QR "scan to connect" pairing has no wallet number, so nothing registered
the request with the relay and the scanning device's response was rejected
("unknown or expired request"). The backend now pre-registers the pairing
request over the relay's new hub:expect event on POST /pair, and
re-registers still-pending requests on hub (re)connect so routing survives
a relay restart. /pair now requires the clinic to have joined the network
(clear 409, surfaced in the import dialog, localized in all five langs).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The relay is now multi-clinic. Each clinic authenticates to the /hub
namespace by signing a challenge with its own Ed25519 clinic signing key
(a per-clinic identity, not a shared RELAY_TOKEN), and the relay routes
every device response back to only the clinic that originated the request
(keyed by requestId) — so clinics never see each other's traffic.
Backend:
- clinic_signing_keys.network_enabled + GET/PUT /api/signing/network
(owner/admin) to join/leave the network.
- relay-client keeps one authenticated hub connection per network-enabled
org (connectOrg/disconnectOrg, hubs map keyed by orgId); emitToWallet/
sendToWallet take orgId; offline flush is org-scoped.
- Wallet import/push return 409 until a clinic joins.
- RELAY_TOKEN is now optional/legacy (open relay needs no shared secret).
Frontend:
- "Join Temetro Network" toggle in Settings → Signing, localized in all
five languages (en, fr, de, so, ar).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Devices no longer connect to the backend directly. The /wallet Socket.io
namespace is removed from realtime.ts; a new services/relay-client.ts connects
to the standalone Temetro Network relay's /hub namespace (RELAY_TOKEN-auth),
emitToWallet delegates to its sendToWallet, and device responses + wallet:online
replay are handled there via the same wallet-share/wallet-updates services.
- Add RELAY_URL + RELAY_TOKEN env (env.ts, .env.example, docker-compose.yml);
the wallet-import QR (resolveRelayUrl) now points at RELAY_URL.
- Add socket.io-client dependency.
- Document the Temetro Network folder/service in root + backend CLAUDE.md.
- Bump root/backend/frontend to 0.7.0; CHANGELOG entry.
The relay service itself lives in github.com/temetro/temetro-network.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Expose temetro's own records as a read-only FHIR R4 server at /fhir,
authenticated with per-clinic API keys (tmf_… bearer tokens, SHA-256
hashed, shown once). Serves Patient, Observation (labs + vitals),
AllergyIntolerance, Condition, MedicationRequest, Encounter and
Appointment as text-only CodeableConcepts (temetro stores free-text
clinical values); CapabilityStatement at /fhir/metadata (unauth).
Searchset Bundles with _count/_offset pagination and self/next/prev
links; every request is org-scoped and written to the activity log.
Keys are created/revoked under Settings → Integrations (owner/admin).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A clinician can push an updated record to a wallet-linked patient (permanent
share). The snapshot is signed with the clinic Ed25519 key and sealed to the
wallet's X25519 key — derived from its Ed25519 wallet number via the birational
map, verified byte-for-byte against the wallet's own derivation. Stored pending,
delivered over the /wallet relay live and on the wallet's next authenticated
connect (offline catch-up). The patient approves/denies in-app; the wallet signs
its decision, the backend verifies it, and the record is replaced only on
approval. Wallet pins the clinic key (TOFU) and warns on change.
Backend: walletRecordUpdates table + service, ed25519PubToX25519Hex helper,
POST /api/patients/wallet/push, GET .../link/:fileNumber|updates|updates/:id,
wallet:update-request / wallet:update-response relay events.
Frontend: "Push to wallet" dialog with live status, wallet-link gating on the
patient sheet, "Sent updates" list under Settings → Signing, walletPush /
walletUpdatesList locale namespaces across all five languages. Bumps to v0.5.0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Record a clinician↔patient visit (or paste a transcript) on the patient
sheet; the backend transcribes it (OpenAI Whisper / Gemini), de-identifies
the transcript + context through Veil, and drafts a structured SOAP note
the clinician reviews and edits before saving — the same write-approval
gate as the chat agent.
