generated: '2026-07-21'
method: searched
source: >-
  https://docs.tenyks.ai/reference/ and https://docs.tenyks.ai/docs/ — the
  cross-cutting request/response semantics of the Tenyks API, plus what
  openapi/tenyks-openapi.json implies.
description: >-
  How the Tenyks API behaves across operations: authentication style, resource
  identity, the cloud-storage credential model, the two-phase upload→ingest
  pattern, versioning, and the error envelope. Tenyks is an alpha API with a
  small, upload-oriented surface, so several conventions common to mature REST
  APIs (documented idempotency-key header, cursor pagination, rate-limit
  headers) are NOT published and are recorded here as not-documented rather than
  invented.
base_url: https://dashboard.tenyks.ai   # or https://sandbox.tenyks.ai for the freemium tier
api_style: REST over HTTPS, JSON request/response bodies.
authentication:
  scheme: Bearer access token, exchanged from an API key + secret.
  detail: authentication/tenyks-authentication.yml
  token_ttl_seconds: 3600
resource_identity:
  mechanism: >-
    Datasets and models are addressed by a client-supplied unique `key`
    (e.g. dataset_key "face_detection", model_key "yolo_v8"), not a
    server-generated opaque id. Re-using a key targets the same resource.
  note: >-
    This key-based identity gives creates a natural upsert/dedupe seam, but
    Tenyks does not document a dedicated Idempotency-Key request header, so no
    formal idempotency contract is asserted.
data_ingestion_pattern:
  supported: true
  description: >-
    Uploads are two-phase: first upload (annotations, predictions, images) from
    a cloud-storage location, then call the corresponding `/ingest` endpoint to
    start asynchronous processing. Ingestion is a background job.
  operations:
    - openapi/tenyks-openapi.json#uploadDatasetAnnotations
    - openapi/tenyks-openapi.json#ingestDatasetAnnotations
    - openapi/tenyks-openapi.json#uploadModelPredictions
    - openapi/tenyks-openapi.json#ingestModelPredictions
cloud_storage_credentials:
  supported: true
  providers: [aws_s3, gcs, azure]
  description: >-
    Data is referenced by a storage URI (e.g. s3_uri) plus per-request
    credentials (AwsCredentials: access key id, secret access key, region), so
    Tenyks pulls data directly from your bucket. GCS and Azure follow the same
    shape (see the GCS/Azure workflow reference pages).
pagination:
  style: not-documented
  note: The published Retrieve datasets endpoint returns all datasets without documented pagination parameters.
versioning:
  scheme: single-version
  current: "1.0.1"
  status: alpha
  note: >-
    "This API is currently in its alpha version and may undergo changes."
    (docs.tenyks.ai/reference/introduction). No dated versioning/deprecation
    policy is published — see lifecycle/tenyks-lifecycle.yml.
error_envelope:
  shape: >-
    Endpoints document a 200 successful response and a 400 "Bad request" for
    invalid input; the API returns application/json. No RFC 9457
    problem+json media type is declared. See errors/tenyks-problem-types.yml.
rate_limit_signaling:
  style: not-documented
cross_links:
  authentication: authentication/tenyks-authentication.yml
  errors: errors/tenyks-problem-types.yml
  lifecycle: lifecycle/tenyks-lifecycle.yml
  data_model: data-model/tenyks-data-model.yml