A cluster of related correctness gaps in the compliance denominator and the
insight surfaces that read it:
- Paused medications. A doctor-ordered pause flipped `active` off and stamped
`pausedAt`, but resume nulled `pausedAt`, so the interval was irrecoverable
and every paused day became a missed dose after resume — collapsing
adherence across the Health Score, the BP gate, the Coach snapshot and the
report. Persist pause intervals as durable eras (new `MedicationPauseEra`,
migration 0211), close the open era on resume, thread the eras through the
compliance context and drop slots inside a paused interval from the
denominator. Every compliance call site now fetches and passes the eras.
- Health Score eras. The Health-Score fast path was the only compliance
caller that skipped the shared schedule select and schedule revisions, so
after a schedule edit it scored past days against the current schedule
while every other surface used the historical era. Switch it to the shared
select and thread revisions (and pause eras) like the canonical path.
- Insight bucketing timezone. The per-metric status series bucketed every
user in Europe/Berlin, so a near-midnight reading landed in the wrong day
and the relative-day phrasing was off for non-Berlin self-hosters. Thread
the resolved user timezone and key off `userDayKey`.
- Correlation significance. User-facing correlations surfaced from five
paired points with strength read straight off the coefficient. Route them
through the rigorous engine (at least twenty pairs, Student-t p below 0.05)
so small-N coincidences no longer render as confident findings.
- Resting-pulse input. The mood-versus-"resting pulse" correlation fed the
raw PULSE day-mean, which folds in workout HR, under a "Resting pulse"
label. Feed `resolveRestingPulseSeries` (clean RESTING_HEART_RATE, else a
daily low-percentile proxy) so the label matches the input.
Code-review and design follow-ups before the v1.22 cut. No behaviour change
beyond the items below.
Coach / insights:
- Coach SSE route: a non-provider error raised inside the open producer (e.g. a
DB failure persisting the reply) now records on the wide event and forwards to
GlitchTip instead of masquerading as a provider outage; it still surfaces an
error frame since the stream is open.
- Lab ↔ outcome correlations: hedge faint-tier pairs ("a faint hint, if
anything … too small to lean on") so the narrated confidence never outruns the
effect, matching the daily engine.
- Reminder sweep: mint the plan-review reminder and clear the plan's reviewDate
in one transaction so a half-applied write can't duplicate the reminder.
- Fold the two raw provider-fallback runners onto one shared chain walker.
Design / accessibility:
- Lift sub-44px tap targets to the 44px mobile floor (suggested-action card,
Coach bubble affordances), brighten the resting timestamp clock for non-text
contrast, and route the lab chart height through the shared constant.
- Replace ad-hoc arbitrary type sizes with the scale token and static size-11
icon buttons with the named icon-lg variant.
- Settings: add the missing About card-header descriptions, normalise the
low-stock numeric widths, and relax the notification-status rows from nested
cards to a divided list.
Housekeeping:
- Drop three unused exported types (and the now-orphaned source const), re-stamp
v1.21.5 comments to v1.22, and register the live-judge eval harness as a knip
entry so the script-only module resolves.
Add three per-user AI settings, all configured in the app (no new env vars),
behind one migration (0198):
- Response timeout (seconds). A per-user upstream timeout threaded onto the
provider call. A local/self-hosted backend can take more than the legacy
60 s default to load a model on the first request, which timed the Coach
out before the first token. Defaults to a generous ~180 s when unset;
validated to 10–600 s.
- True token streaming for the local OpenAI-compatible client. The client now
POSTs with stream:true and parses the SSE delta chunks, emitting tokens as
they arrive, with a per-idle-gap timeout instead of a whole-call ceiling and
a graceful fallback to the buffered path when a server cannot stream. The
Coach route runs the provider call inside the response stream behind a
keepalive heartbeat so a reverse proxy no longer idle-drops the connection
during model load or generation. Grounding, refusal, and sentinel-stripping
still run on the complete reply before any token reaches the client.
- Dedicated document-scan (Lab-OCR) provider. Off by default — the OCR path
uses the main provider chain. When enabled, lab-report processing resolves a
separate provider/model/key, with a hint when the chosen model is not
vision-capable. The new key column is AES-256-GCM at rest and registered in
the encrypted-columns registry and the rotation script.
i18n: EN + DE translations for the new Settings → AI strings (other locales
carry interim copies to keep locale parity green).
