Thanks. You seem to be connecting to some proprietary server using S3 protocol.
The server pages the directory contents by 1000 objects/files.
The directory contains exactly 18000 files, so exactly 18 pages.
With the actual Amazon S3 server, the last (18th) page would say:
<IsTruncated>false</IsTruncated>
while your server says:
<IsTruncated>true</IsTruncated>
So WinSCP reads one more "page".
The server returns an empty page.
That's another difference to the actual Amazon S3 server, which never returns an empty page, unless you are asking for listing of directory/prefix that actually
does not exist. WinSCP gets confused here by these two differences and hence the
"File or folder ... does not exist" error.
I can improve WinSCP not to error, if it gets an empty page, unless it is the first page.
But I'm not sure how your server responds to an actual empty directory/prefix. Can you send a log for that?