DevelopSec

pen testing

by Leave a Comment

Proxying localhost on FireFox

When you think of application security testing, one of the most common tools is a web proxy. Whether it is Burp Suite from Portswigger, ZAP from OWASP, Fiddler, or Charles Proxy, a proxy is heavily used. From time to time, you may find yourself testing a locally running application. Outside of some test labs or local development, this isn’t really that common. But if you do find yourself testing a site on localhost, you may run into a roadblock in your browser. If you are using a recent version of FireFox, when you go into your preferences screen and click on the Network Settings “Settings” button, you might notice the following image:

DenyProxy

When configuring your proxy, there is a box to list exceptions to not proxy traffic for. In the old days, localhost used to be pre-populated in this box. However, that is not the case anymore. Instead, localhost is explicitly blocked from being proxied. You can see this in the highlighted area of the image above.

So how do you you proxy your localhost application? There are a few ways to handle this.

You could set up your hosts file to give a different name to your local website. In this case, you would access the application using your defined hostname, rather than “localhost”.

Another way to get around this would be to modify the about:config of Firefox and update the network.proxy.allow_hijacking_localhost property true as shown in the following image:

AllowHijacking

Once this change is made, it will update the network settings screen to no longer block localhost from proxying. The following image shows that this statement is no longer there:

AllowProxy

by Leave a Comment

XSS in Script Tag

Cross-site scripting is a pretty common vulnerability, even with many of the new advances in UI frameworks. One of the first things we mention when discussing the vulnerability is to understand the context. Is it HTML, Attribute, JavaScript, etc.? This understanding helps us better understand the types of characters that can be used to expose the vulnerability.

In this post, I want to take a quick look at placing data within a <script> tag. In particular, I want to look at how embedded <script> tags are processed. Let’s use a simple web page as our example.

<html>
	<head>
	</head>
	<body>
	<script>
		var x = "<a href=test.html>test</a>";
	</script>
	</body>
</html>

The above example works as we expect. When you load the page, nothing is displayed. The link tag embedded in the variable is rated as a string, not parsed as a link tag. What happens, though, when we embed a <script> tag? 

<html>
	<head>
	</head>
	<body>
	<script>
		var x = "<script>alert(9)</script>";
	</script>
	</body>
</html>

In the above snippet, actually nothing happens on the screen. Meaning that the alert box does not actually trigger. This often misleads people into thinking the code is not vulnerable to cross-site scripting. if the link tag is not processed, why would the script tag be. In many situations, the understanding is that we need to break out of the (“) delimiter to start writing our own JavaScript commands. For example, if I submitted a payload of (test”;alert(9);t = “). This type of payload would break out of the x variable and add new JavaScript commands. Of course, this doesn’t work if the (“) character is properly encoded to not allow breaking out.

Going back to our previous example, we may have overlooked something very simple. It wasn’t that the script wasn’t executing because it wasn’t being parsed. Instead, it wasn’t executing because our JavaScript was bad. Our issue was that we were attempting to open a <script> within a <script>. What if we modify our value to the following:

<html>
	<head>
	</head>
	<body>
	<script>
		var x = "</script><script>alert(9)</script>";
	</script>
	</body>
</html>

In the above code, we are first closing out the original <script> tag and then we are starting a new one. This removes the embedded nuance and when the page is loaded, the alert box will appear.

This technique works in many places where a user can control the text returned within the <script> element. Of course, the important remediation step is to make sure that data is properly encoded when returned to the browser. By default, Content Security Policy may not be an immediate solution since this situation would indicate that inline scripts are allowed. However, if you are limiting the use of inline scripts to ones with a registered nonce would help prevent this technique. This reference shows setting the nonce (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Security-Policy/script-src).

When testing our applications, it is important to focus on the lack of output encoding and less on the ability to fully exploit a situation. Our secure coding standards should identify the types of encoding that should be applied to outputs. If the encodings are not properly implemented then we are citing a violation of our standards.

by Leave a Comment

What is the difference between Brute Force and Credential Stuffing?

Many people get confused between brute force attacks and credentials stuffing. To help clear this up, here is a simple description of the two. These are both in regards to the login form only.

Brute Force
Brute force attacks on the login form consist of the attacker having a defined list (called a dictionary) of potential passwords. The attacker will then try each of these defined passwords with each username the attacker is trying to brute force. Put simply, this is a 1 (username) too many (password) attack.

A common mitigation to brute force attacks is the implementation of account lockout. In this case, after 3, or 5, or 10 failed attempts for a single username, the user account is locked to block any more attempts. This drastically reduces the number of passwords that may be tried in a short period of time.

Credential Stuffing
Credential Stuffing is another attack on the login form but it differs from a brute force attack in that the list used contains both a username and a password. This list is often obtained through a data breach at another organization. The purpose is to find accounts that are re-used across multiple sites. In this case, the attack is a 1 (username) to 1 (password) attack. For each username, only one password will be attempted.

Unlike with Brute Force, account lockout doesn’t have much affect on credential stuffing. Multi-factor authentication is a good mitigation as it will limit the use of valid credentials.

by Leave a Comment

Thinking about starting a bug bounty? Do this first.

