> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mantrixflow.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Strapi source

> Connect Strapi 5 with a read-only API token, validate dynamic Content API collections, preview records, and run Full Table extraction.

Strapi is Available Now as a source-only connector for Strapi 5. MantrixFlow
reads selected REST Content API collections, normalizes nested records with
dlt, stages each resource in DuckDB, applies published SQL transformations,
and delivers the result to a configured destination.

The connector never creates, updates, deletes, publishes, or unpublishes Strapi
content. It does not use the admin API, execute GraphQL mutations, upload media,
or download media binaries.

<Warning>
  Strapi 4 is not supported. The connector expects the flattened Strapi 5
  response shape and rejects the older `attributes` envelope rather than
  silently producing an incorrect schema.
</Warning>

## Before you begin

You need:

* a Strapi 5 project reachable through an HTTPS origin;
* a **Read Only** API token, or a custom token with `find` and `findOne`
  permission for every collection you intend to sync; and
* the Content API ID for each collection, such as `articles`, `authors`, or
  `product-categories`.

The Strapi URL is the origin only:

```text theme={"theme":{"light":"github-light","dark":"github-dark"}}
https://cms.example.com
```

Do not append `/api`. MantrixFlow constructs `/api/{api_id}` server-side.

## Create the source connection

1. Open **Connections** in your MantrixFlow workspace.
2. Click **+ New Connection** and choose **Source**.
3. Select **Strapi**.
4. Enter a descriptive connection name.
5. Enter the Strapi HTTPS origin without `/api`, a path, query, or fragment.
6. Enter the read-only API token.
7. Click **Test Connection** and save the connection.

The initial connection test validates the URL and host reachability. After a
Content API ID is configured, MantrixFlow also requests one record from that
collection to verify the token and its content permission. Host reachability
alone does not prove access to every collection.

The token is encrypted at rest and returned to the browser only as a masked
value. Leaving that masked value unchanged during an edit preserves the stored
token.

## Add Content API collections

Strapi projects define their own content types, so there is no universal stream
catalog. Configure the collections used by this pipeline:

1. Create or open a pipeline that uses the Strapi connection.
2. Open the **Source** tab.
3. In **Strapi content types**, enter a Content API ID such as `articles`.
4. Click **Add** and repeat for the other required collections.
5. Click **Save content types**.
6. Discover the catalog, preview each collection, enable the required streams,
   and save the stream configuration.

Content API IDs accept lowercase letters, numbers, and single hyphens. Enter
`product-categories`, not `/api/product-categories` or a full URL. MantrixFlow
rejects paths, traversal, admin/plugin routes, arbitrary query strings, and
endpoint URLs.

Each collection has a stable staging relation:

```text theme={"theme":{"light":"github-light","dark":"github-dark"}}
strapi__<api_id>
```

For example, `articles` is staged as `strapi__articles`.

## Sync behavior

Strapi currently supports **Full Table** only. Each run paginates every selected
collection visible to the token and replaces that collection's staging
resource. Unselected collections receive no extraction requests.

Incremental sync and CDC are disabled. MantrixFlow does not yet advertise an
`updatedAt` cursor or merge key because `documentId`, locale variants, inclusive
timestamp boundaries, and failed-run recovery require additional live Strapi 5
verification.

Normal collection requests use published content by default. Draft ingestion
is not currently exposed as a pipeline option.

## Record shape and nested data

Strapi 5 root records preserve fields returned by the Content API, including
`id`, `documentId`, `createdAt`, `updatedAt`, `publishedAt`, and `locale` when
present. dlt normalizes nested objects and arrays such as components, rich-text
blocks, and dynamic zones into related staging tables.

MantrixFlow does not automatically add `populate=*`. Relations and media fields
contain only the data Strapi returns for the authorized collection request.
Media metadata can pass through the source payload, but the connector does not
follow or download media URLs.

Use the exact root and child table names shown by discovery and preview when
writing a transformation. A minimal article model can start with:

```sql theme={"theme":{"light":"github-light","dark":"github-dark"}}
SELECT
  document_id,
  title,
  locale,
  published_at
FROM {{ source('raw', 'strapi__articles') }}
```

Column normalization depends on the actual Strapi payload. Copy the column
names from the transformation editor rather than assuming the example names.

## URL and network security

Hosted MantrixFlow requires HTTPS and blocks localhost, loopback, link-local,
private, reserved, and cloud-metadata network addresses. DNS answers and
redirects are validated before requests, and redirects must remain on the same
validated origin under `/api/`.

Private-network Strapi deployments require an explicit self-hosted MantrixFlow
network policy. A browser-supplied endpoint cannot weaken the hosted SSRF
policy.

## Pagination and retries

The connector uses explicit Strapi page/page-size pagination and treats
`meta.pagination.page` and `pageCount` as authoritative. The default page size
is conservative, and request/page budgets prevent unbounded extraction.

Retries are bounded for network failures, `429`, `502`, `503`, and `504`.
`Retry-After` is honored when Strapi or an upstream proxy returns it. Invalid
queries, authentication failures, permission errors, and missing content types
fail without endless retries.

## Verify a run

After starting a pipeline, open **Runs** and confirm:

* every selected `strapi__<api_id>` resource is listed;
* extracted counts match the published records visible in Strapi;
* **Rows failed** is `0`;
* nested component or dynamic-zone child tables contain the expected rows; and
* each published transformation reaches its intended destination table.

## Current limitations

* Strapi is Available Now, source-only, and supports Strapi 5 only.
* Full Table is the only sync mode; Incremental and CDC are disabled.
* Only collection types are advertised. Single types are not supported.
* Draft selection, locale selection, and explicit relation population are not
  exposed yet.
* OpenAPI import and automatic content-type discovery are not implemented.
* Media binaries are never downloaded.
* Private fields omitted by Strapi remain unavailable.
* Manual live Strapi-to-destination validation remains pending and is not an
  independent production-certification claim.

## Troubleshooting

| Symptom                                | Resolution                                                                                                             |
| -------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- |
| Strapi is missing from Connections     | Confirm the Go API and ELT runtime report the Strapi source capability as available, then refresh.                     |
| The URL is rejected                    | Enter an HTTPS origin such as `https://cms.example.com` without `/api`, another path, credentials, query, or fragment. |
| Host reachable, token not verified     | Add a Content API ID and confirm the token has `find` permission for that collection.                                  |
| Authentication fails                   | Replace the API token with a valid Read Only or correctly scoped custom token.                                         |
| A collection returns `403`             | Grant `find`/`findOne` for that content type to the token.                                                             |
| A collection returns `404`             | Confirm the exact plural Content API ID generated by Strapi.                                                           |
| Content API ID is rejected             | Use lowercase letters, numbers, and single hyphens only; do not enter `/api` or a URL.                                 |
| Relations or media details are missing | Strapi does not populate every relationship automatically. The current connector does not request `populate=*`.        |
| Draft records are absent               | Published content is the current default and draft selection is not exposed.                                           |
| SQL cannot find the source             | Use the exact `strapi__<api_id>` staging relation shown in the transformation editor.                                  |

See the [Strapi 5 documentation](https://docs.strapi.io/) for Content API and
API-token administration guidance.