Scribe

Create scribe job

Creates an organization-scoped scribe job in `UPLOADING` status, validates that the templateId and optional patientId belong to the authenticated organization, stores the supplied audio reference, and immediately attempts to enqueue asynchronous scribe processing.

POSTPublic reference baselineFile or artifactScribeoperationId: createScribeJobhttps://dev-api.quickintell.com/api/v1/scribe/jobs

Public reference baselineFile or artifactPrior auth and clinical

Route status

Public reference baseline

Generated endpoint contract may be shown with caveats, but external publication still needs route registry and release evidence.

Publication status

Baseline reference only

Requires E6 route registry and E9 release evidence before external publication. Also requires E2 side-effect policy and E4 file and artifact governance before production recipes, SDK inclusion, or try-out access.

Authentication

Bearer API key declared

OpenAPI declares bearer API-key security for this operation.

Tenant context

API-key organization

API-key organization context is authoritative; request organizationId is not a public tenant selector.

Side-effect summary

File or artifact

Touches documents, reports, transcripts, PDFs, EDI, or generated artifacts. Metadata is safer than raw content.

Source artifact

docs/openapi/openapi.json

OpenAPI 3.1.0; API 1.0.0.

When to use this endpoint

Use this page as an internal baseline for the prior auth and clinical contract generated for POST /api/v1/scribe/jobs. External use still depends on route classification, release evidence, and endpoint-specific side-effect approval.

What this does now

Documents the generated POST /api/v1/scribe/jobs contract as a file or artifact baseline endpoint.

What this does not do

Does not prove external GA readiness, SDK readiness, live production execution, or complete module behavior. Mutating behavior still needs side-effect and retry evidence.

Not public from this page

Live external side effects, file contents, credentials, webhooks, SDK promises, and production recipes are still gated.

Headers and tenancy

Current headers and tenant authority. Target headers stay labeled as future until OpenAPI and release evidence declare them.

Header	Required	Current status	Notes
Authorization	Yes	Current	Bearer <QUICKRCM_API_KEY>. OpenAPI security: bearerAuth.
Idempotency-Key	No	Target/future	Do not claim universal header support until E3 and OpenAPI metadata exist.
X-Request-Id	No	Target/future	Use for support correlation only after approved; not declared as a standard OpenAPI header today.

Required scopes

Unclassified until E6 route metadata exists; do not infer scopes from route names.

Wrong-org behavior

Tenant-safe 404, empty list, or 403 is endpoint-specific and remains an evidence gap until E1 is closed.

Actor and audit

No public actor, audit, or correlation contract is published from OpenAPI alone.

Idempotency and retries

Retry behavior is documented conservatively because universal idempotency headers are not declared.

Topic	Status	Guidance
Current idempotency	No idempotency mechanism is declared.	Do not retry blindly. Check resource, status, batch, or job state before repeating the request.
Universal header	Not declared	Do not document universal Idempotency-Key support until OpenAPI and E3 evidence declare it.
Conflict behavior	Unresolved	Same-key changed-payload, in-flight retry, fingerprint scope, and TTL behavior remain endpoint-specific evidence gaps.

Async jobs and polling

Queued work is accepted work, not proof of completed workflow execution.

Topic	Status	Guidance
Immediate response	No 202 response declared	OpenAPI does not declare HTTP 202 for this operation.
Polling and terminal states	Endpoint-specific	Use only operation-specific status endpoints visible in OpenAPI; generic job behavior, cancellation, retries, and retention are unresolved.
Later side effects	Not proven	Do not infer background execution beyond the documented response.

Files and artifacts

File-like routes need content-class, ownership, scanning, retention, and URL policy evidence before public recipes.

Topic	Status	Guidance
Content class	Potential file or artifact	Documents, PDFs, reports, transcripts, EDI, OCR, or generated artifacts need E4 classification before public examples.
Ownership and storage keys	Evidence required	Tenant ownership must be proven before retrieval or upload docs. Never expose storage keys as public identifiers.
URLs, scanning, and retention	Unresolved unless endpoint-specific evidence exists	Short-lived URLs, MIME limits, checksums, scanner status, retention, expiration, and legal hold behavior are not implied by OpenAPI examples.

Publication warnings

Universal Idempotency-Key header support is not declared; retry only after checking endpoint-specific state.
File and artifact examples must stay metadata-only until scanning, retention, ownership, and URL policy are approved.

