The outbound screen let a halving with no number through in every
language but Dutch ("Reduce your dose by half.", "Reduza a dose pela
metade.", "Tome meia dose."), let a bare change imperative through
("Zmniejsz dawkę.", "복용량을 늘리세요.") and did not know the Polish
spelling of the generic stems, so "Odstaw ramipryl." passed. Each
language's bank now carries a halving and fraction shape and a bare
change shape on a medication noun, and the stems include the Polish
-pryl, -statyn and -glutyd.
The continuation exemption trusted "as before" on any surface that
passed no schedule, so "Take 40 mg as before." went out to people who
had never taken 40 mg. The screen now takes the dose plan (each active
medication with its dose) instead of a bare list of numbers, and
exempts a stated amount only when it is the planned dose of the
medication the sentence names. Without a plan no stated amount is
exempt.
Ollama serves thinking models such as Gemma 4 on /v1/chat/completions with
the thought in a separate `reasoning` field. A long briefing can come back
with empty `content`: the thought used the whole token budget, or the
answer itself was parsed into `reasoning`. The background briefing always
asks to reason, so every briefing on such a model ended in the empty-reply
sentinel, which the chain logged as status 0 / cause "network" in the same
second the server logged 200, and the read path showed as a timeout.
The Local client and the OpenAI-compatible gateway now treat a reply with
reasoning and no answer as a lesson: the endpoint and model are remembered
in the reasoning dialect cache as answering only unreasoned, the call is
asked once more with reasoning_effort "none", and later calls ask that way
until the lesson expires. A test drives the real client, safeFetch and the
pinned dispatcher against a loopback server that answers the way Ollama
serves Gemma 4, and reproduces the reported log line without the fix.
Failed hops now carry what the call ran into: `failure` (timeout, aborted,
refused, reset, socket, parse, empty_response, reasoning_only, ...) and a
code-shaped `code` on ai.chain.link_failed, closed vocabulary only. The
cause only says "timeout" for a deadline, and the briefing failure marker
records the kind so the empty state suggests a longer timeout only when
time ran out.
Refs #1267
The health-status read scanned five vitals over 90 days for a dated,
sustained level shift. It now scans every vital with a stable daily level
(adding RMSSD, diastolic pressure, pulse, body fat, respiratory rate,
oxygen saturation, walking heart rate and VO2 max) over 180 days, reads one
value per day per metric, leaves a metric unread when its module is off,
and reports how many days stand on each side of the break. The Coach keeps
its five-metric, 90-day scan unchanged.
The discovery matrix gains the families people log but could not test:
water, caffeine and up to four varying micronutrients from the nutrient
log, alcohol days from the mood tag, workout minutes (one per session) and
the day's training impulse, exercise, stand and mindful minutes, State of
Mind as an outcome, and the overnight room climate against that night's
sleep only. Every channel runs through the same seasonal adjustment,
effective-n p-value, Benjamini-Hochberg control and effect-size floor, is
read only while its module is on, and drops out below the paired-day floor.
State of Mind, mindful minutes and alcohol days are statistics for the
person: the Coach, the period narrative and the assessment cards drop every
pair that touches them, after the shared scan so the remaining pairs keep
the statistics the correlations page shows. A guard walks the import
closure of the analysis modules and refuses any path to a provider.
Every write has gone to insights_cached_text_encrypted since 0411, and
the boot-time free-text backfill sealed the rows written before it.
Migration 0430 drops insights_cached_text and the backfill index that
served it. A host that skipped the intervening releases loses only a
cache of model output; the next warm writes a fresh, sealed payload.
The cipher module now reads and writes the sealed column only, the
backfill job loses its insights step, and consent withdrawal, the wipe
plan, the backup plan and the session projection follow the schema.
Backups never carried the column, and restore ignores unknown account
settings; a round-trip test now proves an old backup naming the field
still restores cleanly. The access guard pins exact counts for where
the sealed name may appear. Fixtures, the demo seed and the demo
migration script seed the sealed column instead of plaintext.
The shared write bucket threw on any SQL error, so a lock on rate_limits
answered every authenticated write with 500. It is a loop cutter, not an
access check: when the bucket cannot be read the write now goes ahead with
a warn signal. The login and other auth-surface limits call the limiter
directly and still fail closed. requireAuth now passes readOverPost to the
meter, the draft-only document assist declares it, and the three chat
routes, which meter their turns with buckets of their own, declare that
instead of being counted twice.
