Tre leve per migliorare l'accuratezza del piano oltre Mifflin-St Jeor.
1. Composizione corporea (US Navy + Katch-McArdle)
- Profile: neckCm/waistCm/hipsCm opzionali in input
- bodyFatPct calcolato lato server con la formula US Navy metrica
(Hodgdon-Beckett 1984) e cifrato at-rest come dato sanitario
- BMR usa Katch-McArdle (FFM-based) quando bodyFatPct è noto, Mifflin
altrimenti — più accurato del 5-10% sui casi fuori-norma
- Test: estimateBodyFatPercentageUSNavy + calculateBmrKatchMcArdle
2. Deficit dinamico
- Profile.targetWeeklyLossKg (kg/settimana, opzionale)
- computeDailyKcalTarget calcola il deficit da kg/sett ×7700/7 con due cap
di sicurezza: 30% TDEE (Schoenfeld 2014) e 1% peso/sett (Helms 2014)
- plan.routes lo usa al posto del -500/-200 hard-coded
- Test: 6 casi (mantenimento, transizione, intensiva, cap peso, cap TDEE,
default mancante)
3. Condizioni mediche strutturate
- Profile.medicalConditions (array JSON cifrato)
- 11 condizioni: 7 adattabili (tiroide → BMR -10%, diabete II, IBS,
dislipidemia, ipertensione, reni → cap proteine 0.8 g/kg, fegato →
1.0 g/kg) + 4 escludenti (gravidanza, allattamento, diabete I, disturbi
alimentari)
- PATCH /me/profile/conditions per salvataggio mirato dall'onboarding
- hasExcludingCondition: plan.routes restituisce 409 medical_block se
attiva una condizione incompatibile
Onboarding allargato a 6 step: aggiunto "Condizioni mediche" tra Profile
e Preferences. Step labels e numeri rinumerati.
Schema migration 20260429203451_add_profile_health_fields.
44/44 unit test verdi.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ADR 0003: VAPID self-hosted today, DeviceToken model agnostic to
platform so iOS/Android (Expo/APNs/FCM) plug in as new senders later.
Backend (apps/api/src/modules/notifications)
- sender.ts: NotificationSender interface, WebPushSender via VAPID
- notifications.routes.ts: GET /me/notifications/config, POST/DELETE
/me/device-tokens, PATCH /me/notifications/settings, POST /me/notifications/test
- scheduler.ts: node-cron Mon 09:00 Europe/Rome for weekly weigh-in
reminder; auto-cleanup of expired tokens on 404/410
- env: VAPID_PUBLIC_KEY/PRIVATE_KEY/SUBJECT (all optional → push gracefully off)
Frontend
- public/sw.js minimal (push + notificationclick)
- lib/notifications/push-client.ts: subscribe / unsubscribe / getCurrentSubscription
- profile/notifications-{actions,panel}.tsx: editorial panel with toggles,
device list, "send test", per-device removal
- pushReady requires both permission AND active subscription (covers the
case where the user revoked the SW but kept the browser permission)
Schema
- DeviceToken { userId, platform, endpoint, p256dh, auth, token, userAgent,
createdAt, lastSeenAt } with unique(userId, endpoint)
- ExtendedPrismaClient type exported from @ketopath/db
- NotificationSettings zod schema in @ketopath/shared
Tooling
- lint-staged: split .js out of eslint glob so service worker is only
formatted (it lives outside the TS project)
i18n
- Notifications namespace (it) with typed error keys
Smoke tested: POST /me/device-tokens 201, POST /me/notifications/test 200,
real push delivered to a macOS Chrome device.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Recipe detail
- Seed catalog of 33 italian ingredients with macros, allergens, avg price.
- Each seed recipe now declares its RecipeIngredient links (qty + unit).
- /recipes/[id] page: editorial layout with prep meta, ingredients list,
macros block, preparation, chef's note. Plan-week meal names linkable.
Shopping list
- GET /me/shopping-list aggregates RecipeIngredient quantities of the
ACTIVE plan, grouped by ingredient category, sorted alphabetically.
- /shopping page: total items, estimated cost, per-category checklist
with check-off persisted in localStorage per plan id.
- Home gets a "Spesa" nav item; /plan links the list when a plan exists.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@ketopath/shared:
- tracking/schema.ts — Zod schemas for WeightEntry input (weight, optional
measurements/notes/energy/sleep/hunger, photo URLs) and FastEvent start /
update, plus PROTOCOL_DEFAULT_MINUTES table
- planner/macros.ts — macrosForPhase(): protein 1.6-1.8g/kg, netCarb 25/60/120g
by phase, fat = remainder (PRD §9.3)
- planner/matchmaking.ts — pure matchMeals() that filters by exclusion tags,
phase compatibility and meal category, then scores by euclidean distance
from the meal's macro target with a 1.5 penalty for recently-consumed recipes;
DEFAULT_MEAL_SHARE constant (25/35/10/30 %)
- 5 new unit tests cover exclusion, phase, recency, topN, category mismatch
apps/api:
- modules/tracking/weight.routes.ts — GET /me/weight-entries (last 60),
POST /me/weight-entries with upsert on (userId, date); encrypted fields
serialised to/from JSON strings
- modules/tracking/fast.routes.ts — GET, POST (start) and PATCH (update) on
fast events; symptoms stored as encrypted JSON
- modules/plan/plan.routes.ts — POST /me/meal-plans:
- reads profile + preferences, computes BMR (Mifflin-St Jeor), TDEE, daily
kcal target with the phase-dependent deficit, then macros via macrosForPhase
- upserts a MealPlan rooted at Monday-of-this-week, wipes prior slots, and
fills 28 slots running matchMeals per (day, meal) with a recent-2-day
rolling exclusion list to keep variety
- selected recipe + 4 alternatives per slot
- GET /me/meal-plans/current returns the active plan with selected/alternatives
@ketopath/db:
- prisma/seed-recipes.ts — 16 italian keto recipes (4 per meal type) with
estimated macros per serving and phase compatibility
- prisma/seed.ts — idempotent insertion (skip on existing name match)
Verified end-to-end with sign-up → create profile (Michele PRD persona) →
generate plan: 28 slots filled, daily target 1399 kcal / 137 P / 83 F / 25 C,
matchmaking distributes 4 different breakfasts across the first 4 days as
the recent-consumed penalty kicks in.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Generated by `prisma migrate dev --name onboarding_v1_recipes`.
