The summary and suggest routes accept `Prefer: respond-async`: after the
same quick refusals they answer 202 with a run id, and the background
document AI worker runs the read under two new run kinds
(DOCUMENT_SUMMARY, DOCUMENT_SUGGEST, migration 0370). Without the header
both routes answer synchronously as before; both paths run one shared
body, so they cannot drift.
The worker asks the capability for the record and re-checks the wire for
the provider it picks before every read. A summary that is to be stored
marks the document PENDING while queued, READY when it lands, and
UNAVAILABLE when the run fails or the write does not land. A second
request attaches to a live run only when it would produce the same thing.
The vault asks for both in the background and shows the same calm
waiting line the index uses, worded so a transient read never promises
that leaving the page keeps it. The egress guard and the route inventory
now hold every run kind to its dispatcher re-check and its capability.
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.
Coach chat bubbles and the document OCR text panel wrapped long unbroken
tokens (URLs, IDs) past their column width on narrow viewports — add
break-words. Drop the /90 alpha on the coach error text so it stays on the
AA-retuned status token.
Today hero's loading skeleton had drifted from the v1.29.1 tightened hero
(gap/items-start/space-y/rail-heading/xl:grid-cols-3), causing a visible
shift on every dashboard load; mirror the loaded classes 1:1.
The wellness-scores strip returned null while its shared batch loaded,
so it popped in late and pushed the page below it down — render the
section heading + tile-shaped skeletons instead, matching the vitals
dashboard's loading pattern.
Priority-card actions (coach check-in keep/let-go) had no pending state,
so a slow round trip could be double-tapped into firing the mutation
twice; disable the action row while in flight. Priority-item dismiss is
now optimistic (removes the card from the rail on tap, rolls back on
error) instead of waiting on invalidate-after-success.
Google Health: the fullSync rollup-defer tracker collected its touched keys via
a single spread push — one argument PER READING, so a multi-year heart-rate
history (six figures of samples) overflowed the call stack and aborted the
metrics collection mid-cycle, exactly on the accounts with the most data. A
plain loop replaces the spread. Reported by a self-hoster; the v1.28.21 sync
diagnostics confirmed every other read on that account healthy in one log line.
Documents: the vault card's read-by-AI tag no longer takes its own line — a
small glyph rides inline on the date-and-size row. The detail view drops the
status bar under the preview entirely; with automatic reading on and nothing to
review the whole block collapses, so the content starts straight with the
document's fields. The provenance contract moves with it: the e2e specs now pin
the card marker instead of the sheet pill, and the unmounted status-pill
component is removed rather than left dead.
When "read documents automatically with AI" is on, the detail sheet no longer
shows the manual per-document action row (Read / Suggest / Summarise) — reading
already happens on upload, so the buttons only nagged. The section header and
the searchable status pill stay, and content search keeps working. With the
setting off the manual controls render exactly as before.
Drop the "review before saving" heading and its grey sub-caption from the
suggestion panel; the reviewable fields and the apply / dismiss controls stay.
Remove the now-unused documents.assist.reviewTitle / reviewHint keys from every
locale.
The bundled (Turbopack) copy of pdfjs-dist failed its render path in the
standalone image with a bare Error, so a scanned PDF on a non-Anthropic
provider fell back to 'PDF scanning needs a Claude vision provider' — while the
identical un-bundled module rasterizes the same document fine (verified in the
container end to end on the real document). Externalize pdfjs-dist and
@napi-rs/canvas so the server loads the real modules, and hoist pdfjs-dist to
the top level in the image so the runtime import resolves. Also log the render
error message (not just its name) and drop the redundant 'not saved' note from
the document summary panel.
The PATCH detail route now reads the content-index provenance, so its route
test mocks documentContentIndex.findUnique; the searchable status spinner and
the Read-with-AI pulse pair with motion-reduce:animate-none per the coverage
guard.
Auto-index on upload means content search is silent-by-default, so the old
per-document index chore no longer reads as a task. Consolidate every AI
reading action into one inviting block on the detail sheet, led by a
prominent Read with AI action, and show each document's searchable status as
a pill that distinguishes an AI-read document from a locally-indexed one.
- New Read-with-AI block: prominent read/re-read action plus suggest and
session-only summary, gated on provider availability; the calm settings
pointer replaces the actions (never an error) when no provider is set up.
- Searchable status pill reflects the auto-index: Read by AI (provider read
the original, highlighted), Searchable (local), or a quiet not-yet state.
- Card searchable marker distinguishes an AI-read document with a Sparkles
glyph from a locally-indexed one; both carry an accessible label.
- Drop the manual index/re-index chore copy; keep the corpus baseline
affordance in the filter bar for older un-indexed documents.
The usage endpoint counted every DocumentContentIndex row for the user,
including those whose parent document has been soft-deleted (the index
row survives until the hard purge). Against totalCount, which filters
live documents, the gauge could read past 100 %. Scope the indexed count
to live documents so coverage never exceeds the corpus.
Also drop a redundant identical-branch ternary in the per-document
content-index status row.
Add an optional AI layer to a stored document, gated on the existing AI
provider and consent — no new toggle. All of it rides the shared OCR
capability probe so the client only posts a transport the endpoint accepts
(vision empty-body POST, or local-OCR text posted from the browser).
- "Suggest details" prefills the title / type / date as reviewable DRAFTS.
Nothing is written on its own: the title lands in the editable field and
is saved through the existing commit path; type and date each apply on an
explicit tap. The card carries a visible "review before saving" affordance.
- A summary / extracted-text button opens a transient panel that renders the
result once and always shows a "not saved / not a diagnosis" note; closing
it discards the text. Nothing persists beyond the view.
- When no provider is configured the AI area is a calm pointer to the AI
settings, never an error; the manual metadata form stays fully usable.
The detail-sheet AI area is lifted into a presentational section component so
the availability gate and the review-first wiring render without the sheet's
dialog portal and are pinned by static-render tests. Strings land in all six
locales.
Marc-André Bombeck