feat: AI-added records save with placeholders + "Added by AI" provenance

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>

backend/src/lib/patient-validation.ts

@@ -1,7 +1,12 @@
 import { z } from "zod";
 
+import { initialsFromName } from "./initials.js";
+
 const nonEmpty = z.string().trim().min(1);
 
+const EMPTY_VITALS = { bp: "", hr: "", temp: "", spo2: "", takenAt: "" };
+const EMPTY_TREND = { label: "", unit: "", points: [] as number[] };
+
 export const allergySchema = z.object({
   substance: nonEmpty,
   reaction: nonEmpty,
@@ -49,28 +54,44 @@ export const trendSchema = z.object({
 
 // A full patient payload — the frontend form sends the entire record on both
 // create and edit, so the same schema covers both.
-export const patientInputSchema = z.object({
-  fileNumber: z.string().trim().regex(/^\d+$/, "File number must be digits"),
-  name: nonEmpty,
-  age: z.number().int().min(0).max(150),
-  sex: z.enum(["M", "F"]),
-  pcp: z.string(),
-  // Optional link to the responsible clinician (user id). Empty string ⇒ null.
-  primaryProviderId: z.preprocess(
-    (v) => (v === "" ? null : v),
-    z.string().nullable().optional(),
-  ),
-  status: z.enum(["active", "inpatient", "discharged"]),
-  initials: z.string().trim().min(1).max(4),
-  allergies: z.array(allergySchema).default([]),
-  alerts: z.array(z.string()).default([]),
-  medications: z.array(medicationSchema).default([]),
-  problems: z.array(problemSchema).default([]),
-  vitals: vitalsSchema,
-  vitalsTrend: trendSchema,
-  labs: z.array(labSchema).default([]),
-  labTrend: trendSchema,
-  encounters: z.array(encounterSchema).default([]),
-});
+//
+// Tolerant by design: an AI import from a sparse export (e.g. just a name) still
+// validates. The file number may be empty (the patient service auto-generates
+// one), demographics fall back to safe placeholders, initials are derived from
+// the name, and the clinical sections default to empty. Such rows are stamped
+// `source: "ai"` and surfaced with an "Added by AI" badge for later editing.
+export const patientInputSchema = z
+  .object({
+    fileNumber: z
+      .string()
+      .trim()
+      .regex(/^\d*$/, "File number must be digits")
+      .default(""),
+    name: nonEmpty,
+    age: z.coerce.number().int().min(0).max(150).default(0),
+    sex: z.enum(["M", "F"]).default("M"),
+    pcp: z.string().default(""),
+    // Optional link to the responsible clinician (user id). Empty string ⇒ null.
+    primaryProviderId: z.preprocess(
+      (v) => (v === "" ? null : v),
+      z.string().nullable().optional(),
+    ),
+    status: z.enum(["active", "inpatient", "discharged"]).default("active"),
+    initials: z.string().trim().max(4).default(""),
+    allergies: z.array(allergySchema).default([]),
+    alerts: z.array(z.string()).default([]),
+    medications: z.array(medicationSchema).default([]),
+    problems: z.array(problemSchema).default([]),
+    vitals: vitalsSchema.default(EMPTY_VITALS),
+    vitalsTrend: trendSchema.default(EMPTY_TREND),
+    labs: z.array(labSchema).default([]),
+    labTrend: trendSchema.default(EMPTY_TREND),
+    encounters: z.array(encounterSchema).default([]),
+    source: z.enum(["manual", "ai"]).default("manual"),
+  })
+  .transform((v) => ({
+    ...v,
+    initials: v.initials || initialsFromName(v.name),
+  }));
 
 export type PatientInput = z.infer<typeof patientInputSchema>;