169 of 364 mutating handlers met no rate limit, every PATCH and DELETE on
an [id] among them. The shared per-account write bucket (300 a minute,
keyed on the acting account) was called by hand in a dozen creates; it is
now charged by the identity resolvers for any POST, PUT, PATCH or DELETE,
once per request, so a route is metered by resolving its caller.
The bulk ingest routes the native client syncs through in bursts
(measurements, workouts, intake, mood and cycle bulk, nutrients, the ECG
sweep) declare writeBucket: "own" and keep their dedicated ceilings, so a
large HealthKit backfill cannot lock the person out of logging a dose. A
read-over-POST stays free.
write-rate-limit-guard fails on any mutating route that neither resolves
its caller through a metering resolver nor calls a limiter, unless it is
on a short reasoned exception list; an opt-out must be on its own reviewed
list and call its own limiter.
listPage (src/lib/http/list-query.ts) reads one row past the limit and
cuts it back, on top of the existing list-cap signal rather than beside
it. Every capped list now gets its page there, and every read passed to
it orders by something ending in the row id, so rows that share a sort
value cannot repeat or vanish at a page boundary.
- practitioners, vaccinations and encounters were cut at their take
with no signal; they now answer meta.truncated and accept offset, and
their web readers walk the pages instead of stopping at the first
- symptoms/events reported hasMore on a page that was exactly full; it
is now true only when a row past the page exists
- the document list's hand-rolled limit + 1 goes through the helper
- security-activity answers an out-of-range limit with the 422 its
published schema already promised instead of clamping it
list-page-guard binds every apiListSuccess route to listPage, forbids
hand-cut lists and length-derived hasMore, and requires the id
tiebreaker in every listPage read.
Four per-metric status snapshots read mood context without the tombstone
filter, and the blood-pressure snapshot read measurements the same way, so
a deleted entry kept shaping the text written about it. The breathing
screen and the HRV widget probe had the same gap.
One reader now serves the shared mood context, and a guard derives the
soft-deleting models from the schema and fails on any unfiltered read in
the analysis, insight and model-input directories.
Output of `node scripts/strip-planning-tags.mjs --write`: 3,157 comment
lines in 1,308 files lose the work-item identifier at their head. Release
versions and issue references stay; nothing outside comment text
changes, which the script checks per file against a comment-free
printout. healthlog/no-planning-tags is enabled at error level with this
commit, so the tree it lands on is clean.
users.insights_cached_text held the model's briefing, readings and
recommendations about the account in the clear. Migration 0411 adds
insights_cached_text_encrypted (AES-256-GCM, binary layout, purpose-bound
label) and a partial discovery index; the free-text encryption backfill
seals the existing rows one at a time, fail-closed, carrying updatedAt and
guarded on the value it read. Every reader, writer and clear goes through
cached-text-cipher.ts, and a guard keeps the column names there. The
column joins the encrypted-column registry and the key-rotation script,
stays out of backups as operational, and is reset by the data wipe.
A list that stops at its own take and says nothing lets the last row read
as the oldest there is. Capped lists now fetch one row past the limit and
answer meta.truncated and meta.limit on the success envelope, so data keeps
its shape. Applied to the capped lists without a total count: rhythm
events, ECG recordings, sharing grants and activity, illness episodes, side
effects, personal records, security activity, cycles, family history,
allergies and mental-health assessments. The activity feed no longer
reports a list that is exactly full as cut.
The /insights envelope (weight x BP, mood x BP, mood x weight, mood x
pulse, BP medication continuity) and the AI feature payload computed
their coefficients with an ungated Pearson that banded |r| above five
pairs, so five days of noise sloping the right way read as a strong
correlation there while every other surface refused the same points.
Both now use `significantPearsonCorrelation` (n >= 20 and p < 0.05,
through the engine's exact p-value), and the ungated function is gone.
The allowlist guard that tracked its remaining callers becomes a guard
that there is one Pearson implementation.
The latest canonical reading per type ran as `DISTINCT ON (type)` over
every live row of the account in five places (summaries slice warm and
cold path, comprehensive aggregator warm and cold path, targets). With no
skip scan in Postgres 16 that sorted the whole account to keep one row
per type. `readLatestCanonical` drives the read from the type list
instead: one index probe for each type's latest instant, then a closed
one-day window ranked by the source ladder. The closed window also keeps
the plan under `jit_above_cost`. The targets read reuses `readLatestEver`
with a floor.