Related platform rules

Workflow Guide Route Status Tenant Context Errors Rate Limits PHI and Security Publication Readiness Side-Effect Modes Idempotency Files and Artifacts

Evidence

Evidence	Status	Caveat
OpenAPI artifact	docs/openapi/openapi.json; OpenAPI 3.1.0; API 1.0.0.	Generated baseline contract only.
Route registry	Public reference baseline from documentation classification overlay.	E6 route registry evidence is still required before external publication.
Example validation	Generated examples are internal baseline aids.	No external publication claim; examples must pass no-PHI/no-secret/no-raw-payload/no-storage-key checks.
Last reviewed	Implementation run 20260619T-docs-impl; source boardroom packet 20260619T064820Z.	Refresh this evidence for release workspaces.
Release evidence	E9 required.	No SDK readiness, production readiness, or expanded external docs claim from this endpoint page.

Scribe public API.

Scribe public APIs expose organization-scoped clinical documentation job and template workflows for the AI Clinical Scribe module. The existing module docs describe Scribe as an ambient AI scribe that transcribes encounters and drafts SOAP notes; the public API surface narrows that product workflow to ten bearer-authenticated endpoints for jobs, file metadata attachment, asynchronous processing, attestation, and reusable templates. Tenant context comes from the bearer API key. Public callers should not send `organizationId` or `orgId` tenant selectors. Read endpoints require Scribe read-capable credentials (`scribe:read` or `scribe:write`), while create, update, attach, attest, process, and template-write endpoints require `scribe:write` credentials. The job lifecycle is asynchronous. `createScribeJob` creates a local job in `UPLOADING` status and immediately attempts to enqueue processing with the supplied audio reference. status=QUEUED`. Neither endpoint performs inline transcription, inline note generation, or synchronous clinical validation. The public contract is intentionally conservative around clinical data and files. List responses return job summaries and `hasFinalNote` indicators, not raw transcripts, final notes, audio URLs, signed URLs, storage references, or vendor payloads. Detail and write responses can include `finalNote`, so examples and guidance must treat it as clinical documentation that can contain Protected Health Information (PHI). File attachment records metadata for an already-uploaded organization-scoped audio file and returns sanitized file metadata; it does not upload file bytes, create presigned URLs, or echo storage references. Template endpoints manage reusable prompt configuration and should not be documented as patient-specific note content.

What this endpoint does

Use case

Use this endpoint when an integration already has an organization-authorized audio reference and wants job creation plus first processing enqueue in one call. Use attachScribeJobFile plus queueScribeJobProcessing when the job already exists or the audio reference is attached separately.

Before calling

Authenticate with a Scribe write-capable API key. Obtain an organization-scoped scribe template from listScribeTemplates or createScribeTemplate, and ensure the fileUrl is authorized for the same tenant. If patientId is supplied, it must identify a patient in the authenticated organization.

Request and response behavior

Request guidance

`fileUrl` and `templateId` are required. `fileUrl` is stored as the job audio reference but is never echoed in public API responses. `fileUrl` is capped at 2048 characters, `language` defaults to `en` and is capped at 20 characters, `annotate` defaults to false, and `executionMode` defaults to `AI_ONLY`. Do not include raw audio bytes, transcripts, tokens, signed URL secrets, storage credentials, or broad clinical context in this JSON body.

fileUrl can identify clinical audio access and should not be logged or embedded in public examples as a real URL.
templateId is required and must be organization-scoped.
Create-job already queues processing; do not tell callers that queueScribeJobProcessing is always required after a successful create response.
executionMode must be `AI_ONLY` or `AI_WITH_HITL`.

Response semantics

HTTP 201 means the job was created and the asynchronous processing message was accepted. The returned job has local status `UPLOADING`, not a completed transcript or note. The response can include finalNote as string or null, but create responses normally return null; fileUrl, raw transcripts, audio URLs, signed URLs, storage references, and vendor payloads are not returned.

The successful response omits the submitted fileUrl.
job.status is `UPLOADING` immediately after successful create and enqueue.
finalNote is nullable and should be treated as clinical documentation when present.
Creation is asynchronous queue acceptance, not inline transcription or note generation.

