The /extract endpoint (v1, pages: [{url}], format: json) consistently
fails on Discourse forum topic pages with "No data returned from
crawlers!". Two independent instances, so it looks Discourse-wide rather
than one site's bot blocking:
Discourse serves crawler-friendly HTML to plain user agents, and the
same topics are indexed by search engines fine. Interestingly,
Discourse's plain-text endpoint for the same content works through
/extract (e.g. https://community.home-assistant.io/raw/1020609), so it's
something about the HTML topic pages specifically — maybe the crawler is
getting the empty JS app shell.
Seen 2026-08-13. Single-page requests, so it's not the batching issue
from my earlier report (d/11176).
I expected the Extract API to return the text from the source.