The summaries slice was built inside three cache cells (the tile-strip
route, the thick analytics body, the dashboard snapshot). It now has one
cell, read through `readSummariesSlice`; composing callers read it
fresh-only so a refresh after a write cannot cache the pre-write slice
again, and the snapshot folds the water tile into a copy instead of the
cached object.
Covers the day section, the model projection without the travel country and
the two correlation channels; the backup and correlation mocks learn the new
delegate. Adds the operator page for DAWARICH_PRIVATE_ORIGINS and regenerates
the API contract.
Connect, status, places, settings, test, sync and disconnect replace their
stubs. A nightly job (03:20) and the connect backfill read the timeline,
visits and residency endpoints through safeFetch and store one row of daily
values per local day: distances, places visited, minutes away from the home
place and the sealed country with its travel-day verdict. A private instance
needs its exact origin in DAWARICH_PRIVATE_ORIGINS.
The connection joins the shared ledger and the connection health model,
watched for errors only. The day carries a movement block, the model
projection drops the travel country, the two movement channels join the
correlation matrix under the environment module, travel days become trip
suggestions, and the days travel in the environment backup section.
The level now sets, per record: the status-card regeneration gap after a
sync (frugal 24 h, standard the operator's gap, full 1 h) and after a hand
entry (frugal 6 h, otherwise 1 h); the status snapshot cap (10 000 compact
characters on frugal, 14 000 above); background reasoning for the briefing
and the period narratives (off on frugal); the daily briefing's form (a
short form on frugal that keeps every field and at most one
recommendation, the full contract on standard and full); and whether the
automatic document read may act (frugal reads on request only, the stored
switch is kept).
A deterministic simulation of a watch-synced day through the invalidator
counts 122 ingest-driven status generations on full (the v1.42 behaviour),
25 on standard and 2 on frugal, plus the nightly warm on each.
Refs #1217
The opener hint, the first-name clause, resolved follow-ups, the table rules
and the final-round line used to be written into the system prompt, so the
cached prefix (tools plus system) was rewritten on every turn and again on
the forced final round. They travel on the user side now: the per-turn lines
in the user message, the table rules as a note after the round that produced
the first table, the final-round reason as the last user line. The system
prompt depends on the locale, the Coach preferences and the about-me text
only.
Refs #1217
The status cards, the briefing, the derived assessments and the Coach
snapshot drop the two-space indentation from the JSON they hand the model.
The content is unchanged (the hash gate hashes the object, so stored hashes
stay valid). MAX_SNAPSHOT_CHARS is now measured on the compact string: 14 000
for the status cards and the Coach, which every snapshot that fitted the old
24 000 pretty-printed characters still fits, so the model reads the same data
for fewer tokens.
Refs #1217
A converted reading's bounds move with its value, but the printed range text
stays in the unit the report used, and every reader preferred that text: the
list showed "Report range 3.9 to 5.5" beside 100.9 mg/dL, and the doctor
report and the FHIR referenceRange.text did the same. The text is still
stored verbatim. The range resolver now takes the reading's sourceUnit (as a
required field, so no reader can forget to select it) and, for a converted
reading, leaves the window's text empty so every surface states the converted
bounds in the marker's unit, and hands the printed text over separately with
its unit. The lab DTO carries it as printedReferenceText, and the note beside
a reading shows it as "as printed in mmol/L: ...".
The health score had no history anywhere, readiness and the sleep score
showed a fixed sparkline, and only the stored nightly scores offered the
range tabs. Every score page now charts its course with the same
7 / 30 / 90 / All tabs, remembered per page, and each point opens its day.
- GET /api/insights/score-history?score=&days= serves the daily course of
the health score (its stored day rows), readiness (the nightly blend,
filed on its wake day) and the sleep score (each night against the
trailing window ending on it), so a point equals what the day view shows
for that day. MANAGE record read, module-gated for readiness and sleep,
days 1 to 3650, additive.
- The health score's line breaks where its recipe changed (configuration
or algorithm version); the usual range is formed from the newest side
of the last seam.
- The range tabs move into one shared component used by every chart;
recovery, stress and strain keep their measurement chart, now with the
range remembered and no overlay dropdown.
- New page /insights/health-score, linked from the score panel and from
the day view's score tile.
