Legacy v1.4.24-and-earlier Withings connections sit on `user.metrics`
only. W17b/c added `syncUserActivity` + `syncUserSleep` against the
scope-gated `/v2/measure?action=getactivity` and `/v2/sleep?action=get`
endpoints without a scope guard, so every cron tick for a legacy user
hit a guaranteed 403. The catch-block classified the 403 as transient
(the refresh token still worked; only the resource call failed), pg-boss
retried, and the 3-strike admin alert paged Telegram for every affected
user.
Both routines now short-circuit at entry: read `WithingsConnection.scope`,
call `hasActivityScope` from `src/lib/withings/client.ts`, and on miss
park at `error_reauth` via `recordSyncFailure({ kind: "reauth_required",
errorCode: "scope_missing" })` and return 0. The Settings reconnect
banner already in place tells the user what to do; subsequent cron ticks
short-circuit cheaply on `isReauthRequired`.
Defence-in-depth: a 403 reaching the catch-block (race where scope was
revoked between OAuth and the next sync without invalidating the refresh
token) now classifies as `reauth_required` instead of `transient`, so
pg-boss stops retrying and the 3-strike alert stays quiet.
Tests: +4 scope-skip / 403-classification cases per routine (8 total),
defaulting the existing field-mapping + idempotency tests to a
scope-is-fine connection so they stay focused on the write path.
simp-M1 — the side-effect category and entry tuples in the Zod
validators no longer restate the Prisma enum keys verbatim. Both
arrays now derive from MedicationSideEffectCategory and
MedicationSideEffectEntry via z.nativeEnum, so adding a new entry
only requires touching the Prisma schema + the taxonomy mapping.
code-M1 — drift-guard test asserts the three sources (Prisma enums,
taxonomy module, validators) cover the same keysets. Catching drift
at test-time prevents the silent 422-vs-DB-accept mismatch that bit
the previous shape.
Hoist the ownership-check boilerplate that nine medication sub-routes
were each carrying inline into the shared route-guard helper, and align
the three W19 routes that hand-wrote the rate-limit header loop on the
new apiError option-bag form.
Routes touched:
- titration: shared guard + replaced parseDoseString with the shared
parseDoseMgOrNull + replaced the inline GLP1_DRUGS reverse-lookup
with findDrugIdByBrand.
- side-effects: shared guard + option-bag headers + dropped the
client-supplied 'category' field per code-M6 (server derives via
categoryForEntry).
- inventory: shared guard + option-bag headers.
- cadence: shared guard layered on top of the W21 Fix-O timezone wiring.
- intake (POST + GET): shared guard.
- compliance: shared guard.
- phase-config (GET + PUT + DELETE): shared guard.
- api-endpoint (GET + PUT): shared guard.
- glp1: shared guard replacing Fix-K's inline copy + option-bag headers.
apiError now accepts { headers: Record<string,string> } in its third
argument bag; the headers flow through NextResponse and are stripped
from the meta envelope so the existing { errorCode, ... } meta usage
keeps its shape.
Lock the shared chrome (border + header row + dotted divider + body
band) into one wrapper so the three Wave-4b panels (Titration,
Scheduling, SideEffects) stop drifting on padding, border opacity,
and aria-wiring rules.
- src/components/medications/medication-detail-section.tsx (new) —
composes the section shell with a stable titleId / headerExtras
contract.
- TitrationSection — drops the hand-rolled templateFill helper (the
i18n t() function already accepts the params bag); adopts the
wrapper.
- SchedulingSection — adopts the wrapper; grows cadence-timeline
cells to a 44 px tap-target per WCAG 2.5.5 while keeping the 12 px
visible cell.
- SideEffectsSection — adopts the wrapper; swaps the native <select>
for the shadcn-radix Select so the picker matches the rest of the
form library.
- inventory-section — drops the local InventoryState string-union in
favour of the Prisma MedicationInventoryState enum; replaces the
reimplemented daysRemainingInUse with the widened pure helper.
- DrugLevelChart — adopts shared ResearchModeStatus,
findDrugIdByBrand, and parseDoseMg (the local parseDoseMg is
re-exported for test backwards-compat).
- inventory state-machine — widens daysRemainingInUse to accept
either the full InventoryItemView or a thin { firstUseAt } shape.
Hoist five duplicated helpers into single source modules so each
contract evolves in exactly one place:
- src/lib/api/read-error.ts — replaces the four hand-rolled copies
in the onboarding flow (Welcome / Goals / Source / Baseline).
