This violates your "no code" premise, but if your tags are all on one line, this seems reasonably robust and should be fairly efficient.

#! perl -sw
use strict;

while( <DATA> ) {
    s[^(<([^>]+?)>)(.*)(</\2>)]{
        (my $x = $3) =~ tr[<>][[]];
        "$1$x$4";
    }e;
    print;
}


__DATA__
<sometag> my data here </sometag>
<anothertag> further text </anothertag>
<furthertag> not good as contains <this> in angle brackets </furtherta
+g>
<byetag> another possibility is <more> <than> one <angle pairs> in her
+e </byetag>
<a tag with spaces> and content containing a false </a tag with spaces
+> and a real </a tag with spaces>
[download]

Produces

C:\Perl\test>junk2
<sometag> my data here </sometag>
<anothertag> further text </anothertag>
<furthertag> not good as contains [this] in angle brackets </furtherta
+g>
<byetag> another possibility is [more] [than] one [angle pairs] in her
+e </byetag>
<a tag with spaces> and content containing a false [/a tag with spaces
+] and a real </a tag with spaces>
[download]

Of course, it doesn't attempt to deal with attributes, or multi-line elements or nested tags or any of that good stuff.