A 202 counts as a challenge only when it answers a read

Yesterday's change counted any 202 answered to a crawler as a captcha challenge. That was right for the page, robots.txt and sitemap requests it was written for, and wrong for one case: a 202 to a POST is the ordinary "accepted" answer. On a WordPress site, Googlebot posts to a plugin endpoint while it renders the page, the endpoint answers 202 every time, and Start here told the site that Googlebot was not getting in cleanly.

What changed

• A 202 counts as a challenge (challenge_basis of status_202) only when it answers a GET or HEAD, or a request whose method the feed did not record. A 202 to a POST, PUT, PATCH or DELETE is served.
• The bots access page, Start here, the public API and MCP all read the same rule.