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>
- new org_ai_policy table + GET/PUT /api/ai/policy (read for any member,
write owner/admin only); migration 0018
- /api/chat hard-blocks (403) when AI is off for the caller
- Settings → AI "Availability" section: enable AI, or disable for
employees only (owners/admins keep access); read-only for non-admins
- sidebar, command palette and route guard hide/redirect the AI chat
when it's disabled for the current user
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Stop blocking AI imports/proposals on missing non-critical fields. Records
the chat agent drafts now save with safe placeholders, auto-generated file
numbers, and a source="ai" marker that surfaces an "Added by AI" badge so a
clinician can review/edit them later.
Backend:
- add `source` (manual|ai) column to patients/appointments/prescriptions
(migration 0014) + canonical types, services, validation schemas
- relax patient/appointment validation: empty file number allowed, demographic
+ type/provider/initials fall back to placeholders (initials derived from name)
- patients.generateFileNumber() auto-assigns an MRN when one is missing
- proposeAppointment accepts a name when no file number resolves; AI commits +
/api/ai/import stamp source="ai"
Frontend:
- `source` on Appointment/Patient/Prescription types; AI commits send source="ai"
- reusable <AiBadge> shown on the Patients table/detail and prescriptions list
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Real LLM chat replacing the mock, backend-centric per the plan:
- Multi-provider API-key mode (OpenAI / Anthropic / Gemini via the AI SDK) plus
local Ollama (OpenAI-compatible endpoint). Provider is derived from the
picked model id; the matching stored key is used. New user_ai_settings table
holds per-user config with provider API keys encrypted at rest (AES-256-GCM,
src/lib/crypto.ts, keyed by AI_CREDENTIALS_KEY).
- POST /api/ai/config (get/put, secrets never returned), POST /api/ai/test
(Ollama ping / key presence), POST /api/ai/import (approved migration commit,
re-validated server-side, reuses the audited patient service).
- POST /api/chat: streamText agent with tools (getPatient, getPatientLabs,
searchPatients, previewImport). Real record data streams to the clinician as
custom data parts (cards) while the model sees only Veil-redacted results.
- Veil (src/services/ai/veil.ts): de-identifies patient identifiers to tokens
before external calls, resolves tokens on tool args, and rehydrates the final
answer. Bypassed for local Ollama. External mode runs non-streamed so the
rehydrated text is correct. Every call is audited (provider + Veil level).
- Shared role-scoping helpers extracted to src/lib/role-scope.ts (reused by the
patient routes and chat tools so visibility rules match).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Khalid Abdi