Add the medication-class to target-vital map and the adherence-to-vital
storyline so the Coach can surface, association-only, when an adherence dip
lines up with a drift in the vital that medication targets. Never causal,
never a dose-change suggestion; dose questions defer to the prescriber.
Add an in-repo CUSUM / binary-segmentation changepoint detector over DAY
rollups with a high firing bar (persistence plus within-segment-spread
magnitude) so the Coach can name a dated, sustained level shift instead of a
vague "lately", and stays silent when no break clears the bar.
Surface which source recovery is read from when two sources materially
diverge, with the honest reason, instead of silently averaging.
Lay the n-of-1 experiment substrate: a nullable encrypted outcome column on
coach_plans (migration 0197), the review_due / reviewed lifecycle statuses
wired into the encrypted-column registry and rotation script, and a daily
worker that reads the experiment result back in the user's own before/after
numbers — association-only, honest-null, no cheerlead on an adverse trend,
and routing a worsened vital to the clinician. The user-visible read-back
stays behind COACH_EXPERIMENT_VERDICT until its review cases clear.
Extend the outbound dose screen to catch experiment-shaped dose changes
(halve / skip / stop a medication to test it). Add evaluation cases and unit
tests for each new surface.
Feed the user's own usual blood-pressure range — the median plus or minus
MAD band, computed from the readings already in the snapshot — into the Coach
context with a clause to quote those bounds verbatim, so a band is never
invented from window means. Thread the stored first name into the prompt and
surface it sparingly on roughly one turn in three, never as a rote greeting.
Add the context-layer rule that names the relevant context before a
suggestion, a single-chart-token allowance that works on every provider, and
a per-turn opener hint.
Wire the already-shipped suggested-action SSE frame to render: the streaming
hook collects it and the message thread shows the action card, mirroring the
reminder-suggestion path. Extend the eval golden set with band-verbatim,
chart-token, and no-overclaim cases, and pin a red-team case proving the
chart-token syntax cannot smuggle a dose past the outbound screen.
Add unit coverage for the closed when-grammar and `---REMEMBER---` capture
(propose-free active write, recall-only note, malformed drop, per-user cap),
the closed action-card allowlist (off-list and bad-interval rejection, no
auto-apply), the daily sweep (due flip, plan-review mint, undecryptable
skip), and the routes (owner-scoped no-IDOR 404, field-by-field build, the
off-allowlist 422, and the module gate).
Capture the casual "remember this / remind me about X" utterance the Coach
used to drop. The model emits a provider-neutral `---REMEMBER---` sentinel
that the chat route strips inline on EVERY turn — so a short conversation,
where the request usually happens, no longer waits for the long-chat memory
worker. The note is the user's own explicit request, so it is stored
silently as active; a validated closed grammar (ISO date, relative +Nd/+Nw,
or a context cue) sets the trigger, and a malformed block is dropped.
Surface the stored reminders two ways. A snapshot recall sub-block injects
near, overdue, and already-surfaced reminders so the Coach references them
unprompted next turn. A daily sweep flips overdue reminders to due and
mints a one-off review reminder from a plan whose review date has passed,
finally activating the plan review-date column. In-app surfacing only here;
push delivery follows later.
Also generalise the propose-then-confirm action card: a closed
`---SUGGEST-ACTION---` allowlist (a preventive-care item or a reminder note)
rides an additive SSE frame. Nothing is created until the user confirms.
Thread the rolling early-detection pass and the labs -> outcome pass through the
correlations route and the coach correlations reader, both of which already
build the same daily series. The route response gains an optional `emerging`
block (recent-window pairs not yet established, flagged provisional) and a
`labCorrelations` block; the coach reader gains matching `emerging` and
`labDrivers` fields. Regenerate the OpenAPI contract for the new response shape.
Drop the unused extractCoachClaims helper (+ its CoachClaim types) and the
unused CoachPlanStatus type — the deterministic grader uses the boolean claim
helpers directly, and the plan status enum is consumed via the const array.
Update the dashboard-snapshot suite for the locale-keyed cell: stub the i18n
context and add getQueriesData to the fake query client so the locale-aware
preload probe is exercised.
