API reference

https://api.native.no/v1

Every request goes over HTTPS. Plain HTTP is refused.

Send an access token as a bearer token in the Authorization header of every request. You get it with your client ID and secret and the Norn API as the audience, as Authentication shows:

POST /v1/ideas HTTP/1.1
Host: api.native.no
Authorization: Bearer <ACCESS_TOKEN>
Content-Type: application/json

Your credentials belong to your team and reach every brand your team creates, and no others. A token lasts at most an hour: cache it and fetch a new one before it runs out.

• Request bodies are JSON, sent with Content-Type: application/json. Responses are JSON too.
• Field names are snake_case, and times are ISO 8601.
• Ids start with the kind of thing they name: brand_, batch_, idea_, fb_.
• Every object has an object field that names its kind. Lists come back as an object with "object": "list" and the items in data.
• New fields can appear at any time. Write code that ignores the ones it does not use.

An error comes back with an HTTP status and a body that says what went wrong. When it is about one field, param names it:

{
  "error": {
    "type": "invalid_request",
    "code": "brand_not_ready",
    "message": "Norn is still reading https://pipedream.example. Ask again when the brand’s status is ready.",
    "param": "brand"
  }
}

• Your client ID has a request limit during the preview, agreed with you when access opens. Past it, the API answers 429.
• Every call answers within seconds. The slow work, reading a website and thinking up ideas, runs in the background: you get a brand or a batch back at once and ask for it again until it is ready. Ask no more often than every 10 seconds.
• Up to 10 ideas per request, and up to 20 drafts per prediction.

The version is part of the path: /v1. A change that would break your code gets a new version. The model field in each response names the model that answered, so you can tell when Norn itself has changed.