Securing a Publicly Accessible Folder

This website contains links to software which is either no longer maintained or will be supported only until the end of 2019 (CKFinder 2). For the latest documentation about current CKSource projects, including software like CKEditor 4/CKEditor 5, CKFinder 3, Cloud Services, Letters, Accessibility Checker, please visit the new documentation website.

If you look for an information about very old versions of CKEditor, FCKeditor and CKFinder check also the CKEditor forum, which was closed in 2015. If not, please head to StackOverflow for support.

When integrating CKFinder, you will often want to give users access to uploaded files, so they can insert images or links to files into the edited content. This requires to make the folder publicly accessible, so all the files are served through the web server. If you rely on your web server to serve the files uploaded with CKFinder, you should take additional steps to make sure the files are served in a secure way.

Let us assume that you have configured your CKFinder to allow uploading of avi files. Even if the avi file is then served with a valid Content-Type: video/x-msvideo header, some browsers may ignore this information and perform additional checks on the raw file contents. If any HTML-like data is detected in the file content, the browser may decide to ignore information about the content type and handle the served content as if it was a regular web page. This behavior is called content sniffing (also known as media type sniffing or MIME sniffing), and in some circumstances, it may lead to security issues (for example, it may open door for XSS attacks).

To avoid content sniffing, you should make sure that your server adds the X-Content-Type-Options: nosniff header to all HTTP responses when serving files from the publicly available folder. The X-Content-Type-Options response HTTP header is a marker used by the server to indicate that the MIME type set by the Content-Type header should not be changed and should be followed. As a result, the browser does not perform any content sniffing on the received content.

This page was last edited on 21 August 2019, at 12:56.