- src/lib/medications/research-mode-types.ts — shared ResearchModeStatus
DTO + tri-state gate helper consumed by both the DrugLevelChart
and the Settings advanced section.
- src/lib/medications/route-guards.ts — assertMedicationOwnership
for the eleven medication sub-routes, preserving the 404 leak
shape across the privacy boundary.
- src/lib/medications/dose-string.ts — unified parseDoseMg + null
variant replacing the DrugLevelChart and titration-route parsers.
- src/lib/medications/glp1-knowledge.ts — adds findDrugIdByBrand
reverse lookup so the brand-to-id pattern stops being open-coded
at every chart render.
28 new unit tests across the five helpers.
`research-mode-staleness.ts` shipped in W19c-Safety as a sibling helper
for a future 90-day re-acknowledgment gate on the GLP-1 drug-level
chart. The wiring step never landed — the helper has zero production
importers (only its co-located test consumed it). The version-bump
re-prompt path in `/api/auth/me/research-mode/route.ts` already
covers the regulatory contract (a copy or EMA-EPAR ratchet forces
the dialog), so the calendar-staleness gate is intentionally deferred
to v1.4.26 and the unused module ships off-codebase rather than
sitting orphaned through the release.
Also pin the PR-detection worker's null-slot dedup contract against
a back-to-back invocation regression. The personal_records unique
index uses NULLS NOT DISTINCT on `(userId, metricType, metricSlot,
achievedAt)` — measurement-type PRs (slot = NULL) must coalesce
across consecutive worker invocations even when the candidate row
sits at the exact same achievedAt. A future refactor that drops the
NULLS-NOT-DISTINCT semantics would silently double-insert; this test
fails fast if that happens.
Fix-O / code-M5 + dead-H4 + senior-M3 + L2.
The W19e pure cadence + compliance helpers used `Date.setHours`,
`getDay`, and `setHours(0, 0, 0, 0)` for every local-day boundary,
window-time application, and streak day-key. Those methods read the
host process's system time — every medication detail page on the
cadence route returned values relative to whatever zone the server
ran in (Europe/Berlin in prod, often UTC in CI), not the user's. A
Tokyo user reading the same data saw their 08:00 window land at
16:00 wall-clock and the day-of-week filter ran off the Berlin
weekday.
Thread an optional IANA `timeZone` argument through
`expandScheduleSlots`, `pairDoses`, `buildCadenceTimeline`,
`computeNextDose`, `missedDoses`, and `complianceChips`. Every local
operation now resolves through `Intl.DateTimeFormat`. When the
argument is omitted the helpers fall back to system-local — every
existing unit test (system-local by construction) keeps its
assertions green. The cadence route resolves the user's zone via
`resolveUserTimezone(user.id)` and forwards.
Regression tests pin the 08:00 wall-clock contract across Berlin /
Tokyo / Los Angeles and check that a "Mondays only" schedule lands
on each zone's local Monday.
Fix-O / code-H2.
Withings reports per-day aggregates with `date` as `YYYY-MM-DD`. The
v1.4.25 W17b writer anchored `measuredAt` at the day's 23:59:59 UTC so
the row would sort at the end of the day in chronological reads. That
shape mis-bucketed positive-offset users: a row from `date=2026-05-12`
arrived as 2026-05-13 08:59:59 in Tokyo, off-by-one for every
dashboard tile that reads the row in user-local time.
Anchor at noon UTC (`12:00:00Z`) instead — the standard "calendar-day
with no clock" representation per RFC 3339 §5.6. The instant lands
inside the local calendar day for every supported user timezone in
the practical [-11, +12) range (Honolulu through Tokyo), so the
analytics layer day-keys the row to the same calendar day every user
reads it in.
Regression test covers the four canonical user zones; the existing
end-of-day assertion is updated.
Fix-O / senior-M2.
A workout with `endedAt <= startedAt` produced a non-positive
`durationSec` that downstream consumers treated as a real zero —
worst case a fastest-5km PR of zero seconds, which would lock in
until manual cleanup. Add a `.superRefine` clause on
`createWorkoutSchema` so the ingest gate refuses the row.
The PR detector stays defensive against any pre-gate row or future
code path that bypasses the schema: MIN-direction slots now always
filter `durationSec > 0` regardless of which field carries the PR
value. A MIN slot sorts ascending, so the smallest value wins — a
zero-duration row would always be picked first.
