generated: '2026-07-19'
method: searched
source: openapi/evermuse-ingest-v1-openapi-original.yml
docs: https://docs.evermuse.com/
authentication:
  style: api-key-header
  header: x-api-key
  key_format: em_sk_
  detail: authentication/evermuse-authentication.yml
idempotency:
  supported: true
  header: Idempotency-Key
  scope: /api/v1/ingest
  max_length: 256
  semantics: |-
    Optional. If a request with the same key and identical body has already been accepted,
    the original response is returned without reprocessing the data. Recommended when
    retrying failed submissions.
  source: openapi/evermuse-ingest-v1-openapi-original.yml#ingestRecords
deduplication:
  keys: [_type, _vendor_ids, _event_at]
  semantics: |-
    Records identified by the same _type, _vendor_ids, and _event_at are recognized as
    the same record; a re-send overwrites the previous version instead of duplicating.
pagination:
  style: cursor
  request_params: [limit, cursor]
  limit_default: 20
  limit_max: 100
  response_field: nextCursor
  order: newest-first
  source: openapi/evermuse-ingest-v1-openapi-original.yml#listBatches
content_types:
  request:
  - application/json
  - application/x-ndjson
  detail: |-
    JSON accepts an array of records or a { "records": [...] } wrapper. NDJSON (one JSON
    object per line) is recommended for high-volume sends. All records in one request must
    share the same _type and _product_id; max 1,000 records; max 5 MB body.
error_envelope:
  shape: { error: string, message: string, retryAfter: number, details: object }
  format: custom-json
  note: Not RFC 9457 problem+json. See errors/evermuse-problem-types.yml.
rate_limiting:
  signal: 429 response with `retryAfter` (seconds) in the JSON body
  detail: lifecycle/evermuse-lifecycle.yml
versioning:
  scheme: uri-path
  current: v1
  base_path: /api/v1
async_processing:
  model: batch
  status_endpoint: /api/v1/ingest/batches/{batchId}
  states: [LANDING, LANDED, PROCESSING, COMPLETE, FAILED]