Ten scenarios, each in German and English, describe what a turn must do
with a question: which tool it calls and with which arguments, whether it
may read anything new, whether it asks back, declines a dose, offers
chips only for what the record holds, answers a chip without a provider,
offers to keep looking after a forced answer, and keeps an earlier
table's title out of the model's context. evaluateScenario grades a run
against them.
The deterministic suite drives POST /api/insights/chat with a scripted
provider; the tool loop, executor, table tool, chips, clarification
filter, method line, reply guards and stream run for real. Every check
also runs against a misbehaving model or a broken record and must fail
there.
route-followups.test.ts pins the front door for chips and answered
questions: validation, the rate limit and capability gate on a reuse
turn, the stored chip winning over anything the request adds, stale
chips and answers degrading to a plain message, and the continue line.
The outbound screen matched "Erhöhe auf 2,4 mg" but not "Erhöhe deine
Dosis auf 2,4 mg": the German change verbs had no slot for an object
between the verb and the target, unlike the English "increase your dose
to". The raise and lower patterns now accept one.
Asking for an earlier table "as a chart" went through show_result, which
builds the requested view, but collecting the turn's tables then reset
every chart to null and only derived a new one for fresh tables. The table
came back without its chart. A table shown again now keeps the view it was
given.
When the tool loop forces an answer at its round cap, the reply leads
its chips with "Keep looking". Tapping it runs a normal turn, under the
budget gate as ever, with two server-written context lines: the question
that was left unfinished (at most 1000 characters) and what the forced
turn already read, each table by the name show_result takes, so the
continued turn shows those again instead of fetching them twice.
An answer is continued at most once. The continued reply records the
message it continues in metricSource.continuationOf and offers no
further "Keep looking", even when it is forced again; a chip on a reply
that is itself a continuation is refused.
Up to three chips, derived on the server from what the turn read: the
other view of a table with a chart, the period before, a year earlier,
a wider window, and a metric a discovered pattern links to. A chip is
offered only when its read was done this turn and the record holds the
metric; the year-ago and wider-window chips also need the record's
first reading to reach back that far. The model may name chips in a
FOLLOWUPS block, but only to reorder ones the rules already allow; every
label is rendered from the catalog.
A tapped chip resolves against the chips stored on the conversation's
latest reply. "As a chart" and "as a table" are answered from the
stored table without a model call or a budget reservation: the rows are
copied unchanged and the message is tagged providerType "reuse". Other
chips run a normal turn with a server-written context line. A chip that
is no longer current falls back to a plain message.
A setting turns the chips off (absent means on).
A table over time is drawn as a line for a level metric and as bars
for a total, with at most two series in one unit. Category counts get
bars, horizontal above six categories, the top eight and one other.
Fewer than two or more than 400 rows, or no numeric column, stay table
only. show_result's view now counts: table drops the chart, chart turns
an earlier day table into counts per range with a histogram (fixed
widths for the common metrics, Freedman-Diaconis otherwise, none below
ten values).
The method line merged a repeated read by keeping the larger count. With
show_result that could let an earlier answer's count stand over the read
this turn just made. A fresh read of the same source, window and period
now always wins, whichever order the steps ran in, and the two are never
added. An entry that rests only on an earlier table says so ("from an
earlier answer"), including a reused table no step accounts for.
The new catalog string is in all seven locales.
get_metric_table now gets a step like every other read: the domain is
its validated metric, and once the table is in, the window, period and
granularity are the ones the table was built with (after the executor's
defaults). The count is the readings the table folds, taken from the
summary the server computed, so a read past the sixth table still counts.
A step whose call produced a table now names it (resultRef), for the
projected tables of the older tools as well.
show_result gets its own step. It reads nothing new: while it runs it
names no domain, and once the stored table is found it takes the source
of the read that built it and the name the copy carries in this turn. A
table that could not be shown is a failed step with the reason, never an
empty read.