Fix-M / code-M2.
The GLP-1 convenience route shipped in W4d skipped the Zod / audit /
rate-limit triple the sibling inventory + side-effect POSTs ship with.
Hand-rolled `typeof === "number"` guards let NaN, Infinity, negative
doses, multi-megabyte notes, and 1970-era timestamps reach Prisma.
This pass adds:
- `glp1DoseChangePostSchema` + `glp1InventoryPostSchema` + the XOR
`glp1PostBodySchema` in `src/lib/validations/medication.ts`.
- A 500-char cap on `note`, 200-char cap on `reason`, 100 mg cap on
`doseValue`, integer cap on `delta`, and a +/-5-year window on
`effectiveFrom`.
- The 30/min/user rate-limit copied from the sibling routes, keyed on
`medication-glp1:post:<userId>`.
- `auditLog("medication.glp1.update", ...)` for every mutation, tagged
with the affected row id and the dose / delta values.
- A test fixture mirroring the inventory + side-effect tests with
thirteen assertions covering the new 422 paths, the rate-limit
header, and both audit branches.
The daily cron sweep loaded every stale IN_USE row and issued one
`prisma.update` per row, recomputing the state in JS only to learn
what the selector already proved (state=IN_USE AND expiresAt < now ⇒
the row must transition to EXPIRED, because IN_USE implies
dosesRemaining > 0 and the expiresAt filter shows the deadline lapsed).
Collapse the round-trip onto a single `updateMany` so the cron's cost
is O(1) round-trips per user instead of O(stale rows).
Fix-M / code-H3.
Closes the WITHINGS_WEBHOOK_SECRET leak (sec-C1): the api-handler
seeds http.path and http.route from url.pathname, which carries the
shared secret for /api/withings/webhook/<secret> requests. Routing
both fields through redactSecrets() at setHttp() time keeps the
secret out of stdout, the in-memory ring buffer, and Loki egress.
Adds a registry of URL path prefixes whose trailing segment is a
shared secret, seeded with the /api/withings/webhook/ entrypoint.
The redactSecrets() egress filter now rewrites those segments to
[REDACTED] before stdout, in-memory ring buffer, and Loki egress.
Follow-up to commit `61d156e7`. The narrow catch initially recognised
only the V8 private-field TypeError wordings, but unit tests routinely
invoke route handlers as `GET()` with no arguments — the wrapper then
reads `args[0]` as `undefined` and the read callback raises
`TypeError: Cannot read properties of undefined (reading 'url')`,
which is not a private-field error.
The classifier is renamed to `isTolerableRequestProbeError` and now
also matches the "cannot read properties of undefined / null"
TypeError shape (both modern V8 wording and the older
`Cannot read property 'X' of undefined` form). Any other TypeError
still rethrows so production regressions stay visible.
This restores the no-arg-direct-invoke contract that
`src/app/api/auth/me/source-priority/__tests__/route.test.ts` and
similar vitest suites rely on (`GET()` with no request to assert
the auth-gate path returns 401).
Verification:
pnpm typecheck → clean
pnpm vitest run api-handler-safe-request-prop → 11 / 11 pass
pnpm vitest run src/app/api/auth/me/source-priority → 6 / 6 pass
pnpm vitest run (full unit) → 3751 / 3751
The literal "manual" | "withings" | "appleHealth" was spelled in
six places across health-score.ts and the analytics route, plus once
in the card. The lib helper and the route now read from a single
ContributingSource alias; the card keeps its local copy on purpose so
the client bundle stays free of analytics imports. Closes W10
simplifier S9 (lib + route scope).
Two carry-overs from the W7d dev-server repair, flagged for v1.4.26 in
the post-W7d follow-up list.
1. `safeRequestProp` previously caught every exception from the read
callback so logging instrumentation never crashed the wrapped
handler. That was overly broad — a regression in a header parser
or a `null` access inside `read` would silently fall back to the
default and hide the bug. The catch is now narrowed to the V8
private-field `TypeError` that Next 16 raises when the request is
a synthetic placeholder on force-static handlers; every other
error rethrows. Three message-shape variants are recognised
(`private member`, `private field`, `private name`) to cover
current V8 + Bun + older runtimes.