A deterministic golden-set + red-team floor that gates every Coach change
offline, plus an opt-in live-judge layer that scores warmth and groundedness
with a pinned model only when COACH_EVAL_API_KEY is set.
- 42 golden cases across grounding, cross-metric, data-honesty, provider-parity
and own-baseline taxonomies, graded by behaviour-anchored binary criteria.
- Structured-claim grounding helpers extend the numeric verifier from numbers
to threshold / own-baseline / confident-verdict claims; the numeric API is
unchanged so the existing grounding tests stay green.
- Red-team battery asserts the existing inbound refusal, outbound dose screen,
acute red-flag escalation and learn-link guards fire.
- Nightly + manual workflow runs the deterministic floor always; the live judge
step runs only when the secret is present and is non-blocking.
Add a CoachPlan model: durable, user-scoped if-then implementation plans
the Coach proposes and the user confirms. The background memory-refresh
pass extracts plans the user agreed to and writes them as "proposed"; only
the user-facing PATCH on /api/coach/plans/[id] activates one. The free-text
cue / action / target are encrypted at rest with the shared Bytes codec;
metric, status and dates stay plain so the injection picker ranks without a
per-row decrypt. Active plans fold into the Coach snapshot's memory block
and survive the prompt-budget drop alongside facts. List, confirm and
soft-delete routes are Coach-gated and narrow the owner from the session.
Key the dashboard-snapshot query cell by locale so a locale switch reads
freshly localised tile copy on its own cell instead of serving the prior
locale's prose until the slot goes stale. The zero-arg key stays
byte-identical, so every prefix-based invalidation keeps reaching the
locale-keyed cells; the offline-banner and preloader match on the prefix.
Migration 0194 (additive CREATE TABLE). Encryption-rotation registry and
script cover the three new Bytes columns.
Add a deterministic narrative layer the Coach narrates rather than invents:
- Briefing callbacks + forward-looks: the daily briefing recalls the prior
period and points ahead, anchored to a figure, one callback per surface.
- Tension Verdict: when the readiness composite's contributors disagree
(good sleep but a rising resting pulse), surface a brief honest internal
read on the health-score card. A real clinical red-flag is never reconciled
away into a calm verdict.
- Streaks + return-to-baseline: a pure detector over a DAY-bucket series
closes a worry once a metric returns inside the personal MAD band after a
genuine out-of-band run; a gap day breaks a streak.
- Coach prompt: add develop-discrepancy, roll-with-resistance, and an
anti-persuasion clause; ratchet PROMPT_VERSION.
The tension flag and streak event are computed server-side; the card lines
render resolved fields and the model narrates them.
The post-hoc numeric verifier only graded the tool path, so local /
no-tools providers got no grounding net. Grade the no-tools prose
against the snapshot the model was shown — its structured sections,
which already carry the correlations-snapshot block — and flag any
cited number that traces to no snapshot leaf. Non-blocking: the count
is annotated under coach.prose.number_unverified and the unverified
token soft-stripped, never a stream-killer. Only runs when the full
figures were delivered this turn; a cheap follow-up has no fresh set
to grade and no-ops. Tool-path behaviour and tolerance unchanged.
The correlation discovery matrix gained two non-MeasurementType channels
(medication compliance, symptom severity), but the Coach get_correlations
reader still filtered its measurement query by hand and never built the new
channels. The channel keys leaked into the Postgres enum cast and threw for any
user with medication or illness data.
Extract the route's two channel-series fetchers into a shared module
(correlation-channel-series.ts) so both consumers build the channels identically,
then wire the reader to derive its type filter from discoveryMeasurementTypes and
fold the fetched compliance + symptom series into the discovery input. The reader
now humanises the channel keys to "medication adherence" / "symptom severity".
The labs block already surfaces value + unit + range + date, and the Coach
prompt instructs it to quote them verbatim, so labs are cited end-to-end — no
gap. Pin that with a tool-level guard asserting value and unit survive
get_labs, so a future change to the labs pass-through can't quietly strip the
number the Coach needs to cite.
