Understanding XSS Auditor



Understanding XSS Auditor

We see a lot of confusion regarding the X-XSS-Protection header and thought it might be worthwhile to go over exactly what this header is and what it isn’t.

What is XSS Auditor?

XSS Auditor is a built-in function of Chrome and Safari designed to mitigate Cross-site Scripting (XSS) attacks. It aims to identify if query parameters contain malicious JavaScript and block the response if it believes the payloads were injected into the server response. XSS Auditor is enabled by default, but can be configured or disabled with the X-XSS-Protection HTTP header. X-XSS-Protection is a non-standard header, meaning there is no official W3C or IETF specification. Despite this, the common configurations can be seen below.

Valid Configurations

  1. Disable XSS auditor
    X-XSS-Protection: 0
  2. Run in rewrite mode (default if header is not set).
    X-XSS-Protection: 1
  3. Run in “block” mode. Once the auditor is triggered the response is blocked and a blank page is shown to the user.
    X-XSS-Protection: 1; mode=block
  4. Run with reporting. This is a Chromium function utilizing CSP violation reports to send details to a URI of your choice.
    X-XSS-Protection: 1; report=http://example.com/your_report_URI

So what isn’t X-XSS-Protection?

XSS Auditor isn’t a solution to XSS attacks. As Justin Schuh of Google mentions, “XSS auditor is a defense-in-depth mechanism to protect our users against some common XSS vulnerabilities in web sites. We know for a fact it can’t catch all possible XSS variants, and those it does catch still need to be fixed on the affected site. So, the auditor is really an additional safety-net for our users, but not intended as a strong security mechanism.” Because of this, XSS Auditor bypasses are rated as ‘SecSeverity-None’ and, if you we’re wondering, are not eligible for bug bounty payments.

How does the XSS Auditor Work?

XSS Auditor takes a black list approach to identify dangerous characters and tags supplied in request parameters. It also attempts to match query parameters with content to identify injection points. If the query parameter can’t be matched to content in the response, the auditor will not be triggered. Because the browser will never have insight to server-side code, an application that mangles an XSS payload will always render the XSS auditor useless in preventing attacks.
To take a quick look at the code behind Chrome’s XSS auditor, we can get an idea of the inner workings of the detection mechanisms:


Just by looking through the function names we can see that the auditor searches for script tags, valid HTML attributes, and other XSS injection vectors. Before rendering the response in the Document Object Model presented to the user, XSS auditor searches for instances of (malicious) parameters sent in the original request. If a detection is positive, the auditor is triggered and the response is “rewritten” to a non-executable state in the browser DOM. Chrome’s ‘view-source’ has a builtin component to highlight sections on code in red that caused the XSS auditor to fire.


Bypassing XSS Auditor

A bypass of XSS auditor should not be considered a vulnerability. While the Chromium team does actively improve the auditor, there are likely to always be number of bypasses for the auditor. We will not go in depth about specific bypasses as they do change with time and are likely to be outdated fast. At the time of this writing two examples that are functional in the latest version of chrome can be found here and here.

How to Enable Network Level Access for Windows RDP


How to:

Enable Network Level Access for Windows RDP

Chances are you may have arrived here after a vulnerability scan returns a finding called “Terminal Services Doesn’t Use Network Level Authentication (NLA)”. The default configuration of Windows 7, 2008, and 2012 allows remote users to connect over the network and initiate a full RDP session without providing any credentials. This allows an untrusted user to land on the system login page as shown below:

Windows 2008 Login Screen

Several risks are associated with this functionality; an attacker is now able to:

  • Accurately fingerprint the version of Windows
  • Potentially identify user accounts on the system
  • Leverage the RDP service to consume excessive system resources

The default configuration of RDP is similar to letting anyone into the lobby of your building; while they may not have keys to apartments, we generally don’t want strangers milling around the lobby to gather information if it can be avoided.


To enable network level access on Windows 2008 R2 we can do the following:

  1. Open the Group Policy Editor by typing ‘gpedit’

    Group Policy Editor

  2. Navigate to the following:
    • Computer Configuration
    • – Administrative Templates
    • — Windows Components
    • — Remote Desktop Services
    • —- Remote Desktop Session Host
    • —– Security
  3. Doubleclick on “Require user authentication for remote connections by using Network Level Authentication”
  4. Check ‘Enabled’. Apply. Save.

NLA Enabled

Changes are immediate, no reboot is required. Network Level Access should now be enabled.


One of the quickest and easiest ways to verify if NLA is to use the ‘rdesktop’ tool packaged with Kali Linux. When NLA is properly enabled, you will get the following error:

root@kali:~# rdesktop 
Autoselected keyboard map en-us
ERROR: CredSSP: Initialize failed, do you have correct kerberos tgt initialized ?
Failed to connect, CredSSP required by server.

