Contentful delivery

Contentful’s own paths, over the same content the Strapi surface reads. A site written against the CDA needs its base URL changed and nothing else.

GET /spaces/:space/environments/:env/entries
GET /spaces/:space/environments/:env/entries/:id
GET /spaces/:space/environments/:env/content_types

The environment-scoped spellings are served too, for clients configured with a base URL that already carries the environment:

GET /spaces/:space/entries
GET /spaces/:space/entries/:id
GET /spaces/:space/content_types

Spaces and environments

The :space segment selects a content space. An unknown space is 404, and the message lists the spaces the tenant declares.

The :env segment is accepted and is not a selector. A Contentful environment is a copy of a space’s content, which on this platform is a tenant, and the tenant is already established by the request’s CEPT headers. Accepting the segment keeps every client URL working; treating it as a second tenancy axis would let a token choose its own data.

Reading entries

`content_type` is required

GET /spaces/blog/environments/master/entries?content_type=post

A type-less query is 400. There is no single table to read, and fanning out across every content type would turn one request into one query per type. The refusal says so rather than answering with an empty array.

A by-id fetch does not need one, because the id is a document identity:

GET /spaces/blog/environments/master/entries/01J9Y…

The type is found by asking each of the space’s types for that document, which is what the CDA’s own /entries/{id} does.

Query parameters

Parameter	Default	Notes
`content_type`	required on `/entries`	Type name or plural
`locale`	the row’s own	`*` pivots every locale into one entry
`limit`	`100`	Clamped to `1000`
`skip`	`0`
`include`	`1`	Link resolution depth. Above `10` is `400`
`order`	`(documentId, locale)`	`-` prefixes a descending term
`select`	every field	Comma-separated

Filters

Field equality and bracketed operators, in the CDA’s spelling:

?fields.slug=hello
?fields.price[gte]=300
?fields.slug[in]=hello,second
?sys.id=01J9Y…

Operator	Meaning
`ne`	not equal
`in` `nin`	membership
`lt` `lte` `gt` `gte`	ordered comparison
`exists`	presence
`match`	substring

A field the type does not declare is 400.

Ordering

?order=fields.title
?order=-sys.createdAt,fields.title

A leading minus is descending.

Selecting fields

?select=fields.title,fields.slug

sys is always rendered, so naming it does not restrict the projection and is not an error. The columns sys is assembled from are projected whether or not they were listed.

The entry envelope

Every entry is sys, which the platform owns, plus fields, which the author owns.

{
  "sys": {
    "type": "Entry",
    "id": "01J9Y…",
    "space": { "sys": { "type": "Link", "linkType": "Space", "id": "blog" } },
    "contentType": { "sys": { "type": "Link", "linkType": "ContentType", "id": "post" } },
    "locale": "en",
    "createdAt": "2026-09-15T09:12:44Z",
    "updatedAt": "2026-09-15T09:14:02Z",
    "firstPublishedAt": "2026-09-15T09:14:02Z",
    "publishedAt": "2026-09-15T09:14:02Z",
    "revision": 1
  },
  "fields": { "title": "Hello world", "slug": "hello", "price": 100 }
}

sys.id is the document identity, and it is the same value as the Strapi surface’s documentId. An entry resolved through a link by one client is findable by the other under the same name.

A published entry carries revision, firstPublishedAt and publishedAt. A draft carries none of them, which is how a client tells them apart without a status field.

The collection envelope

{
  "sys": { "type": "Array" },
  "total": 2,
  "skip": 0,
  "limit": 100,
  "items": [ /* entries */ ],
  "includes": { "Entry": [ /* linked entries, once each */ ] }
}

includes is present only when links were resolved.

Link resolution

The CDA does not inline a linked entry. The field holds a Link, and the entries themselves arrive once in a top-level includes.Entry[] however many fields point at them.

"fields": {
  "title": "Hello world",
  "author": { "sys": { "type": "Link", "linkType": "Entry", "id": "01J9A…" } }
}

That shape is why include is a depth rather than a list of names: the client walks links against the includes block.

Four properties are worth knowing.

The block is flat at every depth. A post’s author and that author’s publisher are siblings in includes.Entry[], joined by the Link the author carries. Nesting the publisher inside the author would put it out of reach of a client that follows links by id.

An entry appears once. A page of twenty-five posts by the same author carries one author.

Resolution is batched across each level. Every parent sharing a relation at level n contributes its keys to one query, so the cost is depth × distinct relations and never a function of the number of rows. Twenty-five posts, three authors and two publishers is three queries.

A cycle terminates. A model where a page’s parent is a page resolves to exactly the depth asked for.

Each link is resolved through the target type’s own read rule, so include never becomes a way to read a type the caller is not entitled to.

The locale pivot

locale=* collapses a document’s locale rows into one entry whose localized fields are maps of locale to value, which is how Contentful stores content natively.

{
  "sys": { "type": "Entry", "id": "01J9Y…" },
  "fields": {
    "title": { "en": "Hello world", "de": "Hallo Welt" },
    "price": 100
  }
}

Fields marked localized: false are rendered as plain values, since they are the same across every locale by construction. Wrapping them would tell a client they vary.

sys.locale is omitted on this view, because a pivoted entry is not in one locale.

The pivot gathers a document’s rows and is the call to reach for in management and migration tooling. ?locale=en is the delivery path, and is a single-row read.

Preview

The CDA models preview as a separate host with its own token. One instance serves both here, so the draft view is selected by any of:

Signal	For
`X-Contentful-Preview: true`	A client that can set a header
A `/preview/` path prefix	A client pointed at a preview base URL
A preview token	A front end with no editorial credential

Drafts require permission to edit the type, or a token minted by somebody who has it.

Content type discovery

GET /spaces/blog/environments/master/content_types

{
  "sys": { "type": "Array" },
  "total": 1, "skip": 0, "limit": 1,
  "items": [
    {
      "sys": { "type": "ContentType", "id": "post",
               "space": { "sys": { "type": "Link", "linkType": "Space", "id": "blog" } } },
      "name": "post",
      "displayField": "title",
      "fields": [
        { "id": "title", "name": "title", "type": "Symbol", "required": true, "localized": true },
        { "id": "body",  "name": "body",  "type": "Text",   "required": false, "localized": false },
        { "id": "price", "name": "price", "type": "Integer","required": false, "localized": false },
        { "id": "hero",  "name": "hero",  "type": "Link",   "required": false, "localized": false }
      ]
    }
  ]
}

Field types are mapped into Contentful’s vocabulary, which is what a client’s codegen reads:

Declared	Rendered
`string`	`Symbol`
`text`	`Text`
`int`, `bigint`	`Integer`
`float`, `decimal`, `money`	`Number`
`bool`	`Boolean`
`datetime`, `date`, `time`	`Date`
`json`	`Object`
`array`	`Array`
`file`	`Link`

The CMS columns are not listed as fields, since they belong to sys.

Errors

{
  "sys": { "type": "Error", "id": "NotFound" },
  "message": "The resource could not be found.",
  "requestId": ""
}

Status	`id`	When
`400`	`BadRequest`	`content_type` missing, or a malformed query string
`400`	`InvalidQuery`	An unknown field, or `include` above the ceiling
`403`	`AccessDenied`	Well formed, and the caller is not entitled
`404`	`NotFound`	Unknown space, unknown type, or no such entry