The property test now drives show_result too and checks resultRef; the
coverage test gives each tool the smallest arguments it accepts instead
of passing empty arguments to a tool that requires a metric.
get_metric_table reads one metric as days, weeks or months through the
chart reader, cut in the user's zone and collapsed to one source per day,
with null for a period without a reading. The model reads a bounded
summary; the full table rides the turn, encrypted at rest. show_result
shows a table an earlier reply of the same conversation fetched, resolved
only against that conversation's own messages. Workouts, labs, sleep and
adherence results project into tables, earlier tables are named in the
context by metadata alone, and result:rN marks are stripped from the reply.
The evidence disclosure under a Coach answer now carries one muted
line: which sources and windows the turn read, how many readings, and
how they were aggregated ("Blood pressure, last 90 days: 142 readings,
weekly averages · Sleep, last 30 days: no readings in this window").
The server builds it from the recorded steps and tables in the request
locale; it holds catalog strings and counts only, never a reading, and
a source that turned up nothing says so.
The tool-mode prompt gains a recheck rule: when the person questions a
figure, name the source and window, read it again (or reuse the stored
table), and correct the answer if the two differ.
Each tool call now emits a running step as it starts and its final
status as it settles: found with a server-counted number of readings,
empty with the reason, or failed. The no-tools path shows one step for
the record summary. Steps carry catalog labels rendered in the request
language, closed-enum domains and windows, and integers only; a lab
analyte, a value or a model-written argument never reaches one. At most
twelve per turn. They persist on the provenance, so a reload shows the
same list.
The conversation window reaches the step so a call without a window
names the one it actually read. The no-tools transcripts gain the
summary step; a new transcript pins found, empty and failed steps ahead
of the first token.
A reply may end with a ---CLARIFY--- block (kind metric, window or
context, 512-byte cap). The server keeps only metric choices the record
holds, drops the question when fewer than two remain, takes window
choices from the presets, screens the question text like any reply, and
never offers a second question right after one. Choice labels come from
the catalog, never from the model.
An answer arrives as clarification { messageId, choiceId? }; it resolves
against the stored question on the conversation's latest message and
adds one server-written line to the next turn's context. The card sits
above the composer, announces politely and takes no focus.
A reply that is only the block, with no question before it, is now an
empty reply instead of falling back to the unstripped text.
Adds every string the dialog needs up front: step labels, table titles,
columns and controls, the method line, the follow-up chips and their
setting, the clarification card, and the caption of an answer reused from
a stored table. A key catalog shared by server and client names each one;
a test holds the catalog and the bundles equal in both directions.
GET /api/insights/chat/{id}/messages/{messageId}/results returns the
tables one assistant message read, narrowed to the conversation's owner
like the conversation read, with a table withheld as module_disabled when
its domain's module is off and as unavailable when it cannot be read. A
data read: no AI capability.
The published document now describes every stream frame as the
CoachStreamEvent union, including the suggestedAction frame it never
named, the new step, result, followUps and clarification frames, their
components, the provenance fields, the followUp and clarification request
fields, and the reuse provider type.
Each later piece of the dialog gets its own module, answering nothing for
now: the step builder, the table projection and chart choice, the result
ref marks, the follow-up derivation, sentinel, resolution and continue
chip, the reuse turn, the clarification parse and answer, the method line,
and the four prompt addenda. The turn pipeline calls them where they
belong, so the frames go out in the documented order (step, token,
provenance, result, cards, clarification, followUps, done) and a blocked
turn drops its tables, chips and clarification. With every piece empty
the stream and the prompt stay byte for byte as before.
The retrieval loop takes onCallStart and onCallSettled callbacks, fired
for every call with a turn-wide index; a throwing callback is swallowed.
The trace carries each call's schema-validated arguments (never the raw
string, never persisted), the result says whether the answer was forced
at the round cap, and a seedCalls parameter is accepted but not read yet.
A new nullable column, coach_messages.results_encrypted, holds the tables
of values a turn read as AES-GCM ciphertext; only their metadata rides the
plaintext provenance. readMessageResults serves them to the owner, each
table whole or withheld with a reason. The provenance restore learns the
steps, method, table metadata, chips, clarification and the forced-answer
marker, entry by entry against the wire schema.