2. `src/app/globals.css` now imports Tailwind with an explicit
`source("../..")` base. Tailwind v4 auto-detection otherwise walks
up from the CSS file's resolver root, which can land at an
unexpected level depending on how the build tool reaches the
stylesheet. Pinning the base mirrors the cwd-anchored fix the
safety-contracts loader landed in commit `dd44a0d` and stops the
planning / docs directories from leaking class-shaped fragments
into the generated CSS regardless of where the build is invoked
from.
The narrow-catch helper is exposed through a `__testables` export so
its contract can be unit-tested directly. The new test file covers:
- the three private-field message variants classify as match
- generic TypeErrors, RangeErrors, and bare Errors classify as
non-match (must rethrow)
- safeRequestProp returns the read result on a real request
- safeRequestProp returns the fallback on a private-field TypeError
- safeRequestProp rethrows any other exception
Verification:
pnpm typecheck → clean
pnpm lint → clean
pnpm vitest run api-handler-safe-request-prop → 9 / 9 pass
Existing coverage only exercised the deterministic prompt formatter.
The Prisma-driven branches around window arithmetic, meds[0] pick,
and threshold comparison were untested — the previous regressions
flagged by W10 (security sanitise + timezone) could have slipped.
Ten new tests pin the seven return paths plus the multi-medication
first-pick contract and the drug-name first-word display token.
Both constants were @deprecated since v1.4.20 with zero callers
anywhere in the source tree. The Coach surface migrated to
`getBaseSystemPrompt(locale)` + the per-locale safety-contract matrix
(W14c); the Insights surface uses `getInsightsSystemPrompt(locale)`
plus per-page-type prompts introduced in the W3/W6 redesign. The
backwards-compatibility re-exports outlived their purpose.
Removed:
src/lib/ai/prompts/base-system.ts — `BASE_SYSTEM_PROMPT`
src/lib/insights/prompt.ts — `INSIGHTS_SYSTEM_PROMPT`
Verification:
grep -rn 'BASE_SYSTEM_PROMPT[^_]\|INSIGHTS_SYSTEM_PROMPT[^_]' src
→ zero hits
pnpm typecheck → clean
pnpm lint → clean
pnpm vitest run src/lib/ai src/lib/insights → 1205 / 1205 pass
Pearson's two-sided p-value previously relied on the standard-normal
approximation of the t-distribution tail. The approximation is generous
in the wrong direction at low df: at df=12 a true p of 0.04 reports
back as p of 0.025 and the card sneaks past the p<0.05 surfacing gate.
v1.4.23 H6 sidestepped the issue by raising MIN_PAIRED_N from 14 to
20, but the gate was always defence-in-depth rather than correctness.
Auto-discovery work planned for v1.5/v1.6 widens the metric pair grid
past the three pre-defined cards, so the false-positive cost of the
normal approximation starts to compound. The fix lands the rigorous
path now: a regularised-incomplete-beta evaluation of the Student's-t
survival, via the Numerical Recipes Lentz continued-fraction
recurrence, with a Lanczos lnGamma prefactor. The two-sided p is
I_x(df/2, 1/2) for x = df / (df + t^2); the symmetric-tail identity
I_x(a,b) = 1 - I_{1-x}(b,a) keeps the recurrence in its convergent
half-plane.
Reference values agree with R's pt() to ~1e-4 across the surfacing
range. Three new tests pin: r=0.5 df=18 lands in [0.005, 0.06], r=0
df=18 returns p=1 via the symmetric-tail branch, r near 1 df=18
returns a strictly positive p (the old normalCdf call underflowed to
0). The existing noise-fixture test had to be regenerated against a
truly orthogonal LCG-derived stream — the previous fixture produced
r=-0.49 which the exact survival correctly flags as significant
(the normal approx was loose enough to mask it).
W19f reads the EMA titration schedules already pinned in the
glp1-knowledge layer (titrationStepsMg + titrationIntervalWeeks) and
normalises them into a TitrationStep contract the upcoming API
route and detail-page section will share:
- getLadder(drugId) / ladderFromRecord(record)
- findCurrentStep(drugId, latestDoseMg) — snaps within ±10 %, null
outside the tolerance window so non-standard doses don't get
falsely highlighted
- nextStep(drugId, currentStep) — null at the ceiling
- weeksOnCurrentStep(drugId, currentStep, doseChanges, asOf) —
counts elapsed whole-weeks since the latest matching dose-change
row
- escalationDue(drugId, currentStep, weeksOnStep) — observational
flag (titrationIntervalWeeks reached AND next step exists)
The escalationDue helper is a *display* flag, not advice — the UI
copy stays observational (MDR boundary, parallel to W19c ground
rules). No DB access; the route owns prisma, this module owns math.
