To achieve this I would create an xwiki class "shellscript" that contains a
single field, named "shcode" which is pure text. Then I would create velocity
script that dispalys the page or downloads the code based on some http servlet $request
parameters.
I would not try to "dig out" the shell code from the wiki page.
Note that you can access the servlet request as well as response object so you can make
downloading just anything, not only text/html.
Regards
Peter
Verhás Péter
+36(30)9306805
peter(a)verhas.com