Backend: POST /api/scribe/{transcribe,draft,save} (routes/scribe.ts,
services/ai/transcribe.ts), veil.redactText() free-text redactor,
appendEncounter service, audio MIME types on attachments. Gated by
patient:write + the clinic AI policy (reception/disabled-AI excluded).
Frontend: ScribeDialog + lib/scribe.ts, "Record visit" on the patient
detail, gated by clinical access + AI availability. New `scribe` locale
namespace across all five languages. Bumps to v0.4.0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add three UI locales (so/ar/de) with full ~1,660-key translations alongside
en/fr, selectable in Settings → Profile. Arabic gets full right-to-left support:
- config.ts registers the locales and exports a `dirFor` helper; an inline
<head> script in layout.tsx sets <html dir/lang> before first paint (no RTL
flash), and i18n-provider keeps them in sync on language change.
- ~160 physical direction utilities converted to logical (ms/me/ps/pe/
start/end/text-start/text-end); directional chevrons/arrows get rtl:rotate-180;
chat-bubble align variants fixed to logical.
- IBM Plex Sans Arabic appended to the sans/heading font stacks for
per-character Arabic fallback.
- Language persists to the backend user_settings and re-applies on sign-in so it
roams across devices (localStorage stays the offline source of truth).
- New scripts/check-locales.mjs (npm run check-locales) enforces key/placeholder
parity and Arabic CLDR plural completeness.
Bump to 0.3.0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Mirror the existing POSTGRES_PORT override for the backend, frontend and adminer
host ports (BACKEND_PORT / FRONTEND_PORT / ADMINER_PORT) so a port clash on
`docker compose up -d` can be fixed via .env without editing the compose file.
Documented in .env.example and the README.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
GET /api/version now reads the latest version from Docker Hub image tags (the
actual update channel clinics pull), falling back to the GitHub release if
Docker Hub is unreachable, with a shorter 1h cache and a `?refresh=1` bypass.
Settings → About & updates gains a "Check for updates" button that forces a
fresh, cache-bypassing lookup.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
"Show me <name>'s medical record" relied on the model chaining
searchPatients → getPatient, but Gemini Flash often calls searchPatients
then emits the canned closing line ("Here's the record.") without the
second tool call, so no data-patientCard part is ever written and no card
renders. (The prior Gemini fix only covered the empty-schema list tools.)
searchPatients now writes the record card (data-patientCard, or
data-recordGraph in graph mode) and a source directly when EXACTLY ONE
patient matches — mirroring getPatient — so the common name-lookup flow no
longer depends on a second tool call. Multiple/zero matches keep returning
the disambiguation list. The tool description and system prompt tell the
model the card is already shown on a unique match so it doesn't double-render.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The card-emitting chat tools (listAppointments, listTasks,
listPrescriptions, getClinicInfo, getAnalytics, listInventory) used an
empty `z.object({})` parameter schema. Google Gemini can't emit a
function call for a tool whose JSON schema has no properties — it prints
the call as `tool_code` text instead of invoking it — so the tool never
ran and the frontend never received the data part to render a card.
Give those tools a shared `emptyToolArgs` schema with one optional,
ignored field so the schema is non-empty and Gemini calls them. execute
bodies are unchanged; other providers ignore the extra field. This is the
same fix already documented for previewImport.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bump root, backend, and frontend to 0.2.1 and add the 0.2.1 changelog section.