The Coach may point at a published /learn guide, but only a slug in the
catalog. That was a prompt instruction with no enforcement — a model could
hallucinate /learn/lower-your-cortisol and ship a dead link. Add a deterministic
post-filter on the assembled reply that keeps a published slug verbatim and
strips any reference whose slug is not in the catalog, tidying the prose around
the removed link. Makes the catalog's impossible-by-construction claim an
enforced guarantee.
The illness-recovery tool and the illness context block carried only labels,
lifecycle, and dates plus the recovery composite — never the recovery-gap, the
metric that drove it, the nadir, the pre-onset deviations, or the red flags the
illness card renders. So a recovery question got the composite, not the gap the
user sees, and a sustained-fever or low-SpO2 escalation was invisible
in-conversation.
Run the existing episode-correlation read-layer (the same engine the card and
the red-flag notifier use) for the most relevant episode — active first, else
the most-recently-resolved — and attach the computed scores to get_illness_recovery,
coverage-gated so a thin signal yields nothing rather than a fabricated number.
Document the illnessScores shape in the Coach prompt so it restates those
numbers verbatim and escalates a red flag rather than reassuring.
Each retrieval tool re-scoped the snapshot to a single source, so its cache key
never matched the inventory's full-source build — a four-tool turn paid four
extra full snapshot builds on top of the inventory's. Thread the inventory's
probe scope to every tool so a per-tool read lands the one cache entry the
inventory already primed, collapsing the turn to a single build (a window
override still gets its own correct build). Raise the loop's round cap from two
to three so a sequential cross-metric why-chain isn't starved; the per-round
token budget is unchanged and the absolute ceiling tracks one above so the
final round still forces prose.
The tool-mode path dropped the launch scope's sources: the inventory probes the
full source set, so a Coach opened from a metric page saw the whole inventory
and could roam, while only the no-tools path honoured the narrowing. Inject a
one-line FOCUS hint naming the launched domain(s) into the tool-mode user turn
so the model prioritises that metric and fetches its figures first, with every
other domain still reachable if the question leads there. Empty on a generic
open, so the prompt prefix is unchanged on that path.
The acute red-flag clause hardcoded its blood-pressure numbers as prose
literals and named no glucose or fever floor at all, so the Coach could
silently drift from the dashboard hero and the notification engine. Compose
the systolic/diastolic, glucose, and sustained-fever thresholds straight from
clinical-floors.ts and echo the previously-missing glucose and fever lines, so
every acute number the Coach states is bound to the one source of truth.
Add an outlook contract (gentle forecast, what-to-expect, anticipatory if-then
— all conditional, ranged, association-framed) composed beside the tone
contract on the Coach and the comprehensive briefing, so a reply can sharpen
expectations without a false promise. Tighten the tone contract against
over-validation (affirm the effort, not an unsafe choice; stay neutral when
there is nothing to praise) and route why/conflict questions through
get_correlations and its coincident-deviation flag before answering.
Coverage asserts the clause numbers equal the constants and the new contract
reaches both surfaces.
The FDR cross-metric driver pairs were reachable only through the
get_correlations tool, so a local/no-tools provider got the coincident
flag but zero discovered drivers and the cross-metric layer silently
degraded. The system prompt's fallback to a snapshot driver field
pointed at a field that did not exist on that path.
Attach a bounded top-3 slice of the same gated, ranked driver output the
tool serves to the no-tools snapshot floor, gated on the derived layer
and shed first under prompt-budget pressure. Local users now reach
parity on the flagship cross-metric capability.
The coincident-deviation read grouped the latest day with a UTC date
slice while the sibling readiness read already used the tz-aware day key.
For a user east or west of UTC, a late-evening or early-morning reading
landed on the wrong calendar day, so a fired flag could compare a vital
from the wrong day against its band and narrate '2 or more vitals out of
band today' on the wrong day.
Thread the user's timezone into the latest-day read and mint the day key
with userDayKey, matching readiness. The Coach derived-snapshot and the
correlations reader pass the real account timezone through. Adds a
regression covering a UTC+2 user whose late-night reading rolls to the
next local day.