For long term solutions to this issue, organizations may wish to make this change part of a hardened standard image used to provision new servers.

Proactive Measures in Healthcare Application Security


Health IT:

The Growing Need for Proactive Healthcare Application Security

Web application security is a far more complex issue than many other areas of security. There is a high potential for applications to allow users to extract sensitive data, bypass authorization, and even execute commands on the system. But even applications that prevent these types of attacks are often found vulnerable to other “less direct” attacks. As we see time and time again, criminal hackers will go to extreme lengths of effort to profit in any way possible.

As many know, healthcare is an industry which has (understandably) lagged in security. It has had an extreme need for technology development, and has never been an attractive target like banks were to criminals. Unfortunately, the rising black market value of medical records is changing this, and with it bringing more sophisticated attacks. And to see what is coming, we don’t need to look far. We can take a lesson from the financial industry, which has been the target of cutting edge attacks for the last two decades.

As application security improves and attackers find themselves unable to gain direct access to systems, they will look to the next best thing: getting users to do the dirty work. These types of issues are heavily exploited in the financial world, an industry which has poured millions into ensuring applications are not susceptible to such attacks. These are most commonly known as a “confused deputy problem”, where a malicious actor persuades a victim into performing an action against their knowledge or will. Before we look at these attacks individually, it’s important to understand that all of these issues require several prerequisites in order to pose actual risk:

  1. An application performing a sensitive operation.
  2. An attacker with some prior knowledge of the application.
  3. A user currently logged into the application.

Cross-site Request Forgery (CSRF)

Chances are at some point you’ve seen a URL that looks like this:


It performs a specific sensitive function, and takes predictable parameters. While this probably doesn’t seem like a big problem to most people, an attacker may exploit this to force a user to delete records within an application.

CSRF Illustrated

Let’s take a detailed look at the steps:

  1. The attacker sends an email or message to the victim, convincing them to click a link to a website controlled by the attacker.
  2. The user clicks the link, visiting the malicious website.
  3. The attacker takes the URL that will perform a sensitive operation and embeds that link on the malicious site.
  4. When the victim’s browser loads the content of the malicious site, it attempts to fetch the content embedded on the page. This causes the victim’s browser to make a request for this URL – because the user is already authenticated to the application, the operation is performed successfully without any knowledge of the user.

You may be thinking “How is this a vulnerability? This is the way the web works” and you’re exactly right. For a long time this was just how applications were designed; and that was that. The ultimate weakness here is in the HTTP protocol itself, it was simply not designed for the fact that we would be using it for the most sensitive and life dependent operations that we do today. But in order to face reality, we must build applications that compensate for the weaknesses of all components, even the HTTP protocol itself.

ClickJacking (UI Redress Attack)

Similarly to CSRF, clickjacking also persuades a user into performing an action against their will. In this scenario, the attacker creates a web page that embeds the target application within a frame. The attacker then creates a CSS overlay to mask the encapsulated application. The mask may include a JavaScript game which encourages the victim to click on areas of the screen, or enter data into form fields. What the victim doesn’t know is they are actually clicking and typing into an application they are already logged into.

Clickjacking Illustrated

It’s important to note that the risk of CSRF and Clickjacking can vary wildly, and there are several factors which may increase or decrease the risk:

  • Sensitivity of operations – An attacker must be able to benefit from a user performing certain actions. Can the application transfer money? Send a medical record?
  • Ability to target users – An attacker must be able to persuade users to visiting a website. A broad user base of an application will increase risk.
  • Inside knowledge – An attacker must have some inside knowledge of the application. Open source and widely distributed products have a significantly elevated risk because of this.

The expanding health IT ecosystem drives all three of these factors higher in risk. As patient facing applications become more powerful, the more likely users will face these types of attacks. The good news is that both of these issues have well documented fixes which have been adopted into most development frameworks and web servers. Next week we will publish part 2 detailing mitigation strategies.

Defeating Android Emulator Detection



Defeating Android Emulator Detection

At some point while performing vulnerability assessments on android applications you will encounter apps that don’t want to be run within an emulator. We can’t blame application owners for wanting to ensure that the user interaction they see comes from genuine devices, but it doesn’t help us do any security testing on it.

There are several ways to detect an emulator; however this example is only relevant to the most common way we see. In this application, a check is performed for an IMEI value of ‘000000000000000’ which is the value used by the emulator that ships with the Android SDK.

