I've been reading threads about this topic and have found next comments:

>> What I need is for the filenames to be 20580.html

> Remove that "index" part completely would be possible and looking at your
BBS the side-effects should be minimal:  files/pages that are mirrored but not
having "index.cgi?read=XXX" are going to have odd filenames (this includes any
images).

> You could try a structure like: h%p/%[read:::file:file].%t

> Then you should get:

> file.htm (the start URL)
> file-2.htm (maybe search page)
> file-3.htm (something else)
> file-4.gif
> 20580.htm (the threads)
> 20579.htm
> etc

I think it would be nicer if it were any possibility about using something
similar to: "%h%p/%[read:::%n:%n].%t"
assuming %n is the filename without extension, as shown in help.

Is there any other way to achieve right results? (thread filenames as numbers
and any other files with their respective "real" names)