Patch release: pagination styling, patient-sheet header reflow, message
avatars, and CHANGELOG-based GitHub Release notes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bump root, backend, and frontend to 0.2.0 and move the changelog Unreleased
notes under a dated 0.2.0 heading. Adds Patients pagination, per-patient record
history, patient summary PDF export, the messages bubble/attachment UI, the AI
import approval-card fix, and the Docker network-address fix.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add GET /api/activity/patient/:fileNumber (every audited change on one chart,
newest first, readable by any clinic member) and surface it as a Record
history timeline in the patient detail sheet. Add a Download summary action
that builds a clean, printable one-page clinical summary in the browser
(Save as PDF) — no PDF dependency, nothing leaves the server.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
/api/network reported the container's bridge IP (172.x) when running in
Docker, which surfaced as a broken/"Error" value in Settings → About &
updates. Skip container-internal interfaces (detected via /.dockerenv) so the
endpoint returns no address inside a container, and guard the panel against an
unexpected response shape — both paths fall back to the helpful
"open via the server's IP" hint.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Give previewImport a concrete object schema (was z.array(z.unknown())) so
Google Gemini can emit a real function call — an array-of-unknown serializes
to an empty JSON schema, which made Gemini print the call as a `tool_code`
text block instead of invoking the tool. Validation stays lenient in
execute() via patientInputSchema. Also strengthen the system prompt: never
print tool calls/JSON or re-list record fields; keep prose to one sentence.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Retarget the release workflow, docker-compose image refs, and docs from the
placeholder temetro namespace to khalidxv (GitHub repo refs unchanged).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Reference prebuilt temetro/temetro-{backend,frontend} images in docker-compose
(with a build fallback) and stop baking the frontend API URL. Add a
tag-triggered GitHub Actions release workflow that pushes both images and cuts
a GitHub Release, plus RELEASING.md, CHANGELOG.md, and a root version. Overhaul
the root and frontend READMEs (screenshot, features, prebuilt-image quickstart,
LAN access, updating).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add public GET /api/version (current version + GitHub-release update check,
cached, fail-soft) and GET /api/network (detected LAN addresses). Accept
localhost/private-LAN origins in CORS and Better Auth trusted origins via a
shared isAllowedOrigin helper, so staff can reach the app over the clinic
network. The seed-demo.ts fake test data is removed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two changes so off-network wallet scanning actually works:
- Force the cloudflared edge connection over http2 (TCP/443) instead of
QUIC (UDP/7844) in both the Docker tunnel and the dev-tunnel script.
QUIC is blocked on many networks/Docker setups, which left the tunnel
stuck "Failed to dial a quic connection" and the QR pointing at a dead
URL — the root cause of "must be on the same network" failures.
- Gate the discovered quick-tunnel URL on real end-to-end reachability:
poll the tunnel's own /health until it answers (Cloudflare 1033 clears
in ~30s) before publishing it, and have /pair await that discovery so
the QR never carries a not-yet-live URL.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add `npm run docker:tunnel` (docker-compose.tunnel.yml overlay): a
cloudflared sidecar opens a public quick tunnel to the backend, and the
backend auto-discovers its https://…trycloudflare.com URL from cloudflared's
metrics endpoint (services/relay-url.ts) and bakes it into the wallet-import
QR — so a phone on any network can scan, with no install/account/PUBLIC_RELAY_URL.
PUBLIC_RELAY_URL still wins when set. Verified end-to-end: the backend logs
the discovered tunnel URL on startup.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Scan-to-connect only worked on the same Wi-Fi because the QR carried a LAN
relay URL. Give the relay a public address two ways:
- `npm run dev:tunnel` (scripts/dev-tunnel.mjs) opens a cloudflared tunnel
to localhost:4000 and starts the backend with PUBLIC_RELAY_URL set to the
public https URL, so the QR carries it and a phone on any network connects.
- Deploy configs for Fly.io (fly.toml), Render (render.yaml), and Railway
(railway.json), all building the existing Dockerfile.
Also harden resolveRelayUrl to honor x-forwarded-proto so the derived QR URL
is https behind a TLS proxy (iOS ATS requires wss).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The "Import from a patient app → QR" code baked the web app's own
API_BASE_URL (typically http://localhost:4000) into the pairing URI, so a
real phone could never reach the relay and "Scan to connect" silently
stalled. Use the backend's server-resolved, device-reachable relay URL
instead (PUBLIC_RELAY_URL, else the request host), surface it on the
WalletPairing type, and document PUBLIC_RELAY_URL in .env.example.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Backend: nullable wallet_number on wallet_share_requests (migration 0029);
createPairingRequest + POST /api/patients/wallet/pair; applyShareResponse
binds the authenticated wallet on QR (pairing) requests
- Frontend: Import dialog gains a "Show QR" mode rendering a temetro-pair QR
(react-qr-code) the patient scans; requestWalletPairing helper; i18n keys
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- auth: forgot-password now has Email/Username tabs; new public
/api/auth-helpers/reset-by-username resolves a username and hands off to the
existing reset flow (real email or admin-notify fallback)
- messages: conversations expose isSystem; System notices are read-only (no
composer, no call button) and styled distinctly (shield icon + badge)
- meetings: compact, larger control bar; self tile shows real initials and an
avatar (not black) when the camera is off; delete-room UI with confirm
- sidebar: username-only accounts show @username instead of a synthetic email
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Chat: history pill now shows a History icon + a Start-new-chat (SquarePen)
button; removed the duplicate chat-history list from the sidebar.