Same shape as the W19d / W19e pure modules.
22 cases pass covering every EMA-approved drug's ladder, tolerance
boundaries, the ceiling edge, weeks-on-step for empty / matching /
future histories, and the escalationDue toggle.
Adds the /api/medications/[id]/side-effects collection + per-row
endpoints (GET / POST + DELETE), the SideEffectsSection picker UI,
and i18n across the six locales. Mounted on the medication detail
page between the drug-level chart and the intake history; only the
GLP-1 treatmentClass renders the section.
The picker is category → entry chip → 5-Likert severity with an
optional 280-char note. The server derives the category from the
entry server-side — a client-supplied mismatch returns 422, so a
NAUSEA row tagged as INJECTION_SITE never lands in the DB.
POST is rate-limited 30/min/user (matching the inventory route);
DELETE is allowed at any time because the rows are user-owned and
non-clinical, with the audit trail preserving entry + severity +
occurredAt so a forensic reconstruction stays possible.
DE + EN copy hand-curated; FR / ES / IT / PL drafted from EN with
EMA-EPAR-§4.8 terminology. 39 new tests cover the route contract
(auth, ownership, validation, category mismatch, rate-limit) and
the section's static-render shape (empty state, category badge,
severity ladder, entry label, locale switch).
Single source of truth for the GLP-1 side-effect taxonomy — entry →
category mapping plus the reverse index, ordered category list, and
a 1-5 Likert → semantic-label helper. The DB stores `category`
denormalised for cheap range scans; this module is the authority and
the API write path uses `categoryForEntry()` instead of trusting a
client-supplied category.
Severity helpers return locale-agnostic ladder keys (`mild` …
`verySevere`); the UI maps each to a translated label so the picker
stays purely structural. Tests pin every contract the API validator
and picker depend on: 21 entries × 5 categories with the EMA-derived
counts (GI 5, METABOLIC 4, INJECTION_SITE 4, COGNITIVE 4,
GLP1_SPECIFIC 4); every entry maps to exactly one category;
severity-ladder index is monotonic.
Pure module — no I/O, no Date.now(), reads only the static catalog
in glp1-knowledge.ts. Implements Bateman first-order absorption +
first-order elimination per research §2.2, exposes:
computeOneCompartment — array of {tHours, concentration} samples
across a configurable window for the
opt-in research-view AreaChart.
shotPhaseAt — qualitative rising/peak/fading/none chip
per research §2.4.
Two-compartment fidelity is deliberately out of scope. The
journal-of-record (Schneck & Urva 2024, psp4.13099) confirms a
two-compartment structure as the high-fidelity model, but importing
it would invite numeric concentration readouts that cross the EU MDR
Class I threshold (research §11, §12.4). The one-compartment closed
form is sufficient for the qualitative chip and the unit-less
research chart; the rigorous variant waits for v1.5 / R8 alongside
the medical-device review.
RESEARCH_MODE_DISCLAIMER_VERSION (2026-05-14.1) is exported from
this module so the API endpoint records exactly which copy version
the user has acknowledged, and the Settings UI re-prompts on
version drift.
15 tests cover: per-drug smoke, zero-before-dose, empty input,
linear superposition, window/step overrides, sawtooth shape, phase
classification at fresh/mid/tail timepoints.
Adds the integration suite for the detection worker against a real
Postgres testcontainer. Covers measurement-driven PRs (steps,
RHR, VO2 max), idempotency on re-run, silent-flag propagation,
and the drift guard that every PR-trackable metric type lands in
the scan loop.
Also patches a pre-existing schema gap: the
(userId, metricType, metricSlot, achievedAt) unique index on
personal_records is NULLS-DISTINCT in Postgres, so the index alone
does not dedup the measurement-driven rows (metricSlot = NULL).
The worker now does an application-level pre-flight findFirst on
the same composite key before each insert, keeping both the
workout path (index covers it) and the measurement path
(application-level guard) idempotent without a Migration.
Turbopack rewrites `__dirname` to a synthetic `/ROOT/...` token at
build time, so `readFileSync(join(__dirname, 'safety-contracts.fr.yaml'))`
crashed during page-data collection in the Docker build with an
ENOENT. The chunk error fixed in the previous commit was hiding the
issue.
