What a person told the Coach never reached a tool-mode turn: facts and plans
rode the snapshot's memory block, which tool mode does not send, and the
extraction ran only past twenty turns and failed to parse on every
non-Codex chain, because the status chain runs in JSON-object mode while the
prompt asked for a bare array.
- Memory block: WHAT YOU KNOW ABOUT THIS PERSON, built per turn behind the
wire egress check (nothing is read on a refusal), fenced as data, at most
1,500 characters, never trimmed before the inventory. Confirmed facts
first, then the newest; active plans with a server-computed progress line;
reminders the person asked for. Medication facts stay out while the
medications module is off. Building it stamps last_used_at without moving
updated_at. buildCoachToolRequest and buildCoachProviderPrompts take it as
an optional memoryBlock; the snapshot memory keeps only the narrative and
trend recall.
- remember_fact: preferences, goals and context are saved at once
(source coach, source_message_id set); anything that names a condition,
limitation or medication, by category or by wording, is only proposed.
A fact must come from the person's current message; one note per answer.
- The person's answer reads the proposed fact from the encrypted trail on
the assistant message, never from the request.
- propose_plan writes a proposed plan; the tap activates it with its review
window counted from the tap. One per answer, three open at most.
- Progress: one sentence per active plan from the person's own day series
and units, also offered to the briefing (two lines at most, behind the
briefing capability). The plan review and the progress line read a step
plan as daily totals instead of per-sample means.
- Background extraction asks for {"facts":[]} / {"plans":[]} and reads
either shape, runs once a conversation has been quiet for 30 minutes, and
stores health facts as proposals offered once at the next turn.
- Unanswered plan and fact proposals lapse after 14 days in the reminder
sweep.
- Memory list API: POST saves one of the caller's own messages (the
remember button), PATCH edits a fact's wording, GET reports source, origin
and last use and never lists a waiting proposal.
- A structural guard keeps the egress check ahead of every read and the
callers of both builders to the turn and the briefing.
Add the shared contract the reasoning release builds on, with no change
to how a turn runs today.
- Schema: operator reasoning controls on app_settings (0376), the
encrypted trail text on coach_messages (0377), and source,
last_used_at and source_message_id on coach_facts (0378).
- Provider contract: CompletionParams.reasoning and onReasoning,
CompletionResult.reasoning and providerState, AiMessage.providerState.
- Reasoning levels, thinking budgets and the background-job table in
src/lib/ai/reasoning/levels.ts; coachPrefsJson.reasoning, absent
meaning medium, salvaged field by field like every other pref.
- Stream: activity, memoryNote and planProposal frames, done.stop and
done.withheldResults, result.interim, a compare chart kind, eight
table refs, three more clarification kinds with an assumed choice,
the change_assumption chip, and assumptions, activity, stop,
memoryNote and planProposal on the provenance. Zod mirror, type
parity and the native-client key rule cover every new frame.
- Chat request: memoryDecision and planDecision, parsed and not yet
acted on.
- Stub contracts for the memory functions and the activity recorder,
inert until their implementations land.
- Backup, restore and rotation carry the new columns; a fact's message
reference is resolved against the restored messages like its
conversation reference.
- Message keys for the trail, memory, plans, assumptions, replies,
interim tables, comparison charts and reasoning settings in all seven
locales, each reachable from a catalog and checked both ways.
The panel groups conversations by the browser's local day, so the e2e
fixture's "an hour ago" landed in yesterday when CI ran just after
midnight. The turn-steps test reads its markup by splitting rather than
with a replace CodeQL reads as an incomplete sanitiser.
The today check now reads briefingGeneratedAt from the cached payload
and treats a payload without it as not from today. The integration
and e2e fixtures that seed a fresh briefing write it, as every real
generation does.
Below md the lines under the 80 px dial (the score delta, how long it
has held, what it rests on) were capped at the dial's width. In German
they broke into three to five rows each, and beside a short lead that
column set the height of the read row and left a gap above the facts.
The score column now dissolves into the grid below md: the dial stays
the top-right cell beside the lead, and its meta lines take a
full-width row right after the read row, flush with the dial's edge,
one line each. The facts move one row down. From md up the column is
the same flex column as before (desktop captures are pixel-identical).