The daily AI-token gate was a flat 25,000/day for every provider. That cap
exists to bound the OPERATOR's API bill, but a ChatGPT-OAuth (Codex) or BYOK
chain egresses on the USER's own plan/key and costs the operator nothing, so
gating it on the operator-cost ceiling is wrong. A single gpt-5.x reasoning
turn legitimately reports 20k-40k gross total_tokens (re-sent system prompt +
inventory + tool defs per round + hidden reasoning, summed across rounds), so
one or two turns exhausted the flat cap and locked the user out of a plan they
pay for.
Classify the chain's cost owner by its primary provider via resolveDailyCap:
an admin-openai primary (operator pays) keeps the operator-cost cap; every
user-egress primary (codex/openai/anthropic/local) gets a generous
abuse-only ceiling. Threaded into the Coach chat gate and both OCR-extract
modes.
Raise the operator-cost cap 25,000 -> 200,000 so the operator-key path also
survives a normal day of reasoning turns, keeping gross-token accounting
rather than reworking the reconcile math.
Subtract cached input tokens at reconcile: the Responses-API gross total still
includes prompt-cached input the user did not re-pay for. The codex client
already parsed cached_tokens; sum it through the tool loop and bill
total_tokens minus cached.
The multi-vital coincident-deviation flag (two or more vitals outside
their usual band on the same day, with the illness-explained reframe)
already reached the period narrative but never the Coach. Attach it,
fired-only, to the derived block so both Coach paths can narrate it: the
tool path via the recovery/correlations tools and the no-tools snapshot
floor. A quiet day adds no entry, keeping the snapshot noise-free.
The Daily Briefing strips any number absent from its server-computed
figures; the Coach's tool path had no equivalent, so a transcription or
paraphrase drift (a tool returns systolic 128, the reply says ~138) could
ship. Collect the numeric leaves from every present tool result this turn,
extract the numbers the reply asserts, and soft-correct any that match no
fetched figure to a neutral placeholder — annotated, non-blocking, and a
no-op on a qualitative turn or the no-tools path where there is nothing to
grade against. The bounded retrieval loop now returns the present results'
payloads so the check has an authoritative figure set.
The retrieval-tool DATA INVENTORY was built against the user's default
narration cluster (cardio/body/mood/medication), so sleep, glucose,
activity, body composition, vascular, SpO2, gait and environment series
reported absent even when the data existed — and the grounding rule then
told the model not to fetch them. Probe presence against the full source
set instead, and generate the metric-series inventory rows from the
complete source-to-section map so every series the user has rows for is
advertised. Per-tool reads still re-scope to their own domain, so the
wider probe never widens a figure read.
Add three retrieval tools whose snapshot blocks already existed but were
unreachable in tool mode:
- get_workouts: recent sessions + per-sport rollup
- get_cycle: phase / prediction / correlation (gated by cycle access)
- get_correlations: the FDR-controlled lagged cross-metric drivers plus
the coincident-deviation flag, surfaced descriptively
Also honour the caller's window in get_labs and get_illness_recovery so a
cross-metric answer no longer silently mixes horizons.
Sharpen the Coach's behaviour-change reflexes from the coaching literature,
prompt-only:
- Connect the signals on a why/pattern question — consult correlations and
link them descriptively ("after short-sleep nights your next-morning HRV
tends to read lower") instead of reading metrics in isolation, always an
association worth a small experiment, never a cause.
- Confidence ruler on action turns — after naming one small step, ask how
doable it feels 0-10 and shrink it when that lands low, keeping the choice
the user's.
- Three-beat shape on data-review turns — finding, likely driver, one small
step; name the dominant contributor when citing a derived band.
- Extend the MI micro-moves with developing discrepancy and rolling with
resistance, and add an anti-persuasion check: any suggested change must
serve the user's own stated goal.
One bilingual tone-calibration example added per locale.
The shared contracts cover chronic deferral (dose, diagnosis, drug-level
to a clinician) but had no acute branch. Add a closed-list crisis clause —
chest pain, syncope, sudden severe symptoms, hypertensive-crisis readings,
suicidal ideation — that points to prompt or emergency care without
diagnosing, and keep every other turn calm and non-alarmist. It rides the
single shared-contract source so it lands on the Coach, the per-metric
status cards, the comprehensive briefing, and the period narrative; the
coverage test asserts it on all four surfaces.