Email: deployment-wide email provider config (Resend/Postmark/SendGrid/SMTP) in
Settings → Developers, with encrypted API key and a Send-test action. sendEmail
dispatches via the chosen provider (REST via fetch; SMTP via nodemailer).
Forgot password with no provider: alert the clinic admin(s) via a "System"
message card in Messages + a bell notification (seeded system user + per-clinic
System conversation); clicking deep-links to /settings?tab=careTeam&member=<id>.
Admins can set a member's password directly from the employee dialog
(PATCH /api/staff/:id/password via Better Auth's internal context — no admin
plugin needed).
Patient Portal: "New patient" booking path registers a demographics-only patient
then books; bookings reject double-booked slots (409).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
New chrome-less (portal) route group with /portal/[clinic], a card-style choice
of "Book an appointment" vs "View my results", and step flows wired to a new
public backend router (src/routes/portal.ts):
- GET /api/portal/:clinic -> clinic name
- POST /api/portal/:clinic/appointments (verifies name+file#, books a confirmed
appointment that appears on the doctor's Appointments page)
- GET /api/portal/:clinic/results (minimal, safe status — no clinical values)
Excluded /portal from the auth proxy redirect so the kiosk loads without a
session. PHI exposure is intentionally minimal (see docs).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backend: GET /api/settings/records/export downloads the clinic's full patient
archive as JSON; POST /api/settings/records/import creates new patients (skips
existing file numbers, drops cross-clinic provider links). Both admin-gated.
Frontend: Records settings now has a working Export button and an Import flow
(choose file -> preview count -> apply) wired to the new endpoints.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backend: add tasks.assignee_user_id (nullable, fk user) + migration; persist
it through create/update and surface tasks to the assigned person in listTasks.
Frontend: New Task dialog gains a three-way Assignee control
(Myself / Department / Person) with a provider picker from /api/staff/providers;
task card + detail show the assignee's name when set.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Live invite: ring a clinic member into a room (socket call:invite + a bell
notification); the invitee gets a toast with a Join action. Notifications of
type "meeting" deep-link to /messages/meetings?room=, which auto-joins.
- Scheduling: new scheduled_meetings table + /api/meetings/events (list mine,
create, delete); a Calendar tab on the Meetings page with a month picker
(meeting-day dots), the day's agenda, and a Schedule-meeting dialog
(title/date/time/participants).
- Redesign: rounded control bar with tooltips (mic/cam/screen/invite + separated
red Leave), speaking ring on tiles (Web Audio), and live room-occupancy counts
via call:presence broadcasts. Migration 0025.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Chat patient cards are now compact previews (header + key stat + "Click for
more"); cards size to content (items-start) instead of stretching to the tallest;
Edit record moved into the summary detail dialog
- AI citations render once per source per message (collapses per-word spam) and the
prompt now asks the model to cite sparingly (once per paragraph, not per list item)
- Sidebar sub-items use exact-match active state so Meetings no longer lights Inbox
- Analysis defaults to All Time, drops the 12m option, and clamps the chart window
to >=2 points so 30d/Today render instead of collapsing
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backend: meeting_rooms table + /api/meetings (list/create/delete, org-scoped),
and WebRTC mesh signaling over the existing authed Socket.io (call:join with a
≤4 cap and org authorization, call:signal relay, peer-joined/left, disconnect
cleanup). Migration 0024.