The code segment below checks for this value and exits if true. While we could easily patch the value from within the application, it may be more efficient in the long run to simply change the IMEI value of our emulator. This way we don’t have to patch the next application that does this.

android emulator check

The IMEI is stored as a text string, so we will search for a ‘text string’ accordingly. Open the binary with hexeditor, hit ^W, and search for the fifteen zeroes. Note that the binary we wish to open is not the “emulator” binary, but the “emulator-arm” binary. If you are using a different architecture you may be using the mips or x86 binary.

cp emulator-arm emulator-arm.bak
hexeditor emulator-arm

hexedit search

Note once again this is an ascii string, so the zeroes are 0x30.


In this case, we just replace four characters with 1234 by updating 0x31, 0x32, 0x33, and 0x34. Do not change the length of data in this segment or overwrite bytes outside this segment or you will corrupt the binary.


Just save and exit. Now our emulator will be using our new custom value.

Preventing Cross-site Scripting in PHP



Preventing Cross-site Scripting in PHP

Preventing Cross-site Scripting (XSS) vulnerabilities in all languages requires two main considerations: the type of sanitization performed on input, and the location in which that input is inserted. It is important to remember that no matter how well input is filtered; there is no single sanitization method that can prevent all Cross-site Scripting (XSS). The filtering required is highly dependent on the context in which the data is inserted. Preventing XSS with data inserted between HTML elements is very straightforward. On the other hand, preventing XSS with data inserted directly into JavaScript code is considerably more difficult and sometimes impossible.

Input Sanitization

For the majority of PHP applications, htmlspecialchars() will be your best friend. htmlspecialchars() supplied with no arguments will convert special characters to HTML entities, below shows the conversions performed:

'&' (ampersand) becomes '&'
'"' (double quote) becomes '"'
'<' (less than) becomes '&lt;'
'>' (greater than) becomes '&gt;'

Eagle eyed readers may notice this does not include single quotes. For this reason we recommend that htmlspecialchars() is always used with the ‘ENT_QUOTES’ to ensure single quotes will be encoded. Below shows the singe quote entity conversion:

"'" (single quote) becomes '&#039;' (or &apos;)  

htmlspecialchars() vs htmlentities()

Another function exists which is almost identical to htmlspecialchars(). htmlenities() performs the same functional sanitization on dangerous characters, however, encodes all character entities when one is available. This may lead to excessive encoding and cause some content to display incorrectly if character sets change.


strip_tags() should NOT be used exclusively for sanitizing data. strip_tags() removes content between HTML tags and cannot prevent XSS instances that exist within HTML entity attributes. strip_tags() also does not filter or encode non-paired closing angle brackets. An attacker may be able to combine this with other weaknesses to inject fully functional JavaScript on the page. We recommended that strip_tags() only be used for its intended functional purpose: to remove HTML tags or content. In these situations, input should be passed through htmlspecialchars() after strip_tags() is used.


addslashes() is often used to escape input when inserted into JavaScript variables. An example is shown below:

 var = "te\"st ";   // addslashes()

As we can see, addslashes() adds a slash in attempt to prevent an attacker from terminating the variable assignment and appending executable code. This works, sort of, but has a critical flaw. Most JavaScript engines will construct code segments from open and closed <script> tags before it parses the code within them. This is done before the browser even cares about the data that resides between the two quotes. So to exploit this, we don’t actually need to “bypass” addslashes(), but simply terminate the script tag.

 var = "test1</script><script>alert(document.cookie);</script>";

As far as the browser is concerned, the code injected is an entire new code segment and contains valid JavaScript.

Where Entity Encoding Fails

We talked before about considerations for the location of data, and will go over some examples where entity encoding with htmlspecialchars() is not enough. One of the most common examples of this is when data is inserted within the actual tag or attribute of an element.

HTML Event Attributes: HTML has a number of elements with attributes that allow for JavaScript to be called after a particular event. For example, the onload attribute can execute JavaScript when an HTML object is loaded.

<body onload=alert(document.cookie);>

This is just one of many somewhat rare situations where extremely strict filtering is required. For an in depth look at many injection scenarios and their prevention methods, take a look at the OWASP XSS Prevention Cheat Sheet.

Third Party PHP Libraries

Virtue Security makes no recommendation or provides any warranty for third party products or software; however, we are aware that several third party PHP libraries are commonly used to assist in XSS prevention. Below are projects that may assist developers building suitable whitelists:

HTML Purifier – http://htmlpurifier.org/
PHP Anti-XSS – https://code.google.com/p/php-antixss/
htmLawed – http://www.bioinformatics.org/phplabware/internal_utilities/htmLawed/

Other Things to Remember

A great rule of thumb to go by is simply not to insert user controlled data unless its explicitly needed for the application to function. It’s often surprising to see XSS vulnerabilities exist because parameters are inserted into HTML or JavaScript comments. Not only does this serve no functional purpose to the application, but it can introduce serious security vulnerabilities.

 Page 1 of 5  1  2  3  4  5 »