The column is registered for key rotation, carried through both backup
formats (ciphertext in a disaster-recovery file, readable JSON in a
portable one) and goes with its row on conversation delete and wipe.
Adds the TypeScript types for the four additive stream frames and their
provenance fields, a Zod mirror for the published document held equal to
the types by a type test, and a guard that no newer frame reuses a key the
shipped native client decodes. The chat request accepts followUp and
clarification, inert for now.
The route keeps the front door: auth, the coach capability, the body
read and schema, the rate limit, the locale and the inbound refusal
screen. Everything after it moves into one module per step:
conversation, context, chain, budget, model, reply-guards, cards,
provenance, persist, sse, errors and pipeline. The capability re-check
at the egress site stays owned by the route and is handed to the
pipeline, so the capability inventory still reads both calls there.
The stream producer now writes through a TurnEmitter created inside
createSseStream and passed to the model step, which is where live
progress frames will come from. Nothing writes to it early yet.
No behaviour change: the golden transcripts pass byte-identical, the
existing route tests pass unmodified, and every annotate, audit and
error-code name appears as often as before.
Guards follow the code. The fence call-site check reads turn/context.ts,
the German-default locale ban covers every file in turn/, and the
fenced module-graph test now asserts that turn/sse.ts and turn/errors.ts
reach neither the tool registry nor the snapshot builder. That test
named lib/ai/coach/tools.ts, which stopped existing when the registry
became a directory, so its negative check could not fail; it points at
tools/index.ts now, and a new sanity check fails if the tool route can
no longer reach the path it names.
Making `=true` grant nothing would stop a single-user self-host's local
model on a patch update. It is deprecated for one release cycle instead,
in a narrowed and safe form: it still grants a private host, but only to
the operator's own configurations (the instance-wide admin provider and
AI settings saved on an admin account), and only through the pinned
operator-approved dispatcher, so redirects stay forbidden and metadata,
link-local and unspecified addresses stay refused. A non-admin account's
own base URL needs an exact origin in AI_PRIVATE_ORIGINS.
Provider resolution passes the owning account's role to the client
(`operatorTrusted`); the save route and the connection test apply the
same rule to the account being configured. The boot summary now says the
setting is deprecated and names the replacement, and the admin AI
settings page shows the exact AI_PRIVATE_ORIGINS line for the private
origins saved on the instance (every account's Local and gateway URL and
the admin base URL, scheme, host and port only), so the operator can copy
it and drop the old setting.
The list of provider tags that send data out through a credential the
operator holds (admin-openai, admin-codex) existed four times: in the
consent gate, in the capability resolver, in the shared-provider health
read and in the admin provider-health ordering. A tag added to one and not
the others would be consented for in one place and not the other. The list
and the local-provider test now live in src/lib/ai/provider-egress.ts and
every reader imports them.
The guard meant to keep one copy matched a variable name, so the copy in
the resolver, under another name, never tripped it. It now matches the
list itself in either order and names the one file allowed to hold it.
ALLOW_LOCAL_AI_PRIVATE_HOSTS=true let any user of the instance point the
Local provider or the OpenAI-compatible gateway at any private address
the server could reach, the metadata range included, and a granted host
was dialled with no connect-time pin at all.
AI_PRIVATE_ORIGINS grants a private AI endpoint with the exact-origin
grammar NOTIFICATION_PRIVATE_ORIGINS and NIGHTSCOUT_PRIVATE_ORIGINS use.
A granted call goes through the pinned operator-approved dispatcher:
redirects forbidden, and metadata, link-local and unspecified answers
dropped at dial time even for a granted name. Everything else stays on
the public pin. The Local client (buffered and streaming) and the
gateway and admin-key tags of the OpenAI client all take their policy
from one function.
