# Domain Breakdown

*Updated: 2026-04-05*

The Angels platform decomposes into 7 bounded domains. Each domain owns its data, API surface, and business rules. Domains communicate through well-defined contracts.

---

## Domain Map

```
┌─────────────────────────────────────────────────────────────┐
│                        ANGELS PLATFORM                       │
├──────────┬──────────┬──────────┬──────────┬────────────────┤
│ Identity │ Wellbeing│  Safety  │  Circle  │   Consent      │
│          │          │          │          │                │
│ Auth     │ Check-ins│ Plans    │ Angels   │ Sharing rules  │
│ Personas │ Journals │ Crisis   │ Invites  │ Privacy guards │
│ Profiles │ Streaks  │ Recovery │ Nudges   │ Coercion det.  │
│ Settings │ Trends   │          │          │                │
├──────────┴──────────┴──────────┴──────────┴────────────────┤
│              Insights (cross-domain reads)                  │
├────────────────────────────────────────────────────────────┤
│              Notifications (cross-domain events)            │
└────────────────────────────────────────────────────────────┘
```

---

## 1. Identity

**Owns:** Users, personas, profiles, authentication, settings.

**Core entities:**
- `User` — account with email, auth method, created_at
- `PersonaProfile` — one per user, type: survivor | guardian | recoverer | connector
- `Settings` — notification prefs, quiet hours, data export config

**Key rules:**
- A user has exactly one persona (chosen during onboarding, changeable later)
- Persona determines which onboarding flow runs
- Settings are private to the user, never shared

**Journeys served:** persona-selection, all onboarding flows, settings

---

## 2. Wellbeing

**Owns:** Check-ins, mood data, journal entries, streaks, trends.

**Core entities:**
- `CheckIn` — mood (great|good|mixed|struggling), timestamp, optional journal
- `JournalEntry` — freetext tied to a check-in, always private unless explicitly shared
- `Streak` — current consecutive days, longest streak, last check-in date
- `MoodTrend` — computed 7/30-day aggregates

**Key rules:**
- One check-in per day (re-check-in overwrites, doesn't duplicate)
- Journal entries are NEVER shared by default — require explicit consent per-angel
- Streaks reset after 48 hours without check-in (not 24 — timezone grace)
- Mood trend computation is read-only derived data

**Journeys served:** daily-checkin, home-navigation (check-in state, nudge)

---

## 3. Safety

**Owns:** Safety plans, crisis events, post-crisis recovery.

**Core entities:**
- `SafetyPlan` — triggers[], copingStrategies[], emergencyContacts[], versioned
- `CrisisEvent` — activated_at, resolved_at, type, false_alarm flag
- `RecoveryCheckpoint` — day 1/3/7/14 post-crisis check-in with notes

**Key rules:**
- Safety plan is versioned — edits create new versions, old versions preserved
- Crisis activation sends alerts to ALL connected angels (no consent check — crisis overrides consent)
- Post-crisis recovery is a scheduled sequence (system events at day 1, 3, 7, 14)
- Emotional readiness check gates trigger editing (REQ-001)

**Journeys served:** safety-plan, crisis-flow, survivor-onboarding (initial plan creation)

---

## 4. Circle

**Owns:** Angel connections, invitations, relationship management.

**Core entities:**
- `Circle` — a user's set of angel connections
- `AngelConnection` — link between a user and an angel, with role and status
- `Invitation` — pending/accepted/declined/deferred, with invite link and metadata
- `Nudge` — system-generated prompt to check on someone

**Key rules:**
- Free tier: 1 angel connection max. Premium: unlimited.
- Invitations expire after 30 days
- Angels can set boundaries (notification level, quiet hours, health metrics visibility)
- Nudges are suggestions, never obligations — language is "thinking of you", not "check on them"

**Journeys served:** angel-invitation, guardian-onboarding, survivor-onboarding (invite step), home-navigation (nudge)

---

## 5. Consent

**Owns:** Sharing rules, privacy safeguards, coercion detection.

**Core entities:**
- `ConsentRule` — metric + person + visibility level (hidden|summary|detailed)
- `SharingProfile` — aggregated view of what each angel can see
- `CoercionFlag` — system-detected pattern suggesting user is being pressured
- `QuietExit` — record that user exited a sensitive flow without completing it

**Key rules:**
- Default sharing: NOTHING shared. Users opt IN, never opt OUT.
- Consent can be withdrawn at any time, with immediate effect
- Coercion detection triggers if: (a) consent rules change rapidly, (b) all metrics set to "detailed" for one person, (c) consent changed immediately after an angel message
- Quiet exit is always available — no confirmation dialogs when leaving consent flows

**Journeys served:** consent-wizard, privacy-response

---

## 6. Insights

**Owns:** Cross-domain read models, pattern detection, trend visualisation.

**Core entities:**
- `MoodPattern` — detected trend (improving, declining, stable, volatile)
- `CorrelationInsight` — relationship between mood and external factors (e.g., "mood dips on Mondays")
- `StreakInsight` — contextual reflection based on streak data + mood history
- `CareProviderView` — aggregated, anonymised data for therapist dashboard

**Key rules:**
- Read-only domain — never writes to other domains' data
- Insights are generated, not entered — user sees them but can't edit them
- Care provider views are opt-in and require explicit user consent per provider
- All insights are computed from wellbeing + safety domain data

**Journeys served:** home-navigation (insights tab), REQ-003 (streak context)

---

## 7. Notifications

**Owns:** Push notifications, SMS alerts, in-app nudges, email digests.

**Core entities:**
- `NotificationPreference` — per-user channel + type configuration
- `PendingNotification` — queued notification with delivery status
- `CrisisAlert` — priority notification to angels (bypasses quiet hours)
- `CheckInReminder` — daily reminder at user's preferred time

**Key rules:**
- Crisis alerts ALWAYS deliver (bypass quiet hours, bypass do-not-disturb)
- Check-in reminders respect quiet hours and can be snoozed
- Nudges are never repeated within 48 hours
- SMS is only used for crisis alerts (cost control)

**Journeys served:** all journeys (cross-cutting)

---

## Domain Dependencies

```
Identity ──────── required by ALL domains (auth, user context)
Wellbeing ─────── reads from Identity (user profile)
Safety ─────────── reads from Identity, Circle (angel list for alerts)
Circle ─────────── reads from Identity (user profiles)
Consent ─────────── reads from Identity, Circle (who to share with)
Insights ────────── reads from Wellbeing, Safety (aggregation)
Notifications ──── reads from ALL (event-driven delivery)
```

**No circular dependencies.** Identity is the root. Insights and Notifications are leaf domains that only read from others.