Thread the request abort signal through runCoachToolLoop into each
per-round completion so a mid-generation client disconnect tears down
the upstream call instead of paying the full token cost. Completes the
no-tools-path wiring to cover the tool-retrieval path as well.
Regenerate the OpenAPI spec for the trimmed OCR-capability reason enum.
Drop the unreachable text-only-model reason from the OCR capability
enum and its DTO type. Replace the literal NUL delimiter in the Fitbit
dedup map keys with a named constant so text tooling stops treating the
file as binary (behaviour unchanged). Correct the Coach tool-executor
docblock, which overstated cross-tool snapshot-cache sharing.
The F1 coach tools each rebuild a snapshot with a distinct single-source
scope, so the 60s snapshot LRU (keyed on the source list) never shares their
reads. On a multi-metric turn the model can fan out up to six tools that each
fire `extractFeatures` (a user lookup plus a windowed measurement read plus the
all-time extremes) and a prefs `findUnique` concurrently — up to twelve heavy
reads against the shared Prisma pool at once.
Memoise both reads on the request-scoped event cache, keyed on userId (and
windowDays for the features read), so the fan-out collapses to one of each
within a request while staying correct across requests. Project the feature
extractor's measurement read to the six columns its consumers actually use
(type, value, measuredAt, sleepStage, source, deviceType) so the 200k-row-class
PULSE / glucose windows stop pulling every column. Outputs are unchanged.
Replace snapshot-stuffing in the Coach turn with a small base context plus
on-demand retrieval. The model receives the full system prompt, a tool-mode
grounding addendum, and a tiny DATA INVENTORY manifest listing which domains the
user has logged and which tool fetches each; the figures themselves are pulled
only when a question needs them.
Six read-only tools wrap the existing server-authoritative snapshot builder:
get_metric_series (blood pressure / weight / pulse + the additive series),
get_glucose_panel, get_sleep, get_medication_compliance, get_labs, and
get_illness_recovery. Each returns a structured { present: false } for an absent
or opted-out domain rather than throwing or returning an ambiguous empty list,
so the model can always tell missing data from real data. Module and cycle gates
are enforced at the builder, so opted-out domains are unfetchable.
A bounded loop runs the call -> tool-result -> answer cycle: at most two rounds
with a hard cap of three that forces a final answer, parallel tool execution,
and only the final round streamed. The day's budget reserves the per-call
ceiling times the round count and reconciles the summed tokens; the atomic
reserve and reconcile primitives are unchanged. All inbound gates (rate limit,
refusal, replay scan, consent, budget) stay before the loop.
The legacy snapshot path stays alive as the no-tools fallback: tool mode is on
only when every provider in the chain supports tool-calling, so a chain that
includes a local model falls back to the prior behaviour verbatim. The tool
trace is persisted onto the assistant message provenance.
The Coach snapshot builder returns the structured, post-degrade record it
serialises to JSON alongside the JSON string, so a single domain block can be
read out without re-parsing. Add an optional tool-call trace to the provenance
envelope (which retrieval tools ran + whether each found data) and mirror it in
the OpenAPI provenance schema.
Groundwork for on-demand retrieval: the retrieval tools slice one section out
of this record, reusing the builder's gates and I/O verbatim.
Replace the CompletionParams { systemPrompt, userPrompt } string pair with a
structured { system, messages[] } request across all five AI clients. One
refactor unlocks three things: provider prompt-caching (Anthropic cache_control
on the stable system prefix, OpenAI/Codex automatic prefix cache, with
cachedInputTokens surfaced for observability), the tool-calling wire surface,
and clean multi-turn.
Each client maps the new shape to its own wire and degrades unsupported
features silently: the local OpenAI-compatible server never receives a tools
field (most reject unknown keys), and prompt-caching is best-effort everywhere.
CompletionResult gains toolCalls, finishReason and cachedInputTokens; the tool
types and per-provider mapping ship now, the tool-use loop does not.
Vision inputs fold into the user turn's content parts via singleUserTurn, with
each provider's pre-existing block ordering preserved. The corrective-retry
paths append to the last user message rather than a flat string. The Coach
keeps assembling one user turn so its snapshot-once and grounding behaviour is
unchanged. No wire schema visible to API consumers changed.
Behaviour is unchanged across every surface; the full suite stays green.
