Registering `ko` in the `Locale` union left seven locale-keyed tables
without an entry, so the build could not type-check and the safety
screens, the citation-coverage grader and the reference-range parser
read nothing at all for a Korean reader.
Three failures were not missing keys but wrong assumptions:
- The shared fold decomposes with NFD and strips U+0300-U+036F. Latin
accents become plain letters, but Hangul decomposes into jamo, which
are not in that range and survived — so folded Korean text was a jamo
sequence no precomposed table entry could ever match. It recomposes to
NFC now; every Latin, Polish and German fold is byte-identical.
- `normaliseLabKey` kept `[a-z0-9]`, which reduced every Korean analyte
name to the empty string, so the derived index held none of them and a
Korean cholesterol result reached a FHIR export uncoded.
- The reference-range parser had the hyphen and the dashes but not the
tilde a Korean lab actually prints (`3.5~5.0`).
The Korean safety banks carry no `\b` next to a Hangul token — `\b` is
an ASCII `\w` boundary and Hangul is not `\w`, so `/\b증량/` matches
nothing — and the clause order is the mirror of the Latin banks, since
Korean closes with the verb. The dose patterns require a target-marking
particle for the same reason the Latin banks require "to"/"by": without
it the permitted restatement "7.5mg을 복용 중이고, 체중은 내려가고
있어요" tripped the lowering pattern.
The Coach and Insights prompts route `ko` down the reviewed English body
plus a Korean reply directive; there is no `safety-contracts.ko.yaml`, so
the native builder has no Korean ground rules to compose from.
The gate that decides whether a transcribed document is worth an extraction
call scored four classes of evidence and needed two. Three of them — the
reference-range label, the analyte names, the report header — were written in
English and German, so a French, Spanish, Italian or Polish report scored at
most one: the units, which are the same everywhere. Auto-staging never fired
for it. The document was lost from the flow silently, because the manual
extract button stays available and nothing reads as broken.
The text is folded once — lower case, accents dropped — and every pattern is
matched against that, which on its own recovers a good deal: "Cholestérol",
"Créatinine" and "Emoglobina" reach the existing stems as soon as the accents
are out of the way. The reference-range labels and report headers of the four
locales are added by hand, because they are prose a lab printed and the app
owns no string for them.
The analyte names ARE derived, from `labs.catalog.<slug>` across every shipped
bundle: those are names the app itself displays, and a new locale should not
need an edit here to be recognised. Names under four folded characters are
dropped — "ALT" is German for "old", and a three-letter alternative in a gate
this cheap would make every German letter a lab report. The acronyms worth
having are in the hand-written half, where each was considered on its own.
Which way this errs, stated at the code: a false negative loses a whole
document from auto-staging, while a false positive costs one extraction call
and stages facts a person then declines — nothing commits either way. So the
vocabulary is widened, and the two-distinct-class rule holds the other side.
The derived names extend the analyte class rather than forming a fifth one, so
a document naming one analyte in two languages still scores one. Tests pin
each addition on a case that fails the gate without it, and pin the prose that
must keep failing it.
Document search does whole-token equality over a blind index, so
'Blutdruck' never found the discharge letter that says 'blood pressure'
and 'hypertension' missed every German Arztbrief. A curated German/English
medical synonym list (vitals, common labs, imaging, specialties, document
words — static, no vendor names) now expands the QUERY tokens before they
are hashed: the stored index, its tokenizer version and the schema are
untouched, the expansion happens ahead of the one-way hash, and recall
simply widens. Typed tokens always survive the expansion cap.
- The provider-chain walker now checks the day's recorded spend against
OPERATOR_COST_CAP before invoking an operator-funded admin-* entry. A
request whose primary provider runs on the user's own key reserves
under the generous user-plan ceiling; a fallback hop (or health-ledger
reorder) onto the operator's key could then spend past the operator's
cap unchecked. An exhausted cap skips the candidate like a failed hop.
