generated: '2026-08-12'
method: derived
source: >-
  openapi/_original/mntn-ptv-openapi.json, openapi/_original/mntn-reporting-api-openapi.json,
  openapi/_original/mntn-batch-export-api-openapi.json
docs:
- https://api.mountain.com/docs
- https://api3.mountain.com/docs
- https://help.mountain.com/en/collections/13086998-apis
auth:
  style: 'API key (X-API-Key header on api.mountain.com, key query parameter on api3.mountain.com) or bearer JWT'
  detail: authentication/mntn-authentication.yml
idempotency:
  supported: false
  header: null
  evidence: >-
    No Idempotency-Key, Idempotency-Token, X-Idempotency-Key or equivalent parameter or header appears
    in any of the three OpenAPI definitions, and no idempotency guidance exists in the help center.
  affected_operations:
  - campaigns.create
  - campaigns.launch
  - campaigns.pause
  - creatives.create
  - AudiencesController_create_v1
  - audiences.geoLists.create
  - pmp.dealGroups.create
  - pmp.deals.create
  - pmp.partners.create
  - reporting.submitExport
  - submitExport
  note: >-
    Eleven state-changing POST operations -- including campaign launch and pause, which spend money --
    have no retry-safety contract. A client that times out on campaigns.launch cannot safely retry.
    No type: Idempotency pointer is wired in apis.yml because the support is genuinely absent.
pagination:
  styles:
  - api: MNTN Performance TV (PTV) API
    style: page-number
    params: [page, perPage]
    used_by_operations: 12
    sort_params: [sortBy, sortOrder]
    search_param: search
    note: 'page/perPage appear on 12 list operations; no cursor, no Link header, no declared page-size ceiling.'
  - api: MNTN Async Batch Export API
    style: offset-limit
    params: [limit, offset]
    used_by_operations: 1
    operation: listExports
  - api: MNTN Reporting API 3.0
    style: none-declared
    note: >-
      The /apidata description advises using "pagination (limit / offset) where supported" but the
      OpenAPI declares only query, key and format parameters -- the pagination contract is prose, not spec.
  inconsistency: >-
    Two different pagination idioms across two APIs from the same vendor (page/perPage vs limit/offset),
    with a third surface where pagination is documented in prose only.
field_expansion:
  supported: true
  param: expand
  used_by_operations: 6
  api: MNTN Performance TV (PTV) API
  note: An expand query parameter on six operations lets a caller inline related resources.
filtering:
  ptv: 'per-operation query parameters -- status, objective, campaignIds, segmentTypes, partnershipDealType, geoTypeId, geoTypeName, companyName, companyUrl'
  reporting: >-
    JSON-encoded filter expressions in the query parameter, for example
    {"campaigninfo.id":{"in":["200"]}}, with And / Or / Not composition schemas declared in the spec.
  note: The reporting filter language is a real declared grammar (And, Or, Not, Filter, FilteredColumn, Value, SimpleColumn).
column_naming:
  scheme: fully qualified table.column identifiers
  examples: [campaigninfo.name, graph.impressions, graph.day]
  tables: 'Info tables carry dimensions, Summary tables carry metrics, joined on shared keys'
  source: 'openapi/_original/mntn-reporting-api-openapi.json info.description'
content_negotiation:
  mechanism: format query parameter (not Accept header)
  values: [json, csv, human, excel]
  media_types:
  - application/json
  - text/csv
  - text/plain
  - application/vnd.openxmlformats-officedocument.spreadsheetml.sheet
  note: >-
    Format is selected by query parameter rather than content negotiation, so a standard HTTP client's
    Accept header is ignored.
request_id_tracing:
  supported: true
  field: traceId
  location: error response body
  format: OpenTelemetry trace ID
  note: >-
    Present on the RFC 9457 error envelope for api3 only. There is no request-id request header a
    caller can set, and no correlation identifier on successful responses.
error_envelope:
  api3: 'RFC 9457 application/problem+json with type, title, status, detail, instance, timestamp, errorCode, traceId, errors[]'
  ptv: 'ad-hoc {"error": "<message>"} application/json; 78 of 85 operations declare only a bare default response'
  detail: errors/mntn-problem-types.yml
rate_limit_signaling:
  headers_declared: false
  status_on_exhaustion: 429
  scope: 'POST /batch only'
  retry_after: false
  detail: rate-limits/mntn-rate-limits.yml
  note: >-
    No X-RateLimit-*, RateLimit-* or Retry-After header is declared on any response in any of the
    three specs, so an agent has no runtime backoff signal.
async_pattern:
  name: enqueue-poll-signed-download
  steps:
  - 'POST /batch -> 202 Accepted with batchId and initial status'
  - 'GET /batch/{batchId} -> status; when SUCCEEDED includes a time-limited signed download URL'
  - 'POST /batch/{batchId}/regenerate-url -> reissue a signed URL without re-running the query'
  - 'GET /batch -> list recent jobs with status, limit, offset'
  entitlement: r2ds.exports.enabled
  trigger: 'a 413 Payload Too Large on the synchronous /apidata call'
  mirrored_on_ptv: 'the same shape exists as reporting.submitExport / getExportStatus / regenerateExportUrl under /api/v1/reports/batch'
versioning:
  detail: lifecycle/mntn-lifecycle.yml
  summary: host-based major versions (api / api2 / api3) plus path versioning (/api/v1/) on the platform API
webhooks:
  supported: false
  evidence: >-
    No webhooks object in any OpenAPI, no callbacks, no event or subscription endpoints, and no
    webhook documentation in the help center. Outbound integrations are pull-based ETL connectors.