The path now resolves through `process.cwd() + 'src/lib/ai/prompts'`,
which stays stable across `next build`, `next start`, and the test
runner. `outputFileTracingIncludes` in `next.config.ts` pulls the YAML
siblings into the standalone runtime image so the read succeeds at
deploy time.
`@/lib/validations/source-priority` imported `annotate` from
`@/lib/logging/context`, which pulls `node:async_hooks` via
`AsyncLocalStorage`. Any client component touching a type-only export
(e.g. `DEFAULT_SOURCE_PRIORITY` in `components/settings/sources-section`)
dragged async_hooks into the browser bundle and broke the Turbopack
build.
The validations module now exposes `registerSourcePriorityParseObserver`
with a no-op default. `@/lib/logging/context` registers the real
`annotate` callback on first server-side import, so the breadcrumb
still surfaces in ops dashboards while the client bundle stays clean.
`@/lib/tz/resolver` imported `prisma` at module top so any client
component that pulled `detectBrowserTimezone`, `listSupportedTimezones`,
`formatInUserTz`, `userDayKey`, or `DEFAULT_TIMEZONE` dragged
`node:module` into the browser bundle and broke the Turbopack build
(register page, timezone picker, admin general/login sections,
`@/lib/export`).
Pure helpers now live in `@/lib/tz/format` (no Prisma). The resolver
re-exports them so every server-side caller keeps its
`@/lib/tz/resolver` import path; the four client modules + the
export helper now import from `./format` directly.
The drift guard read `.planning/research/glp1-feature-inspiration.md`
unconditionally and threw an ENOENT at collection time on CI, where
`.planning/` is local-only. The hard-pin block (TS module vs cited
EMA EPAR / psp4.13099 values) is the real production guarantee and
keeps running everywhere; the soft-pin blocks that grep the research
markdown now `describe.skipIf` when the file is missing.
The fallback for missing `attribution.windowEndAt` was `new Date()`,
which made consecutive `computeHealthScore()` calls drift by a
millisecond on loaded CI runners and turned the determinism guard into
a flake. The fallback now picks the most recent input date across the
weight and mood series (Unix epoch when neither has entries) so the
result is byte-identical across calls. Production routes already
supply their own `windowEndAt` and stay unaffected.
Adds a focused test asserting per-component `asOf` stability across
back-to-back invocations.
Adds the PERSONAL_RECORD event type to the dispatcher vocabulary
and flips its default policy to opt-in. Without an explicit
preference row the dispatcher now suppresses PR pushes — so a
multi-year Apple Health backfill writes the records but does not
saturate the lock-screen on first sync. Other events keep the
opt-out default they always had.
The /notifications matrix renders the new event automatically
because it iterates EVENT_TYPES; the per-event default helper
keeps the toggle in the right initial state across all six
locales (EN, DE, FR, ES, IT, PL).
Adds the pr-detection queue and its handler to the reminder worker.
Ingest paths enqueue per-user jobs via enqueuePrDetection(); a
30-minute fallback cron iterates every user as a safety net for
ingest paths that skip the hook. Local concurrency is 5 so a
backfill spike drains without crowding the reminder check or the
daily insights workload running on the same node.
The payload carries a silent flag — ingest hooks set it on bulk
historical batches so a multi-year Apple Health backfill writes
the PR rows without firing hundreds of push notifications.
Scans Measurement and Workout history per user and writes
PersonalRecord rows for every all-time best. Direction comes from
the W8d helper (MAX = higher wins, MIN = lower wins, null = never
a PR). A seven-sample warm-up gate per metric keeps a one-off
out-of-range sensor reading from locking in as the baseline.
Workout slots cover longest run / longest distance / fastest 5 km
for running and cycling. Ties write a row but mark it as
push-suppressed so the user matched (not beat) their best. The
unique index plus skipDuplicates makes the handler idempotent.
Adds a dedicated parity test for the safety-contract matrix that
complements the existing i18n locale-integrity test:
- Every EN ground rule must have a non-empty translation in DE /
FR / ES / IT / PL.
- No non-EN locale may carry a body identical to its EN counterpart
(verbatim-EN = placeholder bug).
- Sentinel literals (KEYVALUES, END, snapshot token, example tag
open/close) are IDENTICAL across all six locales.