- summaryState PENDING gets a giver-upper: the detail serialiser
degrades a PENDING older than an hour to UNAVAILABLE (which offers the
manual generate action), and an hourly reaper persists the same heal
following the orphan-ImportJob reconcile precedent — a job dead past
its retry budget no longer leaves the sheet saying "generating"
forever.
The residual scan from the report is a CCITT Group 4 single-image PDF,
the codec class pdfjs rasterization is known to fail on. Chain checks:
a pdfjs open failure or throw already lands as render-failed with the
error text annotated; the zero-pages shape now annotates too instead
of returning silently. An end-to-end test drives the REAL rasterizer
through prepareVisionInput and pins the rasterFailed mapping; the test
header documents why a tiny in-repo CCITT fixture is deliberately not
checked in (probed against this pdfjs version, it renders the page as
blank instead of throwing, so the fixture cannot deterministically
fail and would be flaky across upgrades). The detail view's
raster-failed line now names scanned fax compression formats as a
known cause, in all six locales.
Refs #776
A content-index attempt that failed (raster failure, a scan with no
text layer, a provider error) recorded its reason only in the server's
observability stream. In the vault, hasContentIndex false looked
identical for a document nobody ever tried and one that failed three
times, and the detail sheet had nothing to say about either.
Apply the summaryState pattern to the index. Two additive columns on
InboundDocument (migration 0333) record when the last attempt finished
and why it produced no index, as a closed reason set: local-empty,
local-unsupported, decrypt-error, raster-failed, provider-error,
pdf-needs-anthropic, empty-transcription. The writer sits at the choke
point the auto job and the corpus backfill share, and in the manual
index route; a success clears the reason. The provider path's specific
failure now survives the local fallback, since "no text" is less
actionable than "the PDF would not render". rasterizePdf names its
failure class so a build that cannot rasterize at all is distinguished
from a PDF that would not render. New guard in both index paths: an
empty provider transcription never becomes a successful empty index.
The detail sheet renders the reason in plain words (muted, six
locales). The DTO and OpenAPI carry the two fields; the disaster
recovery backup and restore carry them verbatim like their summary
siblings.
Refs #776
The document-AI actions (read/index, extract, suggest, summarise) shared a
hard-coded 6/hour bucket that lied twice: the 429 said "try again in a few
minutes" while the window is an hour, and several failure paths burned a
slot before any provider was dispatched.
- DOCUMENT_AI_LIMIT_PER_HOUR configures the shared per-user ceiling
(default 6, clamped 1-1000), wired end to end: compose environment
whitelist, .env.production.example, env manifest, operator docs.
- INSIGHTS_RATE_LIMIT_PER_HOUR joins the compose whitelist and manifest;
it was documented but never listed, so an operator value could not reach
the container.
- The 429 mirrors the bucket's reset instant into meta.retryAt (plus the
existing X-RateLimit-Reset header); the client derives "try again in
about N minutes" from it in all six locales, with a plural-aware key.
- A slot is consumed only when a request actually reaches the provider (or
the local index/echo work): malformed bodies, unreadable originals,
refused consent, and exhausted daily budgets refund the charged slot via
a window-scoped refundRateLimit primitive. Consistent across all four
routes sharing the bucket.
Every staged fact carried a sourceText the model echoed back, stored verbatim
as 'the exact verbatim span of the document the fact came from'. Nothing
checked that the span existed. A paraphrase, a normalised rendering, or an
outright invention was persisted and shown to the reviewer as a transcription
of their own clinical record — an anchor pointing at nothing, or at a span the
document never contained.
Treat the echo as a lookup key instead. anchorSourceText resolves it against
the extracted text (exact, then whitespace-collapsed and case-folded to absorb
OCR reflow) and stores the span the document actually carries, with its offset.
A quote that matches nothing is dropped and the fact is flagged unanchored;
vision mode has no extracted text at all, so its facts are unanchored by
construction. Rows written before this resolve to unanchored on read — their
quote is preserved but never presented as checked.
A READY row whose ciphertext will not open reports UNAVAILABLE, so the view
does not render a heading over nothing. Nothing asserted that, and removing the
degrade left the suite green.