The Today overview e2e renders German at 390 and 360 px with a short
and a long lead and asserts the dial size, one row per meta line, the
lines flush with the dial, the meta row right after the read row, and
exactly one row gap above the facts.
With signals of the day resting on today's readings only, the briefing
had to be written after those readings, and it was only written at the
04:30 Berlin warm, the post-sleep morning refresh, or a page open on a
day-old cache. A morning blood pressure or weigh-in never became a
signal, and since an earlier day's briefing is no longer served, a
reader in Los Angeles or Seoul had none for most of the day: the Berlin
warm lands the previous evening there, or at noon.
The snapshot read every surface goes through (dashboard, iOS, digest,
morning push) now asks for a briefing for today at two moments: the
first read of the reader's local day with no briefing written today,
and the first read after a signal metric's first reading of the day
that the briefing has not read. Each reason is claimed once per user
per local day in a Postgres bucket and remembered in-process, so later
reads cost no query; no write path needs a hook.
A claim enqueues one today warm: immediately for a new day, after a
five-minute window for readings, so a burst (weigh-in, then blood
pressure) folds into one job. The warm writes the comprehensive
briefing only, past the one-hour freshness gate but under the failure
backoff, the daily forced-warm cap (two per 24 h) and the content-hash
gate.
Worst case per user per local day: one enqueue per reason, at most
eight (the day plus seven signal reading types, measured over a day of
reads every two minutes); at most two forced comprehensive attempts
under the existing cap, each at most three completions (the answer, a
JSON retry, a grounding retry), so six provider calls at the ceiling.
A warm whose data has not changed costs at most the existing
once-a-day paragraph re-roll, inside the same cap.
The SDK renders every tool's Zod shape with a fixed draft-07 target and
stamps "$schema": "http://json-schema.org/draft-07/schema#" on each
inputSchema and outputSchema. Clients that validate with a 2020-12-only
validator refuse that dialect when the tool list arrives, so every tool
was unusable there (#1170).
Wrap the tools/list handler once, after the last tool is registered, and
re-render each advertised schema from the same Zod shape with the
draft-2020-12 target, without a $schema key. A schema with no $schema is
2020-12 under the MCP spec and still compiles in a draft-07 client. All
other fields of a tool entry pass through unchanged, and the SDK keeps
validating structuredContent against the Zod shape, so tool results are
unaffected. For the current surface the 54 rendered schemas differ from
the previous ones only by the dropped key.
The new guard reads tools/list through a real client for a write-scoped
session, rejects any draft-07 marker or construct, and compiles every
schema with Ajv's 2020-12 build (ajv added as a dev dependency at the
version the SDK already resolves).
The summary of a condensed block was taken over the block's mix of daily
values and weekly means: over a year the newest fourteen days weighed as
much as fourteen weeks, min and max were extremes of means, and change was
one day minus a weekly mean.
Each series block now hands the budget pass its daily values over the
whole read, computed only when the block is condensed. The summary states
the first and last day, the lowest and highest day, the mean of the days,
and the change from the first calendar week to the latest one. A block
without daily values gets no summary rather than one of what is left. The
coverage note says how the summary is taken.
Only the digest applied briefingForToday. The dashboard snapshot (the
web dashboard, its server prefetch and the iOS cold-launch read) and
the /insights advisor payload still served an earlier day's text and
the raw model deltas ("+33.72 bpm").
The snapshot read now resolves the briefing after the cache, beside
the capability decision, from the payload's generation moment and the
snapshot's own newest reading per type; a withheld briefing reads
"preparing", never "ready" with nothing in it. The digest takes the
resolved briefing from that read on its own clock instead of resolving
again. The advisor route resolves dailyBriefing on every branch it
returns, reading the newest reading per signal type in one grouped
query only when a reading-backed signal is present; text generated on
an earlier day no longer counts as fresh on the read GET (so the warm
is enqueued) and no longer short-circuits a provider-backed POST.
The OpenAPI descriptions of both advisor responses say so.
The check that keeps an earlier day's briefing off the Today surfaces
read the cache row's insightsCachedAt. Every warm moves that column,
including the unchanged-hash short-circuit that keeps yesterday's text
and only refreshes the stamp, so text written the day before passed as
today's after the nightly warm.