The host-list form of ALLOW_LOCAL_AI_PRIVATE_HOSTS keeps working with the
same pinning; a never-grantable literal or a host:port entry is refused
with a one-time warning. Its "true" form grants nothing any more and is
reported in the boot readiness summary with the replacement.
A failed lab-report scan no longer puts the endpoint's response body into
the wide event; status and model stay. The response to the caller never
carried it.
The title of a Coach conversation is the first 80 characters of the
opening message, and the note on a custom-metric reading is free text;
both were stored readable while every other note was encrypted.
Migration 0357 adds coach_conversations.title_encrypted and
custom_metric_entries.note_encrypted (the title loses its NOT NULL) plus
partial indexes on the backfill's discovery predicate. Every writer now
stores ciphertext only: conversation create, proactive nudge, reminder
surfacing, rename, entry create and edit, and both restore formats.
Readers prefer the ciphertext and fall back to the old column for rows
not yet migrated; a ciphertext that does not open throws instead of
falling back. The history title search decrypts in memory, since SQL can
no longer match the title.
free-text-encryption-backfill is a boot-time, idempotent pg-boss job
(staggered past the startup storm, daily discovery tick) that seals each
remaining readable value and nulls it in one transaction per row. A pass
that sees a row twice stops with an error instead of looping.
Both columns are registered for key rotation (registry, CLI script, and
the in-app rotation that walks the registry). A disaster-recovery backup
carries ciphertext only, sealing a not-yet-migrated row on the way out;
a portable export carries readable text. The old columns are marked for
removal in a later release.
Building the coarse buckets from the DAY tier read every day of the window
into the process: a five-year yearly read moved thousands of rows per type,
and the Coach's history bands read several such windows per request. The
per-day source pick and the fold now run in Postgres (DISTINCT ON the day,
ordered by the ladder rank and then the source name, the same order the
in-process pick uses), so a coarse read returns one row per bucket. The
long-range chart picks its tier from the span of the window's days and reads
that tier once. The Coach snapshot hands its already-loaded source order to
the history bands instead of each band reading it again.
The status assessments folded their recent days, weeks, months and years on
Berlin days whatever the user's zone, the mood and cycle crosstabs keyed
their metric days in Berlin, the resting-pulse proxy defaulted to Berlin
days, the briefing's opener rotated at Berlin midnight, and every
assessment prompt labelled today's date as Berlin's. They now use the
user's zone. Rollup months and years are labelled by their UTC start,
which is how they are cut; read in a zone west of UTC the label named the
month before.
"Europe/Berlin" was written out in some eighty places. DEFAULT_TIMEZONE in
lib/tz/format is now the one definition: the second copy in mood/date-key,
DISPLAY_TIMEZONE, the Berlin-only day key helpers and the per-route copies
are gone. An in-repo lint rule rejects the literal outside lib/tz and
tests; four areas another change owns are listed as pending.
The runway estimate read FREQ=MONTHLY as one dose every 30 days whatever
its INTERVAL, so a quarterly injection with two doses left showed about 60
days of supply instead of 180. FREQ=DAILY;INTERVAL=2 counted as daily, a
yearly rule ignored INTERVAL and BYMONTH, and a monthly rule ignored extra
BYMONTHDAY values. A CYCLIC schedule was counted as if it never paused,
and a PRN schedule as if it were due every day. The rate now honours
INTERVAL for every frequency, counts the listed days and months, scales a
cyclic schedule by on / (on + off) and leaves PRN out. The low-stock job
selects the schedule type and cycle lengths so it reads the same rate.
Whole-dose counts no longer lose one to binary floating point: 2.4 mL at
0.8 mL per dose is 2.9999999999999996 in IEEE 754 and used to read as two
doses. The stock summary, both runway estimators, the list and detail
routes, the GLP-1 route, the Coach snapshot and the inventory readout now
floor through one helper that treats anything within 1e-9 of the next whole
number as that number.