The coverage check hung off generateInsight, whose only caller runWithFallback
has no production caller left -- every live surface routes through
runRawCompletionWithFallback instead. The admin quality dashboard was reading a
metric nothing emitted. Widen the input to the structural shape both payloads
share and call it from the comprehensive post-parse step, so the annotation
reflects real generations.
Writing the locale tests found three gaps the English-only banks hid: Polish
verb suffixes fell outside the ASCII word class, a trailing word boundary after
an accented character never closed, and German subordinate clauses put the verb
last so the natural phrasing of a dose instruction never matched. The payload
collector also skipped legacy bare-string recommendations entirely.
Two directions, folding the vault and labs pipelines into one linking path
without weakening either manual flow.
Vault document → lab values (auto-stage). After the per-document auto-index
job transcribes a stored document, a lab-looking document is run through the
same inbound extraction the manual "Extract" button uses, but the facts land
PENDING for the existing human review — nothing is committed. Gated on both
the documents and labs modules plus the auto-AI-read consent gate the
auto-index external path already uses; a text-structuring pass over the
transcription the index just produced, so no second vision call is spent. The
staging is a guarded STORED→EXTRACTED transition, and a keyword heuristic (or
an explicit LAB_RESULT kind) keeps an arbitrary photo from burning a call.
Everyone without the opt-in keeps the manual extract button.
Lab scan → vault. A lab report scanned in the labs area is now also filed
into the Documents vault (kind LAB_RESULT) through the existing upload —
encrypted at rest, EXIF-stripped, thumbnailed, sha256-deduped — and the
committed labs are cross-linked back to it with the same APPROVED-observation
provenance the manual vault → labs path produces. Text-mode scans keep the
image on-device and file nothing. Idempotent: a re-commit inserts no rows and
links nothing new, and the vault dedupe means no duplicate document.
Owner-scoped and module-gated on both sides; the cross-link is best-effort so
it never fails the authoritative write.
Replace the scalar document_id discriminator on a coach conversation with a
CoachConversationDocument join table and a sticky documentScoped flag, so a
conversation can be grounded on several stored documents at once while the
prompt-injection fence stays intact across the data-model change.
The fence is re-established at both enforcement points, fail-closed:
- Client: resolveCoachSendTarget routes any turn on a fenced conversation (or a
fresh chat with staged attachments) to the new /api/insights/chat/fenced
endpoint; the fenced body is strict and never carries scope, guided-question,
or prefill hints.
- Server: the tool route fetches with documentScoped:false AND 404s when the
fetched row still carries an attachment (flag/join drift), emitting a
fence_drift audit alarm. Document text now enters an LLM prompt in exactly one
code path — the fenced pipeline — with no tools and no health snapshot; a
module-graph test pins that the tool loop is unreachable from it.
The fenced turn fences each document separately with a server-built, scrubbed
header line (title and filename are untrusted text), runs a per-document egress
consent check on the single picked provider, and grounds numerically over the
live attachment union only. Attachments are capped at five documents and 192 KB
of combined text with honest truncation.
Attach and detach ride dedicated endpoints: attaching validates the document is
owned, live, indexed, and within cap, then flips the conversation fenced (the
one legal, privilege-reducing transition, audit-logged); detaching removes the
document but never clears the flag. The document-sheet chat keeps its public
wire contract and shares the same fenced pipeline over the join.
Migration 0245 adds the join table and the flag, backfills every legacy
single-document chat into a one-attachment fenced conversation, and drops the
scalar column in the same transaction — no dual-truth window.
The bundled Skia x64 build (document thumbnails, scanned-PDF rasterization)
uses AVX2 instructions; on an x86-64 host without them — Celeron/Atom-class
NAS CPUs, pre-2013 Xeons, VMs masking CPU flags — the first canvas call is an
uncatchable SIGILL that kills the process and crash-loops the container (live
incident: a Synology DS1520+, exit 132 ~18 s after every boot, reported with
kernel traps and the correct root cause in hand).
