This is a fun one, which I believe indicates a bug in Kagi's indexing.
If you search for "curl send multiple request using the same connection" (without the quotes, and using these exact words including the singular "request"), one of the top results is a page titled "multiple POST requests while reusing the TCP connection":
Kagi says this search result is at http://example.com/1.jpg (and the link takes you there), which is obviously not correct since example.com has never hosted any such content.
I'm guessing the page that was actually indexed was either the original post from the curl mailing list at https://curl.se/mail/archive-2013-11/0006.html, or the superuser post at https://superuser.com/questions/670321/curl-multiple-post-requests-while-reusing-the-tcp-connection, or maybe more likely the archived post at https://curl-users.cool.haxx.narkive.com/Ui8M4gYz/multiple-post-requests-while-reusing-the-tcp-connection (because the title matches exactly).
But for some reason, if was attributed to http://example.com/1.jpg, probably because this is the URL that is used in the curl command that is given as an example on the indexed page. So it seems to point to a parsing bug in the indexing engine.
The search result should link to the real page that was indexed, not some unrelated URL that happens to be used somewhere on the page.
