Yes (provided I've understood the question correctly).
Here is the question:
If I want to adapt the code for this web page to host copyrighted content which cannot be redistributed, can I?
If you look inside the GitHub archive, you'll find the entire site there. All of it (markup and its existing contents) is licensed under CC-BY-SA 2.0.
Since this is a copyleft license, you're allowed to adapt it. So whatever you reuse from that site (HTML, CSS, Javascript, images, decorations, etc.) must be made available under CC-BY-SA. Whatever you're left with after doing this is your website template. To simplify things, you may want to create a fork on GitHub and make it clear that the template in this fork is a derivative of the Code for America repo, and that it is made available under CC-BY-SA 2.0.
However, after going through this process you can download your fork from GitHub. You can then proceed by adding your own contents to it. Since this original contents is not derived from anything in the Code for America repo, it is not a derivative work, and you're not obliged to license this original content under CC-BY-SA.
And to expand the answer a little, this not only applies to CC-BY-SA, but to all copyleft licenses I am aware of.
This again boils down to the legal definition of "derivative work".
The contents you place on a web-page does not make the composite (template + contents) a derivative of the template. The reason is that the template and the contents does not require each other to function. You could replace your contents with some other contents, and the template would still work. You could use some other template to display your content, and it would still be the same contents.
Just for the record, I have discussed a similar situation with the FSF, where the use case was a template available under the GNU GPL, and a client wanted to be sure that his proprietary images was not "infected" by the GPL virus. This was the reply from the FSF:
It is the position of the FSF that if a work licensed under the terms of the GNU GPL isn’t dependent on a particular image in any way, that is, the software would effectively function identically with any other image, then the image could be considered as being distributed in mere aggregation with the work. (Source: Private communication.)
(While this is about the GNU GPL and not the CC-BY-SA, I think it is applicable, as the "ShareAlike" provisions of the CC is very similar to the "Copyleft" of the FSF licenses.)
In legal terms, this low level of functional integration between the two means that the composite is not a derivative, but a "mere aggregate". Even strong copyleft licenses, such as the GNU GPL, does not make the copyleft license of a template or CMS apply to the content that is published with the help of the template or CMS.
This also applies in the physical world. If you write a novel, your publisher will probably hire a graphical designer to create the cover art, decide on page format, fonts, etc. Your novel, however, will not be a derivative of the graphical designer's artwork - and vice versa.