The renderer is now gated on a one-time CPU-feature check before the module
ever loads: linux/x64 requires the avx2 cpuinfo flag; arm64 and non-Linux are
unaffected; unreadable cpuinfo fails open. Behind a closed gate, thumbnail
generation and PDF rasterization take their existing fail-soft paths (vault
shows type icons; PDFs read as text) and the boot-time thumbnail backfill
skips enqueueing. NATIVE_CANVAS=off|on overrides detection, documented in the
env example and whitelisted in compose.
Dashboard layout changes take effect immediately: the save now evicts the
locale-keyed home-screen snapshot it previously missed, so a re-ordered or
toggled tile set no longer reverts to the prior selection for the SWR window.
A share link built with a narrowed report-section selection now serves exactly
that selection. The stored selection is the grouped export shape; the shared
view read it through the flat parser, which dropped every grouped toggle and
fell back to the full defaults — an over-share against the owner's choice. The
view now folds the grouped shape down faithfully. Documents-only shares were
never affected.
Hardening: the shared-document download path now sanitises the filename
identically to the owner download path, and thumbnail generation re-checks the
pixel ceiling after decode so a malformed image that slips past the header
sniff still cannot exhaust the worker.
Document vault shows a per-file thumbnail: a server-rendered preview,
downscaled and re-encoded so location and camera metadata are dropped,
served only to the file's owner over an authenticated request. Files that
cannot be previewed keep their type icon. Generation is bounded — a header
sniff refuses an over-large source before decode, so a crafted image cannot
exhaust the preview worker.
Also folds in admin-managed central AI access — an administrator connects
one access on the server and each user opts in or keeps their own provider,
off until an admin turns it on — and trims the redundant documents-search
caption and per-file searchable/read badges.
Render a small JPEG preview for each stored document and show it at the
card's leading edge, falling back to the kind icon when none exists.
- Generate previews in the background on upload (new document-thumbnail
pg-boss queue) plus a boot-time corpus backfill for existing documents;
upload never blocks on or fails because of it. Images are downscaled via
@napi-rs/canvas, a PDF's first page is rasterised then downscaled, every
other type is skipped. Generation never throws.
- Store the preview encrypted at rest in a 1:1 DocumentThumbnail side table
(migration 0236) so the list/detail queries that omit the original blob
never drag the preview either. Same AES-256-GCM string codec + fail-closed
decrypt as the content index; wired into the key-rotation walk.
- Serve it through an owner-scoped GET /api/documents/inbound/[id]/thumbnail
(auth + module gate + userId-scoped query + fail-closed decrypt), always
image/jpeg + nosniff + private, no-store; 404 on miss. The card renders an
authed lazy <img>, onError falls back to the icon.
- Add hasThumbnail to the document DTO so the card only fetches a preview
that exists.
Also drop the "search looks inside your documents" caption from the overview
(the index-all prompt stays) and the per-card searchable marker; both were
noise. Remove the now-unused i18n keys across all six locales.
Introduce an operator-level ChatGPT-subscription (Codex) credential that the
operator connects once and shares server-wide, plus a per-user switch to route
AI egress through it.
- Schema (migration 0235): three encrypted columns on AppSettings for the
operator's central Codex access/refresh tokens and the ChatGPT-Account-ID
claim, plus expiry/connected-at/status; a per-user useCentralCodex flag
(default false). Register the new encrypted columns in the rotation registry
and script.
- Resolution: add the admin-codex chain type; append it to a user's resolved
chain only when they opted in AND the operator connected it. Tokens refresh
server-side against the singleton row (CAS-style), same primitives as the
per-user Codex path.
- Budget: bill an admin-codex primary against the operator cost cap, not the
user plan — it drains the operator's account.
- Consent: admin-codex is server-managed external egress, gated by the existing
consent machinery (self-snapshot and document surfaces alike). No existing
consent path is weakened.
- Document order: rank admin-codex with the subscription tier (last), so the
document class stays local-first.
- UI: admin device-code connect/disconnect card (cookie-only admin routes) with
honest copy on account pooling, shared rate limits, and training-by-default;
a per-user opt-in switch shown only once the operator has connected it, off by
default, behind a one-time honesty confirm.