The "Send raw data" switch promised anonymised single readings from the
last 30 days. It sends bucket averages from the rollup table: per day for
the last 90 days, per week for the year before, per month back to five
years, each with its date and reading count. The label, both
descriptions and the warning now say that, and the "off" text no longer
claims the summaries carry no individual value (they include the latest
one).
The data-posture line and the Coach's own description of the app said
all health data is encrypted at rest and nothing leaves the instance
beyond the AI provider. Free text, documents, questionnaire answers,
AI-written text and tokens are encrypted; measurement values, lab
values, medication names and mood scores are not. Both now say so, and
the Coach is told not to claim otherwise. Two narrower absolutes are
scoped to what they cover: the local-model onboarding title and the
off-host backup card.
All seven locales.
docker-compose.yml forwards its whitelist as "${NAME:-}", so a variable
the operator never set reaches the process as the empty string. Reads
written as a nullish fallback kept that string: the weather client built
an empty base URL and every geocode failed with ERR_INVALID_URL, and the
IP geo lookup requested a relative path.
Reads with a fallback now go through src/lib/env.ts (envValue, envOr,
envFlag), which treats empty and whitespace as unset. IP_GEO_LOOKUP_DISABLED
accepts 1, true, yes and on instead of only "1". A structural test refuses
a ?? fallback written directly against process.env anywhere in src/, and
the compose whitelist guard counts the helper's call shape as a read.
A condition (illness episode) can carry where on the body it sits, in the
person's own words and encrypted at rest, plus an optional side (LEFT,
RIGHT, BOTH). Same enum, codec and reasoning as the procedure's site.
- migration 0354: two nullable columns on illness_episodes, reusing the
laterality type from 0349
- create and edit accept bodySite and laterality; on PATCH an absent key
keeps the stored value, so an edit from the shipped iPhone app, which
never sends either field, leaves them untouched
- GET /api/body-sites: the sites the record holds across visits and
conditions, folded after the decrypt and bounded per kind; with site
(and laterality) the visits and conditions there, one hop to their
documents, lab results and visits. Read in the profile section;
conditions only under an illness grant with the module on, and links
into ungranted sections come back as unlabelled placeholders
- the link service reads document and visit links from the condition's
end of the two existing tables
- /checkups gains a Body sites tab; the visit and condition forms share
one body-site field pair that suggests sites already used; a
condition's page links its site into the view
- the doctor report, the clinician page and the MCP doctor-visit summary
name the site on conditions; the Coach illness block carries it
sanitised, and MCP get_illness_recovery returns it fenced
- backup and restore carry both fields in both file kinds; the site is
registered as ciphertext and rotates with the key
- i18n in all seven locales, OpenAPI regenerated
Refs #1025
A per-medication trackIntake flag (migration 0351, default on for every
existing and new row). With it off the dose, dates and schedule stay stored,
but nothing derives a due dose from them: the reminder worker, the manual
reminder check, the today projector, the auto-miss pass, the low-stock pass,
the doses card, the today list, the outstanding-dose badge, compliance and
its heatmap, the doctor report, the Coach, the BP-status gate, insights,
targets, achievements, the efficacy view and the MCP schedule tool all skip
it.
The medication wire serves such a medication with schedules: [] and the
stored rows in an additive recordedSchedules field, so a client that arms
local reminders from schedules stays silent without an update. Switching
tracking off or back on archives a schedule revision at that instant, the
untracked stretch with an empty payload, so it never counts as missed and
the projector does not put a dose from before the switch back on the card.
A structural guard freezes the set of schedule readers: each one filters on
the flag, derives from dueSchedules, shapes the wire with scheduleWireFields,
or is allowlisted with the reason it derives nothing due. The flag travels
through backup, restore and the medication CSV export.
A visit can be filed as a procedure or surgery (new EncounterKind
PROCEDURE) and carry a body site in the person's own words, encrypted at
rest, plus an optional side (LEFT, RIGHT, BOTH). An existing visit
switches to the new kind by an ordinary edit, keeping its date, reason,
outcome and links.
