generated: '2026-08-29'
method: searched
source: openapi/_original/*.json + https://docs.acceldata.io/api/introduction + https://docs.acceldata.io/api/authentication
docs: https://docs.acceldata.io/api/introduction
scope: >-
  Cross-cutting runtime semantics for the three published Acceldata contracts: Admin API
  (31 ops), Catalog API (104 ops) and Tag Services (5 ops).
auth:
  style: static key pair in headers
  headers:
  - accessKey
  - secretKey
  declared_as: required header parameters on every operation, not an OpenAPI securityScheme
  bearer: false
  oauth: false
  cross_link: authentication/acceldata-authentication.yml
  note: >-
    accessKey and secretKey appear as explicit `in: header` parameters on all 104 Catalog
    operations and all 5 Tag operations. A generated client will therefore expose them as
    per-call arguments rather than as client-level credentials.
base_url:
  templated: true
  form: https://{adoc-host}
  path_prefixes:
  - /catalog-server/api  (Catalog API, and tag associations)
  - /admin/api           (Administration API)
  - /api/tags/v1         (Tag Services)
  source: >-
    https://docs.acceldata.io/api/authentication publishes
    `curl -X GET "https://<HOST>/catalog-server/api/assets"`, and the Python SDK is
    constructed with url="https://<your-adoc-url>". None of the three specs declares a
    servers[] block.
  consequence: >-
    There is no single public API host. Every caller substitutes their own ADOC control-plane
    hostname, supplied by an administrator. An agent cannot discover the host from the
    contract alone.
content_negotiation:
  request: application/json
  response: application/json
  accept_header: required, always application/json
  content_type_header: required on requests with a body
pagination:
  consistent: false
  styles:
  - name: page-size
    applies_to:
    - Catalog API
    - Tag Services
    params:
    - page
    - size
    - sortBy
    operations: 13
  - name: offset-max
    applies_to:
    - Administration API
    params:
    - first
    - max
    defaults:
      first: '0'
      max: '10'
    operations: 6
  docs_guidance:
    quote: 'Use pagination (`limit`, `offset`) to avoid retrieving very large results in one request.'
    source: https://docs.acceldata.io/api/introduction
    note: >-
      The documented guidance names `limit` and `offset`, which are the parameter names of
      NEITHER published style. Three different pagination vocabularies are in circulation
      for one product.
  response_envelope: >-
    Paginated Catalog responses wrap results with a `meta` object (MetaData schema); the
    Administration API returns AllRealmUsersListResponse / GroupsListResponse /
    ServiceUserListResponse wrappers.
filtering_and_lookup:
  dual_addressing: >-
    Assets are addressable by numeric `id` OR by string `uid` (getAssetByIdOrUid). Every
    reliability policy family publishes a matched pair of by-id and by-name operations
    (manage-data-quality-by-id / manage-data-quality-by-name, and the same for
    reconciliation, data drift, schema drift). Name-addressing is a first-class access path,
    not a convenience.
  versioned_reads: >-
    Ten Catalog operations accept a `version` parameter and six accept
    `onlyExecutedVersion`, so policy history is queryable rather than last-write-wins.
field_expansion:
  supported: false
  note: No expand / fields / include parameter appears in any published operation.
metadata:
  custom_metadata: >-
    Assets carry a metadata document (getAssetMetadata) and support custom metadata
    additions via the SDK; tags and labels are the governed extension points.
request_id_tracing:
  header: none published
  note: >-
    Two Catalog operations accept a `requestId` parameter, but no response correlation
    header (X-Request-Id, Traceparent) is declared anywhere. There is no published way to
    correlate a failed call with a support ticket.
versioning:
  api_version_in_path: partial
  detail: >-
    Tag Services versions in the path (/api/tags/v1). The Catalog API and most of the
    Administration API do not (/catalog-server/api, /admin/api); a small slice of the
    Administration API uses /admin/api/v1/service-users. The product itself versions on a
    monthly calendar train (26.3.0 ... 26.8.0) and API changes are published per release.
  spec_versions:
    Admin API: '1.0'
    Catalog API: '1.0'
    Tag Services: 1.0.0
  cross_link: changelog/acceldata-changelog.yml
error_envelope:
  consistent: false
  rfc9457: false
  envelopes: 3
  cross_link: errors/acceldata-problem-types.yml
rate_limit_signaling:
  headers_published: none
  status_on_exhaustion: 429
  documented_only: true
  cross_link: rate-limits/acceldata-rate-limits.yml
  note: >-
    429 is described on the API introduction status-code table but is declared on zero of
    the 140 published operations, and no RateLimit-* or Retry-After header is documented.