The 0–14d raw band folded same-day readings on a UTC calendar key, so a
late-evening reading slipped onto the wrong day for non-UTC users and
across DST transitions. Key it on userDayKey(tz) instead and make the
timezone a required option so a caller cannot silently fall back to UTC.
The coarser DAY/WEEK/MONTH/YEAR bands stay UTC-anchored on purpose: the
rollup writer and the live aggregator both bucket on the UTC calendar
day and are parity-locked for the read-swap fallback. Document the split
and thread the resolved user timezone from the coach snapshot caller.
Add a labs block to the coach snapshot so the assistant can answer
questions about the user's lab results. Each entry is the most recent
reading per biomarker over the last 12 months (capped), carrying the
resolved name, value, unit, reference range, in/out-of-range verdict,
and measured date. Qualitative readings (e.g. negativ) are included
with a neutral range status. Values resolve server-side via the same
helpers the labs API uses, and the system prompt instructs the model to
quote them verbatim and never invent a lab number, so the block stays
grounded. The free-text qualitative result is sanitised before it can
reach the prompt; the encrypted note is never read.
The safety-contract matrix parity test checks rule bodies are non-empty;
this asserts the resolved getCoachSystemPrompt(locale) string carries the
grounding marker and the GLP-1 dose-safety contract for all six UI
locales, closing the gap where a non-EN/DE locale could silently fall
back to a weaker EN-plus-footer body.
The single multi-type window read had no take bound, so a lastYear /
allTime scope over a high-frequency type (PULSE / glucose are
200k-row-class) loaded a year of rows just to fold them into the ~21
daily + ~10 weekly buckets the prompt shows. Read newest-first with a
6000-row cap and restore ascending order; the coarse MONTH/YEAR tail
still comes from the rollup tier, so deep history survives.
detectRefusal guards only the inbound user message; the model's reply
was streamed verbatim. A dose-prescription or a fabricated clinical risk
score that slipped past the system-prompt contracts reached the user
unfiltered.
Add a deterministic outbound screen on the assembled reply (which the
route already buffers before streaming): tight dose-change patterns plus
risk-score patterns, EN/DE. On a trip the turn is replaced with a calm,
grounded fallback and any reminder suggestion / key-value provenance from
the discarded reply is dropped.
Carry an additive usage envelope on the SSE done frame
({ totalTokens, model }) and persist tokensUsed + model onto
coach_messages (migration 0182, nullable, no backfill) so the
per-message token footer survives a conversation reload. Add an
optional reasoning frame the client can render inside the thinking
disclosure when a reasoning-capable provider emits a summary; providers
without reasoning emit none. Extend CoachMessageDTO and regenerate the
OpenAPI contract. Server-authoritative: clients display the count, never
recompute it.
One-shot prettier pass so the new format:check gate starts from a clean
baseline. Formatting only — no behaviour change. Generator-owned and
vendored artifacts stay untouched via .prettierignore.
The daily token cap was read before the provider call and recorded only
after a successful reply, so concurrent requests could all read
`spent < cap` and all hit the provider, and an empty/sentinel reply
recorded zero spend despite burning upstream tokens.
Reserve the budget atomically before the call: a single upsert-increment
returns the new total, so concurrent requests serialise on the row and
cannot both pass the cap; over-cap refunds the reservation and surfaces
`coach.budget.exceeded` as an SSE error frame. Reconcile the reservation
to the actual token count immediately after the provider returns —
before any empty/sentinel short-circuit — so burned tokens are always
recorded. An integration test proves two concurrent requests cannot both
exceed the cap and that an empty reply still records its cost.
The grounding, tone/banned-opener, GLP-1 dose-safety, metric-identifier
ban and forbidden-filler rules were independently encoded across the
comprehensive, status, Coach and narrative prompts and had drifted. Move
the canonical wording into `prompts/shared-contracts.ts` (locale-keyed
fragments) and compose each surface from it.
Where the surfaces had drifted, adopt the stricter wording: most notably
the GLP-1 dose-prescription contract — previously present only in the
comprehensive and Coach prompts — now reaches the status cards and the
period narrative too. A cross-surface coverage test asserts every
enforced fragment appears verbatim in every surface, in de and en.
Marc-André Bombeck