generated: '2026-08-11'
method: searched
source: openapi/qr-code-crafter-openapi-original.json#/components/schemas/GenerateQrRequest/x-agent-rate-limit
docs:
  - https://qrcodecrafter.com/qr-code-api-pricing
  - https://qrcodecrafter.com/ai.txt
model: cost-weighted soft budget per client
limit_count: 5
note: >-
  The prose surfaces (pricing page, ai.txt) describe the SHAPE of the limit but never state a number.
  The numbers are published — inside the OpenAPI, on a vendor extension:
  components.schemas.GenerateQrRequest.x-agent-rate-limit carries the window, both budget ceilings, the
  retry headers, and a full per-format cost table. That is the machine-readable rate-limit contract most
  providers never write down, and it is recorded verbatim below. A reader who only skims the pricing
  page would wrongly conclude the limits are undocumented.
limits:
  - scope: per-client
    surface: QR generation (downloadQr, generateQr, generateVerifiedQr, generateBulkQr)
    window: 60
    window_unit: seconds
    limit: 180
    limit_authenticated: 600
    unit: cost-budget units
    burst: null
    cost_units:
      svg: 1
      png: 3
      eps: 3
      jpg: 4
      webp: 4
      logo: 4
      pdf: 6
      sizeOver2048: 2
      sizeOver3072: 4
    cost_note: >-
      Costs compose — a 3072px PDF with an embedded logo bills 6 + 4 + 4 = 14 units against the budget.
      At the anonymous ceiling that is roughly 12 such requests per minute, against 180 plain SVGs.
    uplift: >-
      A valid issued X-Agent-Api-Key raises the ceiling from 180 to 600 units per 60s window when key
      enforcement is enabled for the deployment.
    guidance: >-
      Provider's own words: "Prefer SVG for high-volume tests, throttle batch jobs, and honor
      Retry-After before retrying 429 or 503 responses."
  - scope: per-request
    surface: generateBulkQr
    limit: 50
    unit: rows per request
    note: Hard cap. Also bounded by a total generated-ZIP size ceiling enforced with 413.
  - scope: per-workflow
    surface: create_dynamic_qr_campaign
    limit: 5
    unit: records per campaign
    note: >-
      Executed as sequential POSTs with no automatic retry; partial success returns a private manifest.
  - scope: per-deployment
    surface: dynamic QR storage
    limit: 100000
    unit: active records
    note: >-
      Hard active-record capacity. Exhaustion returns 503 on createDynamicQr. Vaults hold up to 50
      child records each; daily analytics buckets are retained 90 days with a 100,000 daily count cap.
response_headers:
  - name: Retry-After
    description: Seconds to wait before retrying.
    emitted_on: [429, 503]
  - name: X-RateLimit-Limit
    description: Cost budget units available in the current window for this client class.
    emitted_on: [429]
  - name: X-RateLimit-Remaining
    description: Cost budget units remaining in the current window.
    emitted_on: [429]
exhaustion:
  statuses: [429, 503]
  guidance: >-
    ai.txt instructs agents to back off on 429 or 503, honor Retry-After, and read X-RateLimit-Limit /
    X-RateLimit-Remaining "when present" before retrying batch or agent jobs. The 429 responses in the
    OpenAPI declare all three headers explicitly, which is better than most specs in the catalog.
x-evidence:
  - url: https://qrcodecrafter.com/.netlify/functions/generate-qr?value=https%3A%2F%2Fexample.com&format=svg
    http_status: 200
    fetched: '2026-08-11'
    observed: >-
      No X-RateLimit-* or RateLimit-* headers on an unthrottled 200 — only X-QR-Width/Height/Unit/Bytes.
      The budget signal is reactive (returned at exhaustion), not proactive.
  - url: https://qrcodecrafter.com/qr-code-api-pricing
    http_status: 200
    fetched: '2026-08-11'