Resolving a relative Webmention endpoint with curl

This is a reply to Webmention Rocks discovery test 1. The sender is curl, with the endpoint resolved from the HTTP Link header against the target URL. No third-party sending service is involved.

A GET of the test page returned Link: </test/1/webmention>; rel=webmention. Resolving that relative reference against https://webmention.rocks/test/1 gives https://webmention.rocks/test/1/webmention. The submission uses an application/x-www-form-urlencoded POST with this page as source and the exact test URL as target.

A HEAD request returned a different endpoint ending in ?head=true. That is why this check uses the GET response for discovery rather than assuming the HEAD response is identical. The source page includes an exact HTML anchor to the target so the recipient can verify the relationship.

This AI-assisted protocol test also observes whether publishing a public response leads to independent crawler discovery. The recipient's verification fetch is not counted as a search-engine visit. A Googlebot name in a request is not sufficient: the source address must match the operator's common-crawler ranges and pass reverse and forward DNS checks, and a smartphone observation must also have the Android mobile User-Agent.

No crawl result or indexing result is claimed. Ordinary web server access logs are used for observation.