- migration 0349: enum value, laterality type, two nullable columns
- GET /api/encounters/procedures: the DONE procedures, newest first,
searched after the decrypt over site, side and reason, with body-site
choices and a total computed over the whole history
- create and edit accept bodySite and laterality on any kind
- /checkups gains a Procedures tab with a body-site filter; the visit
form offers the site for a procedure and keeps it in view on any
visit that carries one
- backup and restore carry both fields; the body site is registered as
ciphertext and rotates with the key
- MCP get_visits narrows by kind and reads the whole record for
procedures when no window is named; the Coach visit block names a
procedure's site
- i18n in all seven locales, OpenAPI regenerated
Refs #1025
The today projector mints a pending placeholder for every slot of the
current day, so a medication added in the afternoon carries placeholders
for that day's earlier slots. The hourly auto-miss pass stamped them as
forgotten doses a day later, and every reader that counts raw rows saw a
miss on a day before the medication existed: the miss-free streak broke on
the creation day. The dose history, the rates and the Coach already ignore
such a placeholder; the streak read the row as it was.
The pass now stamps a miss only on slots from the medication's creation on.
A placeholder before it stays pending, so the web and the app can still
offer the slot for logging and a dose recorded for it converges onto it. A
migration returns the placeholders an earlier pass already stamped. It
touches only rows that can be nothing else: server-minted, never acted on,
anchored before the creation but within the day before it, which is where
the projector's placeholders sit. An older auto-miss carried in with
restored history keeps its state.
Refs #1028
The aboutMeQuestions capability admits an ai_insights_only receipt, but the
question generator ran its own gate that demanded a Coach receipt, so a
person whose capability read available still got the fallback questions on
an operator-held chain.
The generator now runs the shared wire re-check (aiEgressRefusal) for the
aboutMeQuestions key, so the receipt kinds and the rule for when one is
needed come from the capability table alone.
The daily sweep and the measurement-ingest hook surfaced due reminders into
a Coach conversation whatever the record's coach capability said, and the nag
cap then dismissed them after three ticks although no badge ever showed. The
app-open path already waited on the capability.
surfaceCoachReminders now holds every row while the coach capability is
unavailable (the same answer that keeps the badge dark), and the nag cap
neither charges nor dismisses a held user's surfaced rows. They surface on
the first tick after the Coach is back. The integration fixture gives its
users a provider and consent, and a held case covers the sweep, the nag cap
and the ingest hook.
A background job opens one wide event for its whole run, and the per-request
memo cached the operator switches, module availability, module preferences,
provider presence and consent kinds on it. The nightly status passes loop
over every candidate inside that one event, so a switch turned off mid-run
did not stop the remaining users, and the wire re-check read the same cached
answer.
memoizePerRequest takes a freshInBackground option; every input that decides
whether AI work may run sets it, so a background caller reads them on every
call while a request still reads them once.
The Coach correlations tool, the per-metric card and the period
narrative scanned every channel, so a switched-off module still fed
their statistics and the text a model was grounded on. The matrix's
modules option is now required, each caller resolves the record's map,
and the Coach skips the lab pass with labs off. A registry guard holds
every caller to resolving the map itself.
The mask's doc comment now states what it does to the false discovery
correction: fewer channels is a smaller family, so q-values of the
remaining pairs move and a borderline pair can cross the threshold.
Brings in the jobs, stored-text and consent package and the extraction
routes. Conflicts resolved:
- The digest keeps its capability-driven reaction line and check-in,
now through the shared `aiCapabilityForRecord`; the local copy of it
is removed.
- The route inventories classify the extraction routes (documents,
labs OCR, medication text, fenced chat and attachments) as actions;
a capability call counts as the gate for the module inventory.
- With every caller moved, `requireAssistantSurface` and
`AssistantDisabledError` are deleted. The `assistant.disabled.*`
codes stay on the capability refusal, and the guard now asserts the
retired gate is gone.
- `aiCapabilityToServe` answers whether stored model text may be shown
from the record's own state, apart from who is reading.
Refs #975
Marc-André Bombeck