The cached payload now carries briefingGeneratedAt, written only where
new text is written: a full generation in the warm and in the
on-demand route, and the daily paragraph re-roll. The stamp refresh
leaves it untouched. briefingForToday reads that moment, and a payload
without it counts as not from today, so the hero falls back to its own
lead until the next generation.
The condensing step read every block as a value series: daily rows under
value, weekly rows under mean and count. Sleep nights carry minutes, so a
condensed sleep summary ended on the last weekly mean weeks before the
newest night; adherence rates fold by doses, not readings; glucose keeps a
series per measurement context; workouts are a sessions list.
Condensing now knows each block's shape: sleep reads minutes, adherence
folds rate weighted by total doses, glucose condenses every context, and a
workouts block keeps its newest sessions beside the whole-window rollup.
formatSignalDelta matched only the locale's decimal mark, so an
English-formatted figure inside German text was cut at the comma
("+1,234.5 kg" read "+1,2.5 kg"), and a difference that rounded to
nothing lost its sign ("+0.04 kg" read "0.0 kg").
A figure with a decimal or grouping mark is now read as a whole: with
a point and a comma the later one is the decimal, thin spaces and
repeated marks are grouping, and a single mark is grouping only when
it is the locale's grouping mark before exactly three digits. The
figure is re-rounded to the metric's precision and re-written with the
locale's own marks. A non-zero difference that rounds to nothing reads
"±0" at that precision. Plain integers stay as written.
Glucose joins the reading types, its precision following the unit the
delta is written in (whole mg/dL, one decimal of mmol/L). The map is
keyed by string, because the published sourceMetric enum has no
glucose value yet and widening it would change the iOS contract.
Tests cover all seven locales.
The Coach's figure check ran before the MCP branch, so an MCP client was
told retrieval_failed for a block it used to receive, and the plain-text
renderer then said no data was on file. Any single-source build also
condensed its own block, which reached MCP through the source snapshot's
fallback to the full builder.
The figure check now runs after the MCP branch, and condensing is an
explicit option of the snapshot build that only the Coach tools set (it is
part of the cache key). MCP output is what it was before.
A block trimmed to its unit and freshness stamp passed the figure check,
because asOf.daysAgo is a number. The check now skips the asOf and scope
fields, so that shape answers retrieval_failed like any other block
without a reading.
Below md the score ring sat centred under the Today facts as a 168 px
dial, where the bottom navigation covered it. The hero is now one
grid with two arrangements: on a phone the ring is a compact 80 px
dial at the top right beside the lead and signal line, flush with the
card's trailing edge, and the Today facts run the full card width
under both; from md up the ring spans both rows of the trailing
column exactly as before (desktop captures are pixel-identical).
Without a lead the facts take the leading cell, so the ring never
sits beside an empty row.
ScoreRing gains a `responsive` size (80 px, md:168 px) whose label is
left out below md; the ring keeps its aria-label and the surrounding
link names it. The meta lines under the phone dial are capped to its
width, so the lead keeps at least 180 px beside it at 360 px.
Each fact now stacks its muted label above its foreground value on
the reading edge at every width, instead of a label-left /
value-right row that broke long values into a ragged right-aligned
block. Rows stay at the 44 px tap floor below md.
The Today overview e2e covers 1440, 390 and 360 px in both themes
for an AI lead, a long AI lead, an AI-off day and a quiet day, and
asserts the dial size, its place beside the lead, symmetric insets,
full-width facts and stacked 44 px rows on a phone.
Inside a chat turn every tool read lands on the turn's shared full-source
snapshot when its window matches. On an account with many synced series
that build passes the prompt budget, and the budget pass collapses blocks
to a bare unit and then to an `{ omitted }` marker. get_metric_series
sliced the marker out and answered `present: true` with it, so the model
was told the readings exist and was given no figure. The same held for
sleep, glucose, compliance and workouts.
- The snapshot result names the blocks the budget pass cut. A tool whose
block was cut in the shared build reads it again from its own source at
the same window.
- A read scoped to one source condenses its own blocks last and never
swaps them for the marker: it keeps a summary (first, last, min, max,
mean, change) taken before anything is cut, the newest daily values,
weekly means folded into monthly means, and names each cut.