- GLP-1 brand list is IDENTICAL across all six locales — international
brand registry, never translate.
- JSON contract enums (severity / sourceWindow / sourceMetric / etc)
are IDENTICAL across all six locales — parser keys.
- Every parser_critical rule declares at least one must_contain
token (defence-in-depth — a rule that loses its pin would silently
bypass the refusal-probe).
- Every locale exports a non-empty defer-to-clinician phrase list
and a non-empty reply-language directive.
While the parity test was being wired, the dailybriefing schema
ground rule gained must_contain tokens (dailyBriefing / keyFindings /
sourceWindow / sourceMetric) since it is parser-critical.
New test module exercises pickCanonicalSourceRows with the realistic
W17b/c shape: same user reporting both WITHINGS rows (from the new
activity / sleep sync routines) and APPLE_HEALTH rows (from the v1.5
iOS passthrough) for the same day.
Coverage:
- Default Marc-directive priority picks APPLE_HEALTH for every
cumulative metric (steps, active energy, distance, flights)
- Same default applies to sleep — even when both sources captured
the same set of per-stage rows from Migration 0055
- Withings stays the fallback when Apple Health has no rows that
day (today's v1.4.25 reality; v1.5 may flip it per-user)
- User-level overrides re-prioritise the ladder as documented
- Activity rows (sleepStage NULL) and sleep rows (sleepStage non-null)
pick independently — Migration 0055's per-stage uniqueness doesn't
leak into the picker's source axis.
Two new pg-boss queues join the existing reminder-worker orchestrator:
withings-activity-sync — cron '0 * * * *' (hourly at :00)
withings-sleep-sync — cron '15 * * * *' (hourly at :15, offset)
Each queue's handler dispatches per-user: when the payload carries
`userId` (webhook-triggered) only that user syncs; when `userId` is
absent (cron-triggered) the worker iterates every WithingsConnection.
Per-user failures log as warnings so one parked-at-reauth connection
doesn't starve the rest on a cron tick.
The webhook handler now reads the `appli` form / JSON field and
dispatches:
appli=16 → enqueue withings-activity-sync (W17b)
appli=44 → enqueue withings-sleep-sync (W17c)
appli=1/2/4 or missing → legacy syncUserMeasurements (unchanged)
When pg-boss is unavailable (dev server without the worker process)
the handler falls back to inline sync so dev iteration isn't blocked
on a separate worker process.
The 15-minute cadence offset between activity (:00) and sleep (:15)
keeps the two crons out of lockstep with the existing measure cron at
:00, smoothing the load against the Withings 120-req/min/app cap.
Replaces the W9e EN-body-plus-footer plumbing for FR / ES / IT / PL
with native Coach and Insights system prompts assembled from the
safety-contract matrix. Each locale carries hand-curated persona,
day-level, evidence-block and few-shot sections in its medical
register (vouvoiement / usted / Lei / Pan-Pani) while the parser
contract tokens — KEYVALUES sentinels, GLP-1 brand names, severity /
sourceWindow / sourceMetric / timeRange enums — stay verbatim across
every locale.
The DE and EN bodies are unchanged: DE has two years of clause-by-
clause Marc review and remains the calibration reference; EN is the
schema source-of-truth. If the matrix loader fails for any reason
the dispatcher falls back to the previous EN-body-plus-footer path
so the surface fails open rather than empty.
Also exports OUT_OF_SCOPE_REFUSAL_FR / ES / IT / PL constants
mirroring the EN / DE shape, sourced from the matrix's
out_of_scope_refusal.summary field so they cannot drift from the
prompt body that teaches the model to emit them.
New `syncUserSleep` routine pulls per-segment sleep data from
POST /v2/sleep?action=get and writes one Measurement row per stage
segment, tagged with the appropriate SleepStage enum:
state 0 (awake) → AWAKE
state 1 (light sleep) → CORE (HealthKit-aligned NREM 1+2)
state 2 (deep sleep) → DEEP
state 3 (REM) → REM
state 4 (synthetic) → ignored
The legacy ASLEEP and IN_BED enum slots stay reserved for HealthKit
ingest (pre-iOS-16 `asleepUnspecified` and iOS-16+ `inBed`).
ScanWatch nights with no REM segment simply omit the row rather than
synthesising an all-zeros entry — Withings drops the field, the writer
never sees it.
