reuse is close to zero today. The point is that it can grow without a version-management problem, and that sharing a repo pushes both apps towards the same solutions.
Layout in pulse-app (from PR #264)
apps/sevenmindnext toapps/native(Gymondo, Buddyfit) andapps/tv. Metro on port 8083 (native 8081, TV 8082).- 7Mind-only package code lives in each package’s
src/sevenmind, exported as a<package>/sevenmindsubpath (for example 7Mind theme tokens). tooling/eslint.product-boundaries.jsforbids Gymondo, Buddyfit and TV code from importing anysevenmindsubpath, so they cannot use 7Mind colour tokens by accident.- Config generation needs
APP=sevenmindnext toTENANT=sevenmindandAPP_ENV; fitness generation never touchesapps/sevenmind. - Identity: the Flutter app’s bundle ids, Firebase projects and Facebook apps (staging
de.7mind.ios.stg/de.seven_mind.android.stg, prodde.7mind.iphone/de.sevenmind.android), display name7Mind [STG]on staging, staging icons with a REACT ribbon. Installing it replaces a Flutter install of the same flavor. - Backend: Gymondo SSO host for login, social token exchange and password reset (
app.6mind.de/app.7mind.de); the legacy gateway for profile (magic.6mind.de/magic.7mind.de). - Not there yet: no Codemagic workflow and no Android upload key (release builds use the debug keystore, so they cannot update an installed Flutter app); Braze, Adjust, PubNub and Thryve keys are not read; no tracking or crash reporting. Status per step in rn-migration-plan.
Ground rules for 7Mind work in pulse-app
Proposed by Ali Kayhan (Slack, 2026-11-10); contribution and approval rules still to be agreed:
- 7Mind code goes only in
apps/sevenmindor a package’ssrc/sevenmind. - Never edit Gymondo, Buddyfit or TV files for 7Mind work.
- In shared packages, only add things. Never change what already exists.
- Gymondo, Buddyfit and TV cannot use 7Mind-specific things in a package’s
src/sevenmind. - 7Mind libraries go into
apps/sevenmindfirst, then get evaluated for a package’ssrc/sevenmindor the shared code. - Do not change library versions, patches or Expo / React Native in 7Mind PRs unless all apps benefit.
With these rules Ali suggested 7Mind PRs could be merged without approval from the Gymondo team.
Constraints from the kickoff (2026-11-09)
- Minimum OS: iOS 17 and Android API 28 (Flutter app today: iOS 15, minSdk 26). iOS 15 and 16 are about 6% of users; API 28 and higher covers about 97% of the Android market.
- Backend unchanged at first, to keep regression risk low: the app calls the same endpoints as the Flutter app.
- Analytics and attribution stay as they are, because company services depend on them (practice history to the statistics service, Braze campaigns, Adjust attribution).
- Retire, do not port: unused feature flags, the legacy DACH token upgrade, the ZPP flavor (rn-migration-not-migrated).
- Health after the migration: mindfulness minutes move to Thryve (pulse-app’s wearables vendor, staging and production accounts exist), covering Apple Health and Android, instead of porting the Flutter
healthintegration.
Why not a tenant of apps/native
pulse-app’s model is “a new brand is config, tokens and native service files, never component changes” (AGENTS.md). That works for Gymondo and Buddyfit because they are the same product. A local spike on 2026-11-06 ran 7Mind as a third tenant of apps/native: branding, consent, Braze, the store and login with a 7Mind staging account worked, but everything after login is the fitness product.
- The tab bar is hardcoded to Home, Fitness, Progress, Challenges, Nutrition; the main stack registers 70 screens, almost all fitness, gated only by two feature flags.
- About 85% of the screen code in
apps/native/src/screens(about 47k lines) is fitness-specific. - The player is a workout video player; 7Mind needs an audio-first meditation player.
A sibling app keeps 7Mind’s own tabs and screens while sharing the package layer.
What “same architecture” means
- Adopt, do not choose. Library choices follow pulse-app; every deviation needs a reason in rn-migration-decision-log. Mapping in rn-migration-dependency-mapping.
- Same folder rules. Screens, navigation and journeys in
apps/sevenmind; packages stay presentation-agnostic and expose barrels (plus thesevenmindsubpath). - Same build axes.
APP_ENV=local,test,prod; no ZPP flavor. - New capabilities as packages. The audio player, the Flutter data import and the 7Mind API modules start in
apps/sevenmindand move tosrc/sevenmindor shared code once they prove generic.
Product shape
7Mind is audio-first: meditations, courses, sleep, breathing, prevention courses (ZPP, German only), Aury, SOS toolkit. Its tabs and screens are its own; only the shell (auth, account, paywall, consent, settings, downloads infrastructure, tracking, push) matches pulse-app.
History of the repo question
- 2026-07-06: pulse-app starts as “7NXT App” with 7Mind listed as a future white-label brand.
- 2026-11-06: first recommendation 7Mind as a tenant; after the tenant spike, a sibling app; then the stakeholders chose a separate repo,
ripple-app. - 2026-11-09: kickoff discusses both layouts and defers the decision.
- 2026-11-10: Ali’s two spikes; recommendation and current direction: sibling app in pulse-app.
If the separate repo comes back, the open problem is importing pulse-app’s packages, which are workspace-only (workspace:*) and not published: options were a private registry, a git dependency, an extracted shared repo, or copying.