This site is the archived OWASP Foundation Wiki and is no longer accepting Account Requests.
To view the new OWASP Foundation website, please visit

CSRFGuard 3 Configuration

Jump to: navigation, search


The most important aspect of deploying OWASP CSRFGuard is configuration of the file. There are a minimum number of configuration settings that users should review and specify before running an instance of OWASP CSRFGuard. Such configurations include specifying the new token landing page, enabling Ajax support for applications making use of XMLHttpRequest, capturing pages that should not be protected, as well as configuring one or more actions that should be invoked when a CSRF attack is identified. The purpose of this article is to provide an overview of key OWASP CSRFGuard configuration settings.

New Token Landing Page

The new token landing page property (org.owasp.csrfguard.NewTokenLandingPage) defines where to send a user if the token is being generated for the first time. CSRFGuard will redirect the user to the current page without any parameters if the property is not specified. Preventing the protected application from consuming a request whose session does not yet have a CSRF token through the use of a redirect prevents the execution of a one-time CSRF attack. The following configuration snippet instructs OWASP CSRFGuard to redirect the user to /Owasp.CsrfGuard.Test/index.html when they visit a protected resource without having a corresponding CSRF token present in the HttpSession object.


Ajax and XMLHttpRequest Support

The Ajax and XMLHttpRequest property (org.owasp.csrfguard.Ajax) indicates whether or not OWASP CSRFGuard should check for the presence of the X-Requested-With (case insensitive) header in the HTTP request. If the header is found, then we have strong evidence that the request was sent using the XMLHttpRequest object. As long as the browser properly enforces the Same Origin Policy with regards to the use of XMLHttpRequest, verification of the header is a strong defense against CSRF attacks for Ajax applications. Note that verification of the X-Requested-With header takes precedence over verification of the CSRF token supplied as an HTTP parameter. More specifically, CSRFGuard does not verify the presence of the CSRF token if the Ajax support property is enabled and the X-Requested-With header is embedded within the request. The following code snippet instructs OWASP CSRFGuard to support Ajax requests by verifying the presence of the X-Requested-With header.


Referrer Checking

The check referrer header property (org.owasp.csrfguard.Referrer) verifies that the supplied referrer header matches a regular expression. When enabled, verification of the referrer header will take precedence over verification of the CSRF token in the HTTP request. Referrer headers are consistently sent by browsers for pure HTTPS communications. Verification of the referrer header is a strong defense against CSRF attacks when consistently using HTTPS as a CSRF attack will not include this header in the request(s) unless the malicious HTML is embedded within the site for which the attack is targeted. Inclusion of the referrer header is less consistent when using traditional HTTP traffic across browsers. CSRFGuard will fall back to verifying the presence of the CSRF token if the referrer header was not included in the HTTP request. The following code snippet instructs OWASP CSRFGuard to verify that the referrer header matches the regular expression .*localhost.*, ensuring the request came from a local resource.


Unprotected Pages

The unprotected pages property (org.owasp.csrfguard.unprotected.*) defines a series of pages that should not be protected by CSRFGuard. Such configurations are useful when the CsrfGuardFilter is aggressively mapped (ex: /*). The syntax of the property name is org.owasp.csrfguard.unprotected.[PageName], where PageName is some arbitrary identifier that can be used to reference a resource. The syntax of defining the uri of unprotected pages is the same as the syntax used by the JavaEE container for uri mapping. Specifically, CSRFGuard will identify the first match (if any) between the requested uri and an unprotected page in order of declaration. Match criteria is as follows:

Case 1: exact match between request uri and unprotected page
Case 2: longest path prefix match, beginning / and ending /*
Case 3: extension match, beginning *.
Default: requested resource must be validated by CSRFGuard

The following code snippet illustrates the three use cases over four examples. The first two examples (Tag and JavaScriptServlet) look for direct URI matches. The third example (Html) looks for all resources ending in a .html extension. The last example (Public) looks for all resources prefixed with the URI path /MySite/Public/*.


Response to Attack: Actions

Miscellaneous Configurations