Frontend: Messages nav is now expandable (Inbox + Meetings); new
/messages/meetings page with a room list and a Discord-style call UI — a
useWebRtcMesh hook (camera/mic, screen share via replaceTrack, ≤4 mesh) and a
tile grid with a mic/camera/screen/leave control bar. New lib/meetings + i18n.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Retrieval tools now register a PHI-free sourceId per record/list and stream a
data-source part with the real clinician-facing title; the system prompt asks
the model to cite facts inline as [[src:id]]. The chat renders those markers as
numbered inline citation chips (PreviewCard hover) via a Streamdown link
override, with a Sources footer fallback when the model omits markers.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
New staff_profile table (org + user, unique) holding a clinician's clinical
specialty. GET /api/staff and /api/staff/providers now include specialty; new
PATCH /api/staff/:userId (member:update) upserts it. Migration 0023.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
In Graph mode, a natural-language request ("graph for John Smith") went
through the getPatient tool, which always emitted record cards — so the
clinician got Summary/Vitals cards instead of a graph (the mode prompt
even wrongly claimed the graph rendered automatically). Thread the
composer mode into the chat tools and have getPatient emit
data-recordGraph in graph mode (matching the client-side /patient
fast-path), with the cards kept for chat/analysis modes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The import preview only showed counts + skipped-row errors with no way to
fix anything. Now every parsed record is editable before import:
- backend: shared validatePatientImport() (used by the previewImport tool)
+ POST /api/ai/import/validate for dry-run re-validation; the import
preview payload now carries the original records (and the source record on
each invalid row) so the UI can edit them.
- frontend: the import card gets a "Review & edit" dialog listing every
record with a ready / needs-fixing badge; opening one reuses the full
PatientFormDialog in a new non-persisting review mode (editable file
number) and re-validates on save, so skipped rows can be fixed and
included.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Migration prompt now tells the agent to map/normalize an uploaded export
into temetro's shape itself (gender words, non-numeric IDs, delimited
allergy/med/problem lists, per-visit encounters) and to fix-and-re-preview
skipped rows rather than telling the clinician to reformat the file.
- Guarantee a non-empty reply: when the model ends on a tool call with no
text, ask it to summarize (external+Veil path) and fall back to a generic
line; the streaming path appends the same fallback. Raise the step cap to 8.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Real-world exports were rejected wholesale (file number must be digits,
sex must be M/F, every allergy/med/problem/encounter a full object), so
AI imports skipped everything. Normalize at the schema edge instead:
strip non-digits from the file number, map gender words to M/F, accept a
bare string for allergies/medications/problems (filling sensible
defaults), and default missing encounter fields (type → "Visit"). The
manual patient form shares this schema and benefits too.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a real file-storage layer to the backend and wire upload UI into the
frontend.
backend:
- new `attachments` table (org-scoped, links to a patient file number and
optionally a lab result) + Drizzle migration
- `/api/attachments` route: upload (multer → disk under UPLOAD_DIR),
list, stream/download, delete; gated by patient:write OR lab:write via
a new requireAnyPermission helper so lab staff can attach analyses
- UPLOAD_DIR env (default ./uploads) + a persistent docker volume
frontend:
- lib/attachments.ts client (multipart upload, list, delete, preview URL)
- staged file picker in the patient Add/Edit dialog (uploaded after save)
and the lab Add-result dialog (linked to the result)
- a Files section in the patient sheet that lists attachments and opens
them in a preview dialog (images inline, others via download)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The composer's SELECT is now a clinician-facing mode picker instead of a list
of LLM vendors (the model still comes from Settings → AI and continues to drive
the Veil consent gate). The mode travels with each send:
- backend appends an Analysis/Graph directive to the agent's system prompt
- Graph mode renders an Obsidian-style RecordGraph for the /patient fast-path
(new data-recordGraph message part reusing the shared graph component)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add optional startDate/endDate columns to prescriptions (schema,
validation, types, service) — when set, endDate drives expiry.
Add a new append-only `dispenses` resource (schema, types, validation,
service, route at /api/dispenses) recording who received which medication,
gated on the existing inventory RBAC statement. Migration 0020.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Khalid Abdi