Adds:
- users.disclaimer_accepted_at (TIMESTAMP NULL) for the medical disclaimer flow
- weight_entries, fast_events with their indexes and FK cascade
- ingredients, recipes, recipe_ingredients with category index on recipes
- meal_plans, meal_slots with composite uniqueness on (planId, dayOfWeek, meal)
- 4 new enums: FastStatus, MealCategory, Difficulty, MealPlanStatus
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Schema changes (require a single migration to apply):
- User gains disclaimerAcceptedAt — null until the user accepts the medical
disclaimer, blocks /profile until set (PRD §14.3)
- New WeightEntry (PRD §5.2) — weight/measurements/notes encrypted at rest,
energy/sleep/hunger left in the clear so we can produce aggregate analytics
- New FastEvent (PRD §5.3) — symptoms/notes encrypted, protocol/status/timer
in the clear; indexed on (userId, startedAt)
- New Ingredient / Recipe / RecipeIngredient (PRD §5.1) — italian keto recipe
database with macros and exclusion groups, ready for the matchmaking algo
- New MealPlan / MealSlot — generated weekly plan with selected recipe and
alternatives per (day, meal)
- New enums: FastStatus, MealCategory, Difficulty, MealPlanStatus
Web — disclaimer flow:
- /welcome page (server component, requires auth) lists the 4 PRD-mandated
points and surfaces the 3 mandatory checkboxes; submit triggers a server
action that stamps disclaimerAcceptedAt and redirects to /profile
- /profile redirects to /welcome whenever disclaimerAcceptedAt is null,
so the disclaimer becomes a hard prerequisite for any sensitive page
- New Italian copy under Welcome.* in messages/it.json
- @ketopath/db added to apps/web dependencies (was indirect via @ketopath/auth)
@ketopath/db re-exports the new models and enums.
Run the migration before testing: `pnpm db:migrate --name onboarding_v1_recipes`.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Generated by `prisma migrate dev --name encrypt_profile_fields`.
ALTER TABLE "profiles" sets weight_start_kg / weight_current_kg /
weight_goal_kg / target_date to TEXT. Existing rows keep their decimal
representation as plaintext text (e.g. "76.00") and prisma-field-encryption
re-encrypts them on the next write — both ciphertext (v1.aesgcm256.<keyId>.
<iv>.<ciphertext>) and legacy plaintext are decrypted transparently on read,
so no app-level fallback was needed.
Verified end-to-end:
- pnpm test:e2e — all 4 specs pass, including profile signup → submit →
BMR 1583 / TDEE 1899 round-trip
- raw SQL on profiles shows ciphertext for new rows; older rows stay in clear
text until their next update
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Wire prisma-field-encryption AES-256-GCM extension on the shared Prisma
client and annotate the four sensitive columns on Profile with @encrypted:
- weightStartKg / weightCurrentKg / weightGoalKg (Decimal → String)
- targetDate (DateTime @db.Date → String, ISO YYYY-MM-DD)
Other Profile fields stay in clear text per ADR 0002 (age, gender,
heightCm, activityLevel) — they're needed for plan generation and
aggregate analytics, and are not strongly identifying on their own.
apps/api profile.routes.ts:
- serialize() now reads the columns as strings and parses them back to
numbers for BMR/TDEE; targetDate is already an ISO string from the DB
- the upsert stringifies numeric inputs and slices the date to YYYY-MM-DD
Env wiring:
- packages/db, apps/api, apps/web .env.example all document
PRISMA_FIELD_ENCRYPTION_KEY (k1.aesgcm256.<base64url>) — must match
across every process that hits the DB
- key generation snippet documented inline
Migration is intentionally NOT in this commit: needs to be created against
a live Postgres instance and applied. The fields change Decimal/Date → text
so prisma migrate dev will require a USING cast — see the follow-up commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- User extended with emailVerified, name, image (Better Auth fields)
- Session: signed session tokens with expiry, ip and user agent
- Account: credential rows for email+password (provider 'credential') and
OAuth providers (e.g. Google) — password hash stored on the credential row
- Verification: single-use tokens for email verification, password reset,
and magic links
- Re-export new types from @ketopath/db
Migration 20260429104721_add_auth_tables run against local Postgres.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Generated by `prisma migrate dev --name init` against the local Postgres.
Creates the 6 enums and the 3 tables with their unique indexes and
foreign keys (profiles.user_id and preferences.user_id with ON DELETE CASCADE).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Initial schema covers only the 3 base entities from PRD §8 — additional
models (MealPlan, Recipe, WeightEntry, FastEvent, ShoppingList, etc.)
will be added alongside their respective features.
- Postgres datasource via DATABASE_URL
- Enums: Role, Gender, ActivityLevel, Phase, CookingTime, FastingProtocol
- Singleton PrismaClient export from @ketopath/db
- Root scripts: db:generate / db:migrate / db:studio / db:seed / db:format
- Seed stub at prisma/seed.ts
- Ignore .claude/settings.local.json (per-user CLI permissions)
The first migration must be run by the developer:
pnpm db:migrate --name init
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
