Skip to main content
The WISEPIM Public API lets your own systems read and update a project’s catalog: an ERP that pushes prices, a storefront that reads enriched content, or a sync job that keeps another PIM in step. Everything the API does, it does on one project, the project the API key is bound to. The API covers 17 catalog resources, including products, categories, attributes, families, vocabularies, prompts, knowledge, dictionaries and import/export templates. You can create, read, update and delete resources, manage their relationships, and launch enrichment, translation and bulk actions with trackable jobs. Start with catalog entities for resource management or actions and jobs for background processing. The endpoint reference and OpenAPI schema define the available operations; template management does not itself run an import or export. For assistant connections, see Connected Apps.

Quickstart

1

Create an API key

In the WISEPIM app, open Settings > API Access and click Create API key. Pick a name and the permissions the integration needs. The full key is shown once; store it in a secrets manager. See API Access for who can create keys and the limits.
2

Check the key

GET /me needs no scope and tells you which project the key is bound to and which scopes it carries.
3

Page through your products

Lists are keyset-paged. Pass next_cursor back as cursor until has_more is false.
4

Write back

Update one product with PATCH /products/{product_id}, or send up to 200 products at once to POST /products/batch, keyed by SKU. Both need the products:write scope.

Endpoints

Paging

Every list returns the same envelope and is ordered by id ascending:
  • limit defaults to 50 and is capped at 200.
  • The cursor is opaque. A malformed cursor is a 422 invalid_cursor, never a silent restart from the top.
  • There is no total count. Drain the list until has_more is false.
  • Keyset paging visits every pre-existing row exactly once, even while products are being inserted.
For incremental sync, call GET /products?updated_since=<ISO-8601> on a schedule, page to the end, then store the highest updated_at you saw as the next updated_since.

Errors

Errors use a JSON envelope. When present, code is a stable machine-readable string; branch on it rather than on error. Request-schema validation errors may instead contain a per-field detail array without a code.
Quote the request_id when you contact support about a failed call.

Rate limits

Each key has its own limit, 600 requests per minute by default, counted in a fixed 60-second window. Every response carries X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset (seconds until the window resets). A request that fails a scope check still counts. Support can raise a key’s limit when a sync needs it; GET /me shows the limit the key currently has.

Retries and idempotency

Networks fail mid-request, and a client that retries a POST after a timeout cannot know whether the first attempt was applied. Send an Idempotency-Key header on every mutation, including resource creation, updates, deletes, action launches and job cancellation and reuse the same key for every retry of that write.
  • The key is any string of 1 to 128 printable ASCII characters without whitespace. A UUID per logical operation works well.
  • A retry with the same key, method, path, query and JSON body returns the original status and body and applies nothing again. The order of keys in the JSON body does not matter.
  • Keys are scoped to the API key that sent them and are kept until that API key is deleted. Rotating a key starts a fresh scope.
  • The same key with a different request is a 409 idempotency_conflict; a malformed key is a 422 invalid_idempotency_key. A write that fails leaves no record, so the next attempt with the same key runs normally.
  • A replayed response describes the resource or job when the request ran. Use its GET endpoint for current state. Deletes return 204 with no response body, including on replay. Authorization is checked again before replay.
  • GET requests need no idempotency header (they still require the API key). POST /products/batch is keyed by SKU, so a retried batch without a header is also safe, but a header gives you the original results and summary back instead of a second run.
The API does not retry anything on your behalf. Retry 5xx responses and timeouts with the same key; honour Retry-After on a 429 before retrying.

Test safely

There is no separate sandbox host. Test against a separate project instead: create a project, fill it with a copy or a sample of your catalog, and create a key bound to that project. Keys cannot reach any other project, so a bug in your integration never touches the catalog your live channels read from. Compare Projects shows the differences between the test project and the live one before you point the integration at production.

Authentication and scopes

How keys work, which scopes exist, and how to rotate a key you think has leaked.

API Access settings

Create, rename and revoke keys in the app, and the limits per project and per person.

Developer portal

Every developer resource in one place: spec, MCP server, llms.txt, guides.

Connected Apps

Let ChatGPT or Codex work with your catalog through a scoped, revocable connection.