Application security has become an important topic within our organizations. We have come to understand that the data that we deem sensitive and critical to our business is made available through these applications. With breaches happening all the time, it is critical to take reasonable steps to help protect that data by ensuring that our applications are implementing strong controls.

Over the years, testing has been the main avenue for “implementing” security into applications. We have seen a shift to the left more recently, leading to doing more throughout the entire development cycle, but we still have a ways to go. I am still a firm believer in embedding security into each of the phases as our main means of securing applications. Testing, however, is still a major component of any security program.

Typically, organizations rely on penetration testing to find the flaws in their applications. This is the de facto standard for understanding your risk. Unfortunately, penetration testing for applications has been watered down from what we think about with network testing. Many of the assessments we call penetration tests these days are just automated scans transposed into a custom report. These types of testing overlook one of the components a penetration test provides, which is the manual testing. Of course, there is much more to a penetration test, but that is not the focus of this post.

Internally, organizations may implement automated tools to help identify security flaws within their applications. These tools are good at finding certain types of flaws, and usually quite quickly. Like many current penetration tests, they lack the manual assessment side.

Not only does manual testing have the ability to find different types of flaws, such as authentication, authorization, CSRF, business logic, etc., it also has the ability to identify flaws that an automation tool overlooks. For example, a tool may not find every instance of cross-site scripting, depending on how that tool analyzes the system. Granted, manual testing is not guaranteed to find every instance either. With each type of testing, there is always a number of issues that will not be identified. The goal is to start reducing these numbers down over time.

Handling the results of all these res ports from the different assessments is critical to how well you start creating more resilient applications. In many organizations, vulnerabilities identified are handled as individual items and patched. In my opinion, the return on investment is when you can analyze these results to review your development process and see what improvements can be made to reduce the chance these types of flaws will be included in the future. Having an expert available to help review the issues and provide insight into how to use that information to improve your process is valuable.

Having a solid application development process in place is important before thinking about implementing a bug bounty program within your organization. If you are not already doing things consistently, there is a better chance the bounty program will fail.

Bug bounty programs have been becoming more prevalent over the last few years. This is especially true for newer technical startups. We have seen much slower adoption with most of the major corporations. There are many reasons for this, which are outside the scope of this post. There have been questions on whether bug bounties can replace penetration testing. The answer is no, because the goal of each of these is different. There are plenty of articles discussing the subject. A bug bounty program has also been seen by many as the evidence to show they are doing application security. Unfortunately, we can’t test ourselves secure. As I stated previously, testing is just a part of our solution for application security.

A key difference between our traditional testing and a bug bounty program is that bug bounties pay by the bug. Our traditional testing is provided at flat fees. For example, that automated tool is a set price for a month or year subscription. A penetration test is a set price per test. A bug bounty is priced per bug, which makes the cost very unpredictable. In addition, if you are not already doing many of the things previously discussed, there could be a lot of bugs to be found, leading to potentially high payouts.

As I have stated before, penetration testing has a different purpose and it can be very expensive. At Jardine Software we offer more budget friendly manual application security testing at a fixed cost. The goal is not necessarily to find every instance of every vulnerability or to exploit vulnerabilities in the way a penetration test would. The focus is on augmenting the automated testing you may already have in place and to provide that missing manual piece. The testing is performed manually by using the application in combination with Burp Suite, to look for weaknesses and provide those in a way that helps prioritize and then remediate them according to your organization’s needs.

The manual application security testing is typically performed over a week to two weeks and includes a broader scope than a typical bug bounty program. The reason for this is that we want to help identify risks that we see based on our years of experience to make you aware. This assessment can then help identify where you may have issues within your application before opening it up for a crowd sourced bounty program where each bug is priced individually.

If you are thinking about implementing a bug bounty program, reach out and lets chat first. Even if you are not considering a bug bounty program, do you have any manual application security testing implemented? We have the expertise to help provide the necessary testing or provide training for your internal teams to start applying manual testing techniques as part of your life cycle.

by Leave a Comment

Installing BeEF on ubuntu 18.04

While working on a VM for a class, I had the opportunity to install BeEF for the students. This was the first VM I have built using Ubuntu 18.04, so I expected there to be a few hiccups along the way. The good news is that the process was pretty straight forward and simple. Here are the steps to getting this up and running on Ubuntu 18.04.

I started off by creating a new virtual machine using Ubuntu 18.04. I won’t go through the steps of creating a new virtual machine image, but the takeaway here is that I am starting with a fresh Ubuntu system.

The BeEF team has put together some simple instructions for installing the application (https://github.com/beefproject/beef/wiki/Installation). This walk through follows these instructions pretty close, with one exception to clear up an error we will see along the way (or at least an error I saw).

