projectId and are listed,
created, and modified through paths scoped to it — for example:
Why every path is project-scoped
Even read-only endpoints take a{projectId}. This is intentional:
- Security — your key’s
projectScopeis checked against the{projectId}in the URL on every request, with no global fallback. - Cardinality — listing across all projects would scan unbounded amounts of data. Project-scoped listing keeps every query bounded.
- Multi-tenancy — most businesses on MarketBox run multiple projects
(per brand, per region, per environment). Forcing the
{projectId}into the path keeps that explicit at every call site.
["*"]) can hit any project the owning user has
access to, but the {projectId} is still required in the URL.
How tenantId relates
The API key carries a userId (the human or service account that minted
it). On every write, MarketBox tags the record with tenantId = userId.
Project membership is a separate concept managed in the dashboard:
“who owns / collaborates on this project.” The API key inherits its
allowed projects from the owning user’s membership — narrowed further by
the key’s projectScope.
Concretely:
Resolving project ids
Project ids are minted in the dashboard. You can also retrieve them via the dashboard UI; a public API endpoint that lists projects is not part of the open API surface — by design, projects are a control-plane concept, not a data-plane one.What’s not in a project
A handful of endpoints are unversioned and project-agnostic:
Every other endpoint — for example
/v1/projects/{projectId}/bookings —
requires a valid API key and a matching project scope. The one exception is
the browser beacon, POST /v1/projects/{projectId}/events/beacon: it takes a
publishable key that belongs to the project instead. See
Report site events.