specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Autopay
providerId: autopay
generated: '2026-09-06'
method: searched
source: https://developer.autopay.io/booking_api/ and https://developer.autopay.io/status_api/
created: '2026-05-04'
modified: '2026-09-06'
tags:
  - Parking
  - Rate Limiting
  - Quotas
description: >-
  Autopay publishes no account-level quota, no per-minute request ceiling and no rate-limit
  response headers. It documents exactly one numeric limit — a per-booking query limit on the
  Booking status endpoint — plus one unquantified "too many queries" condition on the Status
  API, and one behavioural rule about token minting. Exhaustion is signalled as an error_id in
  an ordinary error body, not as a 429 with headers.
limit_count: 1
headers:
  limit: null
  remaining: null
  reset: null
  retryAfter: null
  policy: null
  note: >-
    No RateLimit-*, X-RateLimit-* or Retry-After response header is documented anywhere in the
    reference. An agent cannot read its remaining budget from the wire; it can only detect
    exhaustion after the fact from the error_id.
responseCodes:
  throttled: null
  note: >-
    No 429 appears anywhere in the Autopay reference. query_limit_error is returned inside the
    normal {error_id, message} envelope; the HTTP status carrying it is not documented.
limits:
  - name: Booking status query limit
    scope: per client, per booking
    operation: GET /booking/v3/{id}/status
    limit: 1
    window: 900 seconds
    metric: queries
    burst: null
    error_id: query_limit_error
    quote: 'One status query is allowed per 900 seconds per client per booking'
    prose_quote: 'There is a 15-minute query rate limit per booking on the GET Get a status of a booking endpoint.'
    source: https://developer.autopay.io/booking_api/
    agent_impact: >-
      A polling agent cannot watch a booking more often than once every 15 minutes. There is no
      event or webhook for booking status, so 15 minutes is the floor on booking-state latency.
undocumented_limits:
  - name: Status API query limit
    operation: GET /status/v1/zone/{zone_code}, GET /status/v1/zone_details/{zone_code}
    error_id: query_limit_error
    documented_explanation: 'Too many queries'
    limit: not published
    source: https://developer.autopay.io/status_api/
    note: The error id exists and is documented; the threshold behind it is not.
  - name: Concurrent booking limit
    error_id: availability_limit_error
    documented_explanation: 'Concurrent booking limit exceeded for that time period'
    limit: not published — set per permit definition; read it from GET /booking/v3/permit_definitions ("their joint access limit") and GET /booking/v3/availability
    source: https://developer.autopay.io/booking_api/
    note: A capacity limit rather than a request-rate limit, but it is the one quota an agent can query before it acts.
  - name: Concurrent parking limit
    error_id: parking_validation_concurrent_limit_exceeded
    documented_explanation: 'Concurrent parking limit is reached'
    source: https://developer.autopay.io/tapnpark_api/
  - name: Fleet services date range ceiling
    error_id: vehicle_services_invalid_date_range
    documented_explanation: 'The specified period exceeds the maximum limit'
    limit: not published
    source: https://developer.autopay.io/fleet-api/
policies:
  - name: Token minting
    description: >-
      "Generating excessive access tokens within the expiration time (i.e. requesting a new one
      for each request) may lead to termination of API access." This is the one throttling rule
      Autopay enforces contractually rather than technically — the penalty is loss of access,
      not a 429. Cache the token for its full expires_in (10-24 hours).
    source: https://developer.autopay.io/authentication/
  - name: Export query pacing
    description: >-
      'Data is not propagated to the Accounting API in real-time; therefore, all the queries
      should be delayed as well and not made in real-time.' Walk the export endpoints by time
      window, carrying the previous request's `to` value forward as the next `from`.
    source: https://developer.autopay.io/accounting/
note: >-
  This file previously carried a five-row Free/Professional/Enterprise limit table (10/100/1000
  rpm, 1k/100k monthly quotas) and a full X-RateLimit-* header set, written by the 2026-05-04
  bulk sweep and never sourced from Autopay. None of it is published by Autopay and it has been
  removed.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com