- get_metric_series never answers present with a block holding no figure;
that read reports retrieval_failed instead.
- A chat series read carries the readings it counted, so the step shows
them, and on a long or condensed window a note on what it covers. The
tool description and a new tool-mode rule tell the model to call
get_metric_table itself for longer history instead of asking.
- The screen-reader comma between step parts is unselectable, so a copied
row no longer reads "Weight · ,all time".
MCP output is unchanged: the source-snapshot equality suite stays green.
A statement that says "today", or compares today against a baseline,
now only rests on a reading taken on the current calendar day in the
reader's zone. Yesterday morning's pulse read at nine in the evening
was narrated as today's: the signals of the day accepted any reading
from the last two weeks, and the Today overview, the coincident
deviation flag and the health status counted yesterday as current.
- computeSignalsOfDay drops a metric with no reading today, so the
briefing prompt and its content hash never see a stale value as the
day's; the prompt says so in both languages.
- The digest serves no cached briefing written on an earlier day and
drops a reading-backed signal whose metric was not measured today,
so the web hero, the iOS topSignal and the morning push agree.
- The Today overview's vitals line and deterministic lead, the
coincident flag and the health status deviations use the new
isFromToday rule (calendar day, not elapsed hours).
Signal deltas now carry each metric's display precision: the prompt
input is rounded (whole bpm and mmHg, one decimal of kg), and the
delta string under the hero lead is re-read at that precision in the
reader's locale ("+34 bpm", not "+33,72 bpm").
Every text-entry control (Input, Textarea, Select and NativeSelect
triggers, the date and time fields, the calendar's dropdowns, the Coach
composer and the hand-built note fields) now marks focus by turning its
border the new neutral input-focus tone, about 5:1 on the dark
background and 4:1 on the light one, with no ring. A text field matches
:focus-visible on every mouse click, so the ring was on for everyone.
An invalid field keeps its red border and draws its destructive ring
only while focused. Buttons, links, checkboxes and switches keep their
rings.
text-field-focus-guard.test.ts fails when a text-field class string
brings a focus ring or a primary/ring-coloured focus border back.
A standalone PWA on a tablet reserves the status bar in the top bar but
not in the docked panel's header row, which could tuck its title and
icons under the status bar; the new-chat button sat without the home
indicator inset. Both now reserve the same insets as the top bar.
The docked panel was part of the page, below the top bar. It is now a
column of the shell beside the content column, portalled through a new
side-panel slot: it runs from the top of the window to the bottom, flush
right, and the top bar ends at its left edge. Shut, the panel is gone
entirely and the top bar runs to the window's edge. The panel's header row
is the top bar's band, so the two bottom borders draw one line; the panel
takes the sidebar surface and border, like the rail on the left.
The top bar says where the reader is on desktop: "Coach", then the open
conversation's title, truncated. The panel toggle is its last item, 12 px
from its trailing edge, so it stands against the open panel and at the
window's edge when the panel is shut; it carries a tooltip and is 28 px
beside a fine pointer, 44 px with touch.
New chat is a round 48 px primary button with a plus, 24 px in from the
panel's corner (16 px in the phone sheet), with a tooltip. The top bar's
action slot and the new context and side-panel slots share one portal
helper.
Reading a stored reply back drops a view chip whose table the answer
already shows with its toggle, but the resolver reads the raw stored
chips. A tap from a client that still shows such a chip is answered, by
design: the chip was a valid offer when it was stored. The test pins both
halves for "as a table" and "as a chart".
The condition moves out of the shell into isFullBleedPage(), a pure
helper with its own test: /coach renders edge to edge; /coach/plans,
/coach/conversations, a trailing slash, look-alike paths and other
routes keep the padded frame, and so does a refusal on /coach (outside a
shared record, module off).
Focus opens the details tooltip, and the click that Enter or Space then
fires toggled it shut again. A click with no pointer-down before it is a
keyboard activation and now always opens; a tap or a mouse click still
toggles, and Escape closes it.
The flag helper sat between the module gate's doc comment and the
component, so the comment read as its own. It is now
switchedOffModule(), which returns the owning module when it is off and
has a doc comment of its own above the gate's; the gate narrows on its
result directly instead of looking the module up twice.