A dash between two parts of a sentence ("{what} — tomorrow", "Could not
record {name} — try again", "End {name} — …") stands in for a comma, a
colon or a full stop the sentence should say. Every such string in the
seven bundles now says it in plain words. A range keeps its unspaced en
dash ("70–100 mmHg", "{start}–{end}"); that is how a range is written,
not a pause, and the spaced range forms became unspaced ones.
Labels the components put together themselves follow the same rule:
aria-labels read "Move up: Ramipril", a medication and its dose read
"Ramipril 5 mg", page titles use "|", Coach prefills are two sentences,
and the API's user-facing error messages lost their dashes too.
The bundle guard now refuses " — " and " – " beside " · ", with an empty
allowlist, and a sibling guard walks src/components and src/app for a
string or template literal that joins its parts with a spaced dash.
New conversation leaves the conversations panel's header and floats in
the conversation itself: a round 56 px button with a plus, the floating
Coach button's face, named "New conversation". It sits in the composer's
row, so it always belongs to the chat column, never to the list or a
docked day beside it. Where the column is wide it takes the bottom right
corner beside the composer; where the composer fills the width (a phone,
the drawer, a narrow column) it rises above the composer's edge. The
drawer shows it once there is a conversation to leave, in place of the
plus in its header. An empty conversation shows nothing either way.
The docked conversations panel folds like the day: its header opens with
a fold control, and folded it keeps a 48 px edge whose button opens it
again (aria-expanded, focus moved between the two, Escape inside folds
it). The top bar's toggle remains for the sheet below 1280 px. On /coach
the day docks right of the panel. Below 1600 px only one of the two is
open: a day opening folds the list without touching the remembered
choice, and opening the list folds the day to its edge. From 1600 px both
stay open. Switching or starting a conversation keeps an open day in the
URL.
The empty conversation drops everything under the composer: the daily
seeded opener and the scope pill are gone, and with them the only reader
of GET /api/insights/coach/seeded-question and of the
coach.ambientSuggestions preference. The route, its OpenAPI entry, the
preference and its Settings switch go too.
Since correlations are tested on seasonally adjusted residuals with an
effective sample size, an account with history sees far fewer
relationships, which reads as a defect without a word about it.
The correlations response now carries findingsBeforeSeasonalAdjustment:
true when the record kept a discovery pattern first stored before the
change, one indexed lookup on the pattern store. An account created since
can never qualify. The mood "What stands out" card shows a short note
for those records, closable, remembered per viewer like the AI setup
hint. Copy in all seven locales; OpenAPI regenerated.
The nightly briefing warm completed every night and read "all-failed",
because one run event held every account's annotations and the last
account won. Each candidate now runs in its own wide event and the run
reports every outcome through `reportJobRun`, so ok, dead sign-in,
withheld and failed accounts show side by side.
A dead Codex refresh token threw a bare error with no status, so the chain
read it as a network failure: the health ledger recorded hard_failed, no
"reconnect" hint appeared, and every night tried the same token. The
refresh now throws `CodexRefreshError` with a 401 for a dead grant
(invalid_grant / invalid_refresh_token on a 400 included), as do a missing
or undecryptable stored credential. The ledger records auth_failed, the
chain marks the primary credential expired, and the nightly pass reports
the account as auth_failed without queuing a retry.
A briefing the outbound screen withheld (a causal claim) failed outright,
and the 45-minute provider retry then asked the same question eight times
a night. The generator now makes one bounded repair that names the broken
rule and keeps the repaired payload only if it passes the same parse,
grounding and screen; a withheld outcome queues no retry.
A provider that keeps failing is paused: from five consecutive hard
failures the skip window starts at 30 minutes and doubles per further
failure, capped at a day. A paused provider is walked last and tried again
when the window lifts; one success resumes it. The operator card shows how
many accounts it is paused for and until when. After a 5xx from a
non-canonical OpenAI-compatible endpoint the client asks the free model
listing, and a configured model the endpoint no longer offers is named in
the error, on the hop line and on the card.
Also:
- The tiered context's live day band read `measured_at AT TIME ZONE tz`,
which treats the stored UTC as local time; a far-from-UTC reader's
evening readings landed on the next day. It converts from UTC first now.
- Google Health intraday heart rate re-reads two hours per sync instead of
a day; daily summaries and sleep keep the day (#1023).
- The first filtered daily read that answers 400 and triggers the
snake_case retry is recorded as a note, not an error (#1023).
