Schedules scrape-and-create: the processing pipeline fetches the URL and creates the source entity WITH its processed_content — there is no "create empty, fill later" path. Returns 202 {job_id, status: "queued"}; poll GET /jobs/{job_id} until completed, then find the source via GET /sources?filter[url]=<url>. Attaching to a product is a data_sources reference write on the product (PATCH /products/{id}).

Scope: write:sources.