- i18n across all six locales; unit tests for the chain type, budget cap,
consent gate, vision capability, and document order; a mocked e2e for both
the admin connect surface and the per-user switch.
The central credential's tokens are encrypted at rest and never returned to a
non-admin API; the opt-in only takes effect once the operator has connected the
shared account.
Only Anthropic accepts a native PDF block; codex and other image-only-wire
providers dropped it. Add a bounded, best-effort server-side rasterizer that
renders a PDF's pages to JPEG images those providers read as ordinary image
input, so a scanned/image-only PDF is legible on the auto-read path.
- @napi-rs/canvas promoted to a direct dependency; its musl prebuilts + the
pdfjs loader pinned into the standalone trace (next.config.ts)
- rasterize-pdf.ts: lazy pdfjs import (keeps DOMMatrix out of the eager chunk),
first 10 pages, 1600px long edge, JPEG q80, never throws
- prepareVisionInput rasterizes a PDF for a non-Anthropic vision provider and
keeps the native PDF block for Anthropic; the document capability reports
PDFs readable via raster so the UI offers the read
The auto-index-on-upload job resolved over the cost-first app-wide chain
(codex priority 1) and gated with the chain-based consent check, so a
pure-codex chain could silently egress a freshly uploaded document to the
subscription backend on upload. Reconcile it onto resolveDocumentVisionProvider
(local-first, codex last) and govern external egress with the
documentsAutoAiRead opt-in: a local pick always runs (no egress); an external
pick runs only when the toggle is ON, otherwise the job stays strictly local.
This tightens the default and makes the opt-in path deliberate.
Add GET /api/documents/inbound/capability (document provider order + vendor-blind
egress class), repoint the vault capability hook to it, and show a per-egress
notice in the detail sheet before a document read leaves the machine. Cover the
document provider order, the egress consent gate, and the notice with tests.
i18n x6 + OpenAPI regenerated.
Co-Authored-By: Marc-André Bombeck
Index a document for content search automatically the moment it is stored,
instead of requiring a manual index action. A fire-and-forget per-document
job runs an AI-first decision tree: when a vision-capable provider is
configured and consented, transcribe the stored original (rich, handles
scanned PDFs and images), budget-reserved exactly like the interactive index
route; otherwise fall back to server-side text-layer extraction via pdf-parse
(no provider, no egress, milliseconds) so content search works at a baseline
even with no AI. Budget exhaustion, a provider that cannot read the file
(a text-layer PDF on a non-Anthropic account), and provider errors all fall
through to the free local path rather than leave a document unsearchable.
Both paths write the same blind, encrypted content index (AES-256-GCM text +
opaque HMAC token tags) — nothing readable at rest. The local path adds the
'local-pdf' / 'local-ocr' source values (free String column, no migration).
Server-side OCR for scanned images and scanned PDFs is left as a documented
seam. Pin pdf-parse + pdfjs-dist into the standalone trace so the wasm text
extractor ships in the runtime image.
P2-1. The list GET now issues a documentContentIndex lookup and unions a
content-token OR branch when a search term is present, so the existing route
test gains the mock + the third OR branch in its where assertion. Fix the
backfill test's findFirst mock signature under strict tsc.
P2-1. Unit tests for the token scheme (whole-word match, stopword drop,
de-accent, no plaintext at rest, subkey follows the active key), HKDF subkey
derivation, the suggest route (drafts only, writes nothing, provider-gated),
the summary route (session-only, budget+audit only), the index route (vision
transcribe→upsert, text posts OCR text, provider/local-OCR gated), and the
backfill (no-provider/no-consent no-op, indexes, budget-reached resume).