Errors and retries

Treat 400 as invalid request shape, 404 as a missing or wrong-organization template or patient reference, 401/403 as credential or tenant authorization failure, 429 as a backoff signal, and 5xx as a server or queueing failure. Runtime handler evidence shows a 503 created-but-not-queued failure when enqueueing fails after the job record is created, but the generated OpenAPI response list currently documents 500 rather than an explicit 503. Retrying a failed or timed-out POST can create duplicate jobs unless the client performs external de-duplication.

404 can mean the referenced template or patient was not found in the authenticated organization.
The queue-failure path is runtime-observed as 503 even though generated OpenAPI currently rolls unexpected server failures into 500.
Retried POST calls can create duplicates unless the client coordinates retry behavior externally.

Template boundary before examples

The route class, authentication state, tenancy rule, side-effect mode, idempotency status, async behavior, file policy, warnings, links, and evidence above are part of the endpoint contract. Treat generated examples below as internal baseline aids until release evidence closes.

Path parameters

This endpoint does not define path parameters.

Query parameters

This endpoint does not define query parameters.

OpenAPI-declared header parameters

Standard platform header guidance appears in the governance panel. This table only lists operation-specific headers declared in OpenAPI.

This endpoint does not define operation-specific header parameters.

Request body

JSON fields from the generated OpenAPI schema with pilot enrichment applied where available. Nested object fields are shown with dotted paths.

Name	Location	Required	Type	Description	Validation and safety notes
fileUrl	body	Yes	string	Organization-authorized audio file URL or reference for transcription. Stored by the workflow but not echoed in public API responses. Maximum length is 2048 characters.	Minimum length: 1. Maximum length: 2048. File, artifact, or raw-payload adjacent; keep examples metadata-only.
templateId	body	Yes	string	Required organization-scoped scribe template identifier.	Minimum length: 1.
patientId	body	No	string	Optional organization-scoped patient identifier. Treat as PHI-adjacent workflow context.	Minimum length: 1. Potential PHI; use synthetic examples and avoid logging raw values.
language	body	No	string	Language hint for transcription and note generation. Defaults to `en` and is capped at 20 characters.	Minimum length: 1. Maximum length: 20. Default: en. File, artifact, or raw-payload adjacent; keep examples metadata-only.
annotate	body	No	boolean	Boolean processing hint. Defaults to false.	Default: false.
executionMode	body	No	AI_ONLY \| AI_WITH_HITL	Scribe execution mode. Valid values are `AI_ONLY` and `AI_WITH_HITL`.	Enum values are declared in the Type column. Default: AI_ONLY.

Generated request example

Generated examples are baseline contract aids only. They are not external publication evidence and must not include real PHI, credentials, raw vendor payloads, raw EDI, transcripts, storage keys, or signed URLs.

{
  "fileUrl": "https://example.quickintell.com/resource",
  "templateId": "00000000-0000-4000-8000-000000000001",
  "patientId": "00000000-0000-4000-8000-000000000001",
  "language": "en",
  "annotate": false,
  "executionMode": "AI_ONLY"
}

Successful responses

Generated response examples show the baseline OpenAPI contract. Tenant identifier echo, PHI redaction, pagination, money, date/time, and enum semantics remain endpoint-specific publication checks.

201

Created scribe job.

Generated from the OpenAPI response schema.

{
  "success": true,
  "data": {
    "job": {
      "id": "00000000-0000-4000-8000-000000000001",
      "organizationId": "00000000-0000-4000-8000-000000000001",
      "status": "RECORDING",
      "patientId": "00000000-0000-4000-8000-000000000001",
      "appointmentId": "00000000-0000-4000-8000-000000000001",
      "templateId": "00000000-0000-4000-8000-000000000001",
      "template": {
        "id": "00000000-0000-4000-8000-000000000001",
        "name": "Example scribe_job",
        "specialty": "example-specialty"
      },
      "audioDurationSeconds": 1,
      "executionMode": "AI_ONLY",
      "hasFinalNote": true,
      "createdAt": "2026-06-08T10:15:30Z",
      "updatedAt": "2026-06-08T10:15:30Z",
      "finalNote": "Example scribe_job note"
    }
  },
  "meta": {
    "organizationId": "00000000-0000-4000-8000-000000000001"
  }
}

Error responses

Errors must remain sanitized. Do not include raw payer, EHR, SFTP, SMTP, Stripe, LLM/OCR, browser automation, file, transcript, EDI, worker, or vendor payloads.

400

Invalid request.