- The single derived-score read goes through the derived cache with
stale-while-revalidate, like the batch beside it.
Own-plan turns get 165k tokens, 200 s and 16 rounds, operator turns 55k,
120 s and 8; the final-answer reserve, the concurrent-turn slot and the
step cap move with them.
Codex requests carry a per-conversation prompt_cache_key and a stable
session id, api.openai.com gets the same key, Anthropic usage counts
cache reads and writes in the gross, and local servers report cached
prompt tokens, so the turn budget discounts cached input on every
provider.
remember_fact and propose_plan wrote their rows during the tool round. An
answer that was then blocked, failed or abandoned left the fact in memory
without its note and Undo, and left the plan listed, activatable and fed
into later turns and the briefing. Both are now staged under an id minted
up front and written in the transaction that stores the assistant message,
only when its provenance carries them. A background proposal counts as
offered only once an answer carrying it is stored.
Plan text is model text: ifCue, thenAction and target pass the outbound
dose and risk screen in the person's locale, with their scheduled doses and
medication names, and the injection screen; a trip declines as "unsafe".
The trail route serves recalled facts and the proposal's fact to the owner
whatever the AI capability (only the model-written entries are gated), and
withholds medication facts while the medications module is off.
Accepting a fact proposal is decided under a per-message advisory lock, so
concurrent taps write one fact. Editing a fact re-reads its wording: text
naming a medication moves to "medication", other health wording moves a
non-health fact to "condition", and a health fact the person edited counts
as confirmed.
The usual-range strip on a metric page placed the latest single reading
against a band built from whole-day means. For glucose the day mean climbs
with every meal, so at nine in the morning a fasting reading sat under a
band of whole days and the strip said "below your range" on an ordinary
morning; the MCP baseline read reported the same.
Glucose is now placed by its day mean. A day still in progress, in the
reader's own zone, is held against the earlier days cut at the local time
of today's latest reading, the comparison the health status already makes;
with fewer than seven earlier days that have readings in those hours there
is no verdict. The baseline carries `basis: "sameHours"` on the coach-read
route and the MCP read, and the strip names the range as the one for this
time of day, in all seven locales. Other metrics keep their placement.
The chat route hands the turn the level resolveCoachTurnReasoningLevel
returns: the stored Coach preference, absent read as medium, after the
operator's switch and cap. The turn still applies the operator-paid
ceiling once its chain is known. The controls loader no longer throws
when the request cache cannot serve it; it fails closed to off. The
admin guard pins the route to the resolver.
Build the memory block for the providers the turn sends to, send it on
both paths, keep the facts it recalled in the owner's trail, and offer a
health fact the background found as the answer's one note. Every turn
enqueues the memory refresh; the job's debounce and singleton key decide
when it runs. A remembered fact points back to the stored user message.
Unit tests for the turn budget and the no-progress brake, the budgeted
loop (every stop reason, the round-by-round ledger, repeated calls,
questions, comparisons, reasoning and provider state), the trail recorder
and its screen, clarifications through the tool and the brake, the
comparison table and chart, the trail column and its owner-only read, the
reply frames, and the turn's frames end to end through the route. An
integration test proves the trail against a real database. Eval
scenarios cover a multi-round why question, a period comparison with its
chart, a question asked through the tool, and keep the charts the
existing table scenarios already showed.
Take the effective reasoning level from the route rather than resolving
it in the turn; the turn only caps it when the operator pays.
Give the Coach one statement of how an answer is shaped, in the German and
English bodies and on every other locale: the first sentence answers, one
carrying number with its unit and window, the person's own baseline, an
assumption in one line, at most one next step and no habitual closing
question, 40 to 120 words, and a closed list of filler it never writes.
- Brief is 30 to 70 words in one paragraph, detailed up to 220 in three.
- Motivational-interviewing moves only when the person opens a change topic,
at most one, never as a closing question; the confidence ruler only on an
explicit plan request.
- The few-shot examples are rewritten to model the voice.
- Result tables: show the numbers whenever they carry the answer and span
more than one value, a table for exact values or up to twelve like rows,
one figure alone only for a moment.
- Follow-up chips are the answer's only "what next".
- A voice grader for the eval (length, filler, emoji, next steps, closing
question, warm-up opening) plugs into the golden set as criteria; five
voice cases in German and English.
- A guard holds both prompt bodies to the rules and every example answer to
the grader, each check broken once on purpose.
Marc-André Bombeck