The policy module (src/lib/documents/upload-policy.ts) is the one source of
truth for what the vault accepts and how it serves: Class A (PDF/JPEG/PNG/
WebP/GIF) renders inline; Class B (Office OOXML+CFB, text/CSV/Markdown/RTF,
TIFF, HEIC/HEIF, XML/JSON) is stored verbatim and leaves the origin only as
an opaque attachment download; executables, HTML/XHTML/SVG, and generic
archives are denied. Detection is magic-bytes first — the wire Content-Type
is never trusted; containers (ZIP → OOXML, CFB → legacy Office) and
extension-only text formats gate on the filename plus a no-NUL heuristic.
resolveDocumentLimits() reads the admin-tunable per-file cap (clamped to
the hard 100 MiB ceiling — single-shot GCM and bounded in-memory reads are
load-bearing past that) and the quota (user override ?? instance default).
crypto.ts gains the binary2 codec: encryptBytes/decryptBytes do AES-256-GCM
over raw bytes with a versioned key-id header, skipping the base64 detour
(−33 % at rest, one fewer copy). Same fail-closed discipline as the string
codec. The store dispatches per row on the explicit contentCodec column:
new uploads write binary2, pre-existing base64v1 rows keep decoding
byte-identically, and an unknown codec throws instead of guessing.
Exhaustive unit coverage: every accept class from real magic-byte fixtures,
every deny case (incl. the PDF-with-HTML-head, SVG-as-image, and
ZIP-renamed-docx polyglots), header-matrix source of truth, limit clamping,
codec round-trips both ways, crossed-codec and unknown-codec fail-closed.
Tasks: W1-T3, W1-T4.
The insights alert cards computed medication compliance with the naive
`schedules.length × days` denominator, so a fully-adherent weekly
injectable read ~14% and fired a false low-compliance alert (and a paused
medication mis-fired the same way). Route the alert path through the
canonical `calculateCompliance` engine like every other compliance
surface — pause eras, schedule revisions and dose windows honoured.
Document confirm: committing a numeric observation against an existing
biomarker silently replaced the stated unit with the catalog's, with no
conversion — a stated "6.1 mmol/L" against a mg/dL marker committed
"6.1 mg/dL". Reject on a unit mismatch (fail closed) so the user
reconciles on the review screen, the same posture as the unit-required
guard.
Wire grip strength, waist (with waist-to-height) and pain into the
briefing feature set and the grounding allow-set so the clinical-depth
signals reach the daily narrative; the mental-health screeners stay
excluded by construction.
Biomarker detail: fall back to a generic description for catalog-less
markers, and keep the heading + description with a retry on the error
state. Collapse the medication External-API and Lifecycle groups by
default. Re-enable the insights scroll-to-top regression and assert the
shipped main-content reset.
Align the EN haemoglobin/haematocrit display names to British spelling.
Drop the redundant `(userId, deletedAt)` index on inbound documents — a
strict prefix of the new three-column index. Trim the dead
MedicalDisclaimer variants to the one live `dataPosture`. Move the
crisis-resources card off the retired amber tint to a calm
destructive-subtle border. Add a registry invariant asserting every
biomarker signal key resolves to a real catalog slug.
Ingest an uploaded doctor report / discharge letter through the existing
dedicated OCR/vision provider rails (the v1.22 document-scan provider/key,
reused — no new provider) and transcribe structured facts into staging.
The safety line is extract, never interpret: the prompt frames the document
as untrusted data to reproduce verbatim. Diagnoses map to Condition,
measured values to Observation, medications to MedicationStatement — with
STATED status only. A code survives only when the document writes it and its
system matches the resource (SNOMED/ICD-10 for a diagnosis, LOINC for a
value, RxNorm/ATC for a medication); nothing is inferred, no value is
range-flagged, no condition is linked to a medication. Every fact carries
per-field provenance and a confidence score; a fact below the confidence
floor fails closed — it cannot be approved until the user edits it.
Routes (apiHandler + Zod + session-narrowed userId, opt-in module-gated):
- POST/GET /api/documents/inbound — upload+extract (vision or local-OCR
text), and list. The raw document is stored encrypted at rest.
- GET/DELETE /api/documents/inbound/[id] — review detail; soft-delete.
- PATCH .../facts/[factId] — correct a staged fact before approval.
- POST .../confirm — the only write path: approved facts flow into the
labs / conditions / medications stores through their normal create paths;
rejected facts are discarded; low-confidence facts are refused.
aucun6352