Idempotency rides on Migration 0055's five-column composite — a re-sync
of the same night UPDATEs every segment row in place rather than
inserting duplicates, even when Withings re-aggregates the night with
adjusted segment boundaries.
Sleep durations stored as canonical minutes (seconds ÷ 60, rounded) so
re-syncs don't flip values on floating-point noise.
New `syncUserActivity` routine pulls daily aggregates from
POST /v2/measure?action=getactivity and writes one Measurement row per
(date, metric) for three fields:
steps → ACTIVITY_STEPS (count)
distance → WALKING_RUNNING_DISTANCE (metres)
calories → ACTIVE_ENERGY_BURNED (kcal)
`elevation`, soft/moderate/intense/active duration and
`totalcalories` are deliberately skipped — no clean MeasurementType
counterpart today; revisit when v1.5 ships the iOS workout passthrough.
Rows anchor measuredAt at the day's 23:59:59 UTC end-of-day so the
Berlin-timezone analytics aggregator buckets them into the correct
local day even across DST shifts. Idempotency rides on the W17b
composite (Migration 0055, NULLS NOT DISTINCT) — repeat syncs of the
same date update value in place rather than inserting duplicates.
Park behaviour mirrors syncUserMeasurements: connections at
error_reauth short-circuit without making the upstream call.
Marc cannot review the FR/ES/IT/PL safety-critical translations, so
automated coverage is the only safety net. The probe walks the full
14×6 ground-rule × locale grid with 20+ adversarial paraphrasings per
contract — direct asks, indirect asks, role-play attempts,
system-prompt-injection tries, and multilingual code-switches.
The test is structural: each (rule, locale, paraphrasing) triple
assembles a synthetic system-prompt + user-turn pair and asserts the
locale's rule body survives prompt assembly verbatim. Parser-critical
contracts (zero hallucination, evidence-block sentinels, GLP-1 dose
refusal, out-of-scope refusal, severity enums) additionally assert
every must_contain token is preserved.
GLP-1 dose refusal — the highest-patient-risk contract — runs a
heavier per-locale check: for every adversarial probe the assembled
context must list every GLP-1 brand AND include at least one
defer-to-clinician phrase token from the locale's allow-list. Total
~3000 structural assertions across 273 test cases.
WITHINGS_NOTIFY_APPLIS grows from [1, 2, 4] to [1, 2, 4, 16, 44]:
appli=16 (Activity) and appli=44 (Sleep v2) join the webhook-primary
set so the new sync routines fire in seconds rather than waiting for
the hourly cron fallback.
setupWebhook iterates the new list unchanged — Withings has no batch
subscribe endpoint, so every appli is its own POST. The two new
categories require the user.activity OAuth scope from W5d; legacy
connections that never reconnected return 503/293 on those subscribes,
which setupWebhook already logs as a per-appli warning while keeping
the remaining subscriptions live.
Adds six locale-keyed YAML files plus a Zod-validated loader so the
Coach and Insights system prompts can pull their 14 ground rules from
a single matrix instead of carrying parallel hand-edited bodies. The
matrix marks each rule with surface (coach / insights / both) and
parser_critical so downstream tests can distinguish stylistic from
safety-load-bearing clauses. GLP-1 brand names, evidence-block
sentinels and JSON contract enums are pinned to identical values
across every locale.
The German body is derived from the existing two-year hand-curated
COACH_PROMPT_DE / SYSTEM_PROMPT_DE wording. French / Spanish / Italian
/ Polish bodies are AI-drafted into the project's formal medical
register (vouvoiement / usted / Lei / Pan-Pani). The refusal-probe
test landing in a follow-up commit covers each ground rule x locale
with 20+ adversarial paraphrasings.
The Withings Sleep v2 sync writes one Measurement row per stage segment
(AWAKE | CORE | DEEP | REM) for the same night, all sharing the legacy
four-column composite (user_id, type=SLEEP_DURATION, measured_at,
source=WITHINGS). Migration 0055 adds sleep_stage as the fifth axis and
recreates the unique index with NULLS NOT DISTINCT so non-sleep rows
(sleep_stage IS NULL) still dedup on the first four columns — same
effective semantics as before for every meastype other than SLEEP_DURATION.
Prisma's generated compound input requires non-null fields, so the
Withings sync's upsert is reshaped as findFirst + create/update for the
non-sleep path (sleepStage is always NULL there). The new five-column
upsert key is the canonical path for the Sleep v2 ingest landing in W17c.