Find free times
Free times at the key's own organization, from the same slot engine its booking page uses, so a time this returns is one POST /api/v1/reservations or POST /api/v1/appointments will accept (until someone else books it). Shared: the organization's engine decides the question.
- A table-reservations venue:
dateandpartySize. - An appointments business:
dateandserviceId, optionallyproviderId(omit it for any available team member; each slot then names theproviderIdit was found for).
days walks forward from date, up to 14 days in one call. Each day carries its slots and, for a venue, the reason an empty day is empty (closed, full, too_large and so on), so a fully booked day and a closed one never look alike.
Appointment start times sit on one grid per day (behaviour change, 2026-10-13): every appointmentSlotMinutes from the business's opening time that day, or from midnight when it has no opening hours. A team member whose hours start off that grid joins it at the next tick, so with any available team member the slots no longer mix, say, 9:00 and 9:10 starters. The public MCP's find_appointment_times answers from the same engine and changed the same way.
Slot times are the venue's wall clock wearing a Z, exactly like startsAt on a booking. Send one back to create unchanged; never convert it, or it moves by the venue's whole offset.
Also limited per calling IP, 240 requests a minute, by the booking page's own limiter: a 429 here can come from either limit, and both carry Retry-After.
Authorization
bearerApiKey availability:readBuild with a test key (sk_test_...): it works only in your API sandbox and never emails or texts anyone. Scopes, errors and limits.
In: header
Scope: availability:read
Query Parameters
The first day, YYYY-MM-DD, on the venue's calendar.
dateHow many days from date.
1 <= value <= 141Required for a table-reservations venue.
1 <= value <= 30Required for an appointments business.
uuidAppointments only. Omit for any available team member.
uuidResponse Body
application/json
application/json
application/json
application/json
application/json
application/json
application/json
curl -X GET "https://example.com/api/v1/availability?date=2019-08-24"{ "data": { "businessType": "appointments", "banner": "string", "days": [ { "date": "2019-08-24", "slots": [ { "startsAt": "2019-08-24T14:15:22Z", "providerId": "4834bcdc-4a64-444d-966b-1a6fe381da24", "periodName": "string" } ], "reason": "string", "dateNote": "string", "closedReason": "string" } ] }}Read the organizationGET
The calling organization's own profile and settings, and the operation that makes the rest of the API navigable: `businessType` is how a program learns which engine it is talking to, and so which set of endpoints applies. Shared, so it answers for both engines. A **singleton**: the envelope is `{ data: { … } }` with an object, and there is no `nextCursor`, `limit` or `cursor`. Every v1 response is still unwrapped the same way, from `data`. `timezone` is the organization's IANA zone. The wall-clock times on `GET /api/v1/service-periods` are expressed in it, and `startsAt`/`endsAt` on bookings carry its offset for their date. `currency` and the `deposit*` fields are what a program needs to quote a deposit without guessing. There is deliberately **no `organization:write` scope**: the slug, status and engine are not something a long-lived, program-held credential should be able to change.
Take yourself off the waitlistPOST
The button on the page a waitlist email links to. `token` is the signed value in that link and names one entry. A POST because mail scanners prefetch links, so nothing else may change the entry.