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: Agave
providerId: agave
created: '2026-05-04'
modified: '2026-08-30'
generated: '2026-08-30'
method: searched
source: https://docs.agaveapi.com/agave-api/rate-limits
docs:
- https://docs.agaveapi.com/agave-api/rate-limits
- https://docs.agaveapi.com/agave-api/headers
reconciled: true
tags:
- Accounting
- Construction
- Integration
- Rate Limiting
description: >-
  Agave publishes numeric rate limits and returns runtime rate-limit headers. This artifact replaces
  an earlier generated one that stated "Agave does not publish numeric rate limits" — that was wrong:
  https://docs.agaveapi.com/agave-api/rate-limits has carried both an Agave-imposed ceiling and a
  per-source-system table. Agave layers its own limits on top of each connected source system's, and
  surfaces BOTH in the response headers, which is the useful part for an agent: a caller can read its
  remaining Agave budget and its remaining Procore/QuickBooks/Intacct budget from the same response.
limit_count: 4
responseCodes:
  throttled: 429
  serviceUnavailable: 503
headers:
- name: Agave-RateLimit-Total
  description: The number of requests allowed in a 60-second window.
- name: Agave-RateLimit-Remaining
  description: The number of requests remaining in the 60-second window before a 429 is returned.
- name: '{SourceSystem}-RateLimit-Total'
  description: >-
    Total requests allowed against the underlying source system until its limits reset (typically 60
    seconds). Example observed in the docs - Procore-RateLimit-Total.
- name: '{SourceSystem}-RateLimit-Remaining'
  description: Remaining requests against the source system until its limits reset.
- name: '{SourceSystem}-RateLimit-ResetsAtTimestamp'
  description: UNIX timestamp at which the source-system rate limits reset.
retry_after_header: false
retry_after_note: >-
  Agave does not document a Retry-After header. The documented client guidance is exponential backoff
  on 429.
limits:
- name: Agave per-linked-account request rate
  scope: linked-account
  window: 60s
  limit: 150
  unit: requests
  metric: requests per minute per Linked Account
  applies_to: all source systems
  source: https://docs.agaveapi.com/agave-api/rate-limits
- name: Agave global concurrency
  scope: client
  limit: 15
  unit: concurrent requests
  metric: concurrent requests across all Linked Accounts
  applies_to: >-
    Foundation, QuickBooks Desktop, Sage 100 Contractor, Sage 300 CRE, Viewpoint Spectrum, Viewpoint
    Vista
  source: https://docs.agaveapi.com/agave-api/rate-limits
- name: Agave per-linked-account concurrency
  scope: linked-account
  limit: 10
  unit: concurrent requests
  metric: concurrent requests per Linked Account
  applies_to: >-
    Foundation, QuickBooks Desktop, Sage 100 Contractor, Sage 300 CRE, Viewpoint Spectrum, Viewpoint
    Vista
  source: https://docs.agaveapi.com/agave-api/rate-limits
- name: Serial-execution source systems
  scope: linked-account
  limit: 5
  unit: queued requests
  metric: >-
    Requests are processed serially with a 90-second maximum execution time and a maximum of 5 queued
    requests
  applies_to: QuickBooks Desktop, Sage 100 Contractor, Sage 300 CRE
  source: https://docs.agaveapi.com/agave-api/rate-limits
source_system_limits:
- system: Acumatica
  limit: Varies by plan (License Monitoring Console)
- system: Autodesk Build / BIM 360
  limit: 50-1000 requests per minute depending on endpoint
- system: CMiC
  limit: Not published by CMiC; defaults to Agave's 150/min per linked account
- system: Foundation
  limit: Not applicable; defaults to Agave's 150/min per linked account
  concurrency: 10 per linked account / 15 total
- system: Procore
  limit: 3,600 requests per hour, per user
- system: QuickBooks Online
  limit: 500 requests per minute and 10 concurrent requests, per linked account
- system: QuickBooks Desktop
  limit: Serial execution, 90s max execution, max 5 queued
  concurrency: 10 per linked account / 15 total
- system: Sage 100 Contractor
  limit: Serial execution, 90s max execution, max 5 queued
  concurrency: 10 per linked account / 15 total
- system: Sage 300 CRE
  limit: Serial execution, 90s max execution, max 5 queued
  concurrency: 10 per linked account / 15 total
- system: Sage Intacct
  limit: 900 requests per 18 seconds (1,800 per minute)
- system: Viewpoint Spectrum
  limit: 10,000 write requests per day
  concurrency: 10 per linked account / 15 total
- system: Viewpoint Vista
  limit: Not applicable; defaults to Agave's 150/min per linked account
  concurrency: 10 per linked account / 15 total
introspection:
  endpoint: /rate-limits
  description: >-
    Agave documents a rate limits endpoint that returns the CURRENT rate limits of a Linked Account,
    so a client can read its budget rather than infer it.
  source: https://docs.agaveapi.com/agave-api/rate-limits
policies:
- name: Exponential backoff
  description: >-
    Agave's published guidance - "We recommend adding an exponential backoff to your requests to
    ensure retry requests do not get rate-limited."
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com