Explore emitting a standard machine-readable header that identifies the page as derived output and links back to the Abstract Wikipedia source.

The non-dismissible banner is how we express provenance to human readers; the open question this sub-bullet captures is whether there is a matching standard way to express the same provenance to machines (crawlers, search engines, archive tooling, structured-data consumers), so that "this page is the output of another page" is recorded in a form a consuming tool can actually act on. The leading candidates to evaluate are:

• schema.org's isBasedOn property on the rendered page's CreativeWork entity, expressed via JSON-LD or microdata.
  
  This is the most semantically precise match — isBasedOn is defined as "a resource from which this work is derived or from which it is a modification or adaption" — and schema.org markup is already consumed at scale by the major search engines.
• <link rel="via">, defined in the IANA link relations registry, which is a lightweight HTTP/HTML-level hint that this resource was obtained through another resource.
  
  Less semantically precise than isBasedOn but cheaper to emit and with a broader if shallower consumer base.
• Dublin Core dcterms:source, for completeness and because some archival tooling still reads it.

If one of these (or another standard that emerges during the investigation) is a clean fit for "this page is the output of an abstract article on abstract.wikipedia.org", emit it alongside the other page-level metadata covered in the mainspace rendering sub-bullet. If none is a clean fit, record that finding and do not invent a bespoke header: a non-standard <meta name="abstractwiki-source"> tag would have no consumers and would only add noise. This is an exploration sub-bullet, and its deliverable is either "header emitted" or "investigation documented as concluding nothing fits", whichever is honest.