The panel header buttons shrank to 28 px from md and the conversation rows
to 36 px from sm, so a touch tablet wide enough to dock the panel got
pointer-sized targets. They now shrink only under pointer-fine: 28 px
header buttons, a 36 px gear in the drawer, 36 px rows and row menus
beside a mouse or trackpad, the 44 px floor with touch at any width.
A unit test refuses a width breakpoint on a control size in the three
panel files; an e2e case measures the docked panel at 1280 px with touch
and with a fine pointer.
A table the answer displays with a chart renders its own chart/table
toggle, and the "Show as a chart" / "Show as a table" chips under it
repeated that control. The server no longer derives a view chip for such
a table; a table the answer only used, folded under "Data used", still
gets "Show as a chart", since that brings it into the answer. A reuse
turn shows its table with the toggle and so offers no view chip either.
The rule is one predicate in a leaf module, applied where chips are
derived and where a stored reply is read back, so a reply stored before
the change offers the same chips as a new one on every client.
The tokens and model no longer sit as right-aligned text under a Coach
answer. The row is one left-aligned group: copy, read aloud, try again, a
details icon, then the time. The details icon is a focusable button whose
tooltip names the model and the token count; it opens on hover, on
keyboard focus and on a tap, and a second tap, a tap elsewhere or Escape
closes it. Strings in all seven locales; the joined tokens-with-model
string is gone.
The toggle shows the panel-right open glyph while the panel is open and
the close glyph while it is shut, both mirrored so the chevron points the
way the panel moves on a click. Docked and open, the toggle keeps the
panel's width as its right margin, so it sits beside the panel's leading
edge and travels with it (200 ms, none under reduced motion). Labels and
aria-expanded are unchanged.
The Coach page cancelled the shell's container padding with negative
margins, so <main> kept its reserved scrollbar gutter and painted a dead
strip to the right of the panel on classic-scrollbar platforms. The shell
now renders /coach full bleed: no centred container, no padding and no
reserved gutter, and the page fills what <main> leaves below the top bar.
A refusal card (shared record, module off) keeps the padded frame.
The panel header takes one 12 px inset with a gap of 8 between title and
icons, a text-lg title, and 28 px icon buttons beside a pointer (44 px on
touch). Conversation rows take the control radius and a 36 px pointer
floor.
The same-hours standing averaged each day as sum / n. Pulse already folds
its day as the mean of its local hours' means, and the day-aggregate reader
returns that as dayMean; a type moved onto that reader would have been
compared raw. The standing now reads dayMean where the reader carries one.
Text with no space inside the budget was cut at 160 characters and then
given an ellipsis, so the lead came out 161 long. The ellipsis now counts
toward the budget: a space at the edge still ends a whole word, and
unbroken text is cut one short.
The same-hours comparison cut the earlier days at the current clock time.
At two in the afternoon a day holding only its fasting reading was then
held against mornings plus lunch, and read as below its range again. The
earlier days are now cut at the local time of today's latest reading.
A range read this way is the usual range for this time of day, not the
whole-day band other surfaces show. The deviation carries a basis, the
health-status API publishes it as `basis` (`day` or `sameHours`), and the
Today lead and the health-status card name the range as the one for this
time of day, in all seven locales.
Key rotation rewrote every ciphertext column through a plain Prisma update,
in the in-app corpus walk and in the rotation script alike, so a rotation
stamped the run time on every encrypted row: the Coach panel then listed
old conversations as recent, and every client re-pulled measurements and
mood entries whose sync cursor had moved. Both now read each row's
updatedAt and write it back, matching on it, so a row edited since the
read is left for the next run instead of being stamped. The models that
carry the column are listed beside the registry and checked against the
schema, and the updatedAt guard now covers the rotation writes.
The terms that decide whether the Today lead already names the top
signal's metric were bare stems: "risen" and "chosen" matched the Polish
"sen" (sleep), "impulse" matched "puls", "Passing" matched the Italian
"passi", and the Korean "잠" matched "잠시" (a moment). A lead about
weight then dropped a sleep headline. Every term now has to start a word,
whole-word terms have to end one (Unicode-aware on both sides), and the
Korean terms carry their particle.
Marc-André Bombeck