idempotency:
  supported: false
  header: none
  scope: null
  retention: null
  evidence: >-
    No Idempotency-Key parameter and no idempotency language in any of the three specs or
    in the published API reference. 45 of 140 operations are POST/PUT/DELETE.
  agent_consequence: >-
    A retried policy creation or tag write can duplicate. The one partial protection is
    422 on triggerDataQualityExecution — "Previous execution of rule '{name}' has not
    completed. Not running the rule." — which makes execution triggering effectively
    single-flight per rule, but that is concurrency control, not idempotency.
dry_run_mode:
  supported: partial
  operations:
  - operationId: preCheckPolicyExports
    surface: Pre-Check Policy Definition Export
    docs: https://docs.acceldata.io/api/pre-check-policy-exports
    note: Validates an export before performing it.
  - surface: Validate Persistence Configuration
    docs: https://docs.acceldata.io/api/validate-persistence-configuration
  - surface: Preview Template
    docs: https://docs.acceldata.io/api/preview-template
    note: Renders a template without applying it.
  note: >-
    Three validate/preview surfaces exist, each scoped to one workflow. There is no general
    dry-run flag on write operations.
reversibility:
  grade: documented
  applicable: true
  note: >-
    A real reversal path exists and is published for two of the four policy families. No
    retention WINDOW is stated anywhere for how long an archived policy remains
    restorable, so this grades `documented` and not `verified`. NOTHING here asserts a
    window Acceldata has not published.
  surfaces:
  - action: delete a Data Quality policy
    operationId: deleteDataQualityRule
    reversal:
      operationId: unarchiveDataQualityRule
      path: /catalog-server/api/rules/data-quality/{id}/unarchive
      method: POST
      permission: ASSET_VIEW + POLICY_MODIFY
    window: not stated
    docs: https://docs.acceldata.io/api/manage-data-quality-by-id
    note: >-
      Delete is described in the contract as "Delete (archive)". ADOC also archives a
      policy AUTOMATICALLY when a column it references is deleted or renamed, and archived
      policies cannot execute — so an agent can lose policy coverage without issuing any
      call at all, and unarchive is the recovery.
  - action: delete a Data Quality policy by name
    operationId: deleteDataQualityByName
    reversal:
      operationId: unarchiveDataQualityByName
      path: /catalog-server/api/rules/data-quality/byName/{name}/unarchive
      method: POST
    window: not stated
  - action: delete a Reconciliation policy
    operationId: deleteReconciliationRule
    reversal:
      operationId: unarchiveReconciliationRule
      path: /catalog-server/api/rules/reconciliation/{id}/unarchive
      method: POST
    window: not stated
  - action: delete a Reconciliation policy by name
    operationId: deleteReconciliationByName
    reversal:
      operationId: unarchiveReconciliationByName
      path: /catalog-server/api/rules/reconciliation/byName/{name}/unarchive
      method: POST
      description_quote: Restores a previously archived reconciliation policy by policy name.
    window: not stated
  irreversible:
  - operationId: deleteDataDriftRule
    note: No unarchive operation is published for data drift policies, by id or by name.
  - operationId: deleteDataDriftByName
    note: No unarchive operation published.
  - operationId: deleteSchemaDriftRule
    note: No unarchive operation is published for schema drift policies.
  - operationId: deleteSchemaDriftByName
    note: No unarchive operation published.
  - operationId: deleteApiKey
    note: API key deletion has no published restore. The secret is shown once at issue time and cannot be re-read.
  - operationId: deleteServiceUserApiKey
    note: No published restore.
  - operationId: removeUser
    note: '"Disables and removes a user from the tenant." No published restore operation.'
  - operationId: deleteGroup
    note: No published restore.
  - operationId: deleteSqlView
    note: No published restore.
  - operationId: deleteUdfTemplate
    note: No published restore.
  - operationId: deleteAssetTag
    note: 'Detach only; re-attachable via addTagToAsset. Low blast radius.'
  asymmetry: >-
    The single most important thing for an agent to know here: DELETE means archive for
    data quality and reconciliation policies, and means destroy for data drift and schema
    drift policies. The operations look identical in the contract.
cross_links:
  errors: errors/acceldata-problem-types.yml
  lifecycle: lifecycle/acceldata-lifecycle.yml
  authentication: authentication/acceldata-authentication.yml
  scopes: scopes/acceldata-scopes.yml
  rate_limits: rate-limits/acceldata-rate-limits.yml
  data_model: data-model/acceldata-data-model.yml