The first thing I am going to do is install Ruby on my Ubuntu 18.04 image. To do this we want to run the following command:

sudo apt install ruby ruby-dev

The next step is to get the BeEF source files. We will get these from git. If you haven’t installed git, make sure to run the following command first:

sudo apt install git

Once git is available, we can clone the git project with the following command:

git clone https://github.com/beefproject/beef

This will download the source files for BeEF. Next, move into the beef directory:

cd beef

This is where the I started to run into an issue. The original beef instructions say to just run:

./install

However, when I did this, I received a permission error. To resolve that I ran:

sudo ./install

There is probably a way to fix this permission error without running as sudo.. but I didn’t investigate that further.

Once the installation was successfully completed, I ran:

./beef

I was quickly greeted with the following error:

demo@ubuntu:~/beef$ ./beef
Traceback (most recent call last):
4: from ./beef:44:in `<main>’
3: from /usr/lib/ruby/2.5.0/rubygems/core_ext/kernel_require.rb:59:in `require’
2: from /usr/lib/ruby/2.5.0/rubygems/core_ext/kernel_require.rb:59:in `require’
1: from /home/demo/beef/core/loader.rb:17:in `‘
/home/demo/beef/core/loader.rb:17:in `require’: cannot load such file — xmlrpc/client (LoadError)

After digging around, I found that ubuntu 18.04 by default installs Ruby 2.5, which apparently doesn’t have the xmlrpc/client embedded. To fix this, we just need to tell BeEF that it needs this gem. To fix this, I modified the Gemfile file following these steps:

rm Gemfile.lock – Do this first to remove the lock file. Click Y to remove it.

sudo nano Gemfile

In the file, add the following line:

gem ‘xmlrpc’

Save the file and re-run the installation:

sudo ./install

At this point, the installation should be successful. Try running the following command:

./beef

You should see something like the following:

demo@ubuntu:~/beef$ ./beef
[ 6:41:32][*] Browser Exploitation Framework (BeEF) 0.4.7.0-alpha
[ 6:41:32]    |   Twit: @beefproject
[ 6:41:32]    |   Site: https://beefproject.com
[ 6:41:32]    |   Blog: http://blog.beefproject.com
[ 6:41:32]    |_  Wiki: https://github.com/beefproject/beef/wiki
[ 6:41:32][*] Project Creator: Wade Alcorn (@WadeAlcorn)
[ 6:41:33][*] BeEF is loading. Wait a few seconds...
[ 6:41:38][*] 8 extensions enabled.
[ 6:41:38][*] 301 modules enabled.
[ 6:41:38][*] 2 network interfaces were detected.
[ 6:41:38][*] running on network interface: 127.0.0.1
[ 6:41:38]    |   Hook URL: http://127.0.0.1:3000/hook.js
[ 6:41:38]    |_  UI URL:   http://127.0.0.1:3000/ui/panel
[ 6:41:38][*] running on network interface: 192.168.116.139
[ 6:41:38]    |   Hook URL: http://192.168.116.139:3000/hook.js
[ 6:41:38]    |_  UI URL:   http://192.168.116.139:3000/ui/panel
[ 6:41:38][!] Warning: Default username and weak password in use!
[ 6:41:38]    |_  New password for this instance: ec04906c30d928fb857
[ 6:41:38][*] RESTful API key: 6bd0b11e772df40
[ 6:41:38][*] HTTP Proxy: http://127.0.0.1:6789
[ 6:41:38][*] BeEF server started (press control+c to stop)

Notice that there is a “New password” configured here. This is because by default beef sets the username/password to beef/beef. As this is a default, hard-coded password, it is insecure. To fix this, beef detects the default and creates a new temp password to protect the instance. It is recommended to update the username and password to your instance.

Updating the Password

To change the password stop beef by typing ctrl+c. Now, we will edit the config.yaml file:

sudo nano config.yaml

You should see something like this:


beef:
    version: '0.4.7.0-alpha'
    # More verbose messages (server-side)
    debug: false
    # More verbose messages (client-side)
    client_debug: false
    # Used for generating secure tokens
    crypto_default_value_length: 80

    # Credentials to authenticate in BeEF.
    # Used by both the RESTful API and the Admin interface
    credentials:
        user:   "beef"
        passwd: "beef"

    # Interface / IP restrictions

Modify the user and passwd fields to your own values and then save the file using ctrl+x.

When you restart BeEF and go to the ui panel you should now be able to login with your new credentials.

Notes
There should be a way to install the application without using sudo ./install. This should be checked so you don’t install using root permissions. This is a non-production image for me to use for student training.

Make sure you change the default username and password to help lock down your instance.

This tutorial is for educational purposes only. Hacking without permission is illegal and should not be done.

by Leave a Comment

Choosing Application Security Tools

There are lots of security tools available, so how do you know which one to pick?

If your security team is not including the application teams in the decision, you run a big risk of failure. The security team does get the ability to form relationships with vendors. We see them at conferences. We know people that work there. Because our focus is on security, we know the tools that exist in our space and we have an idea of which ones may be better than others. Of course, this is often due to what we hear from others, rather than our own experience with those tools.

Picking a tool in a vacuum is less than ideal. Sure, the tool may have five stars and has great detection rates. However, if the tool doesn’t fit into the development process and causes more overhead and friction, it will end up failing. A tool with 100% accuracy is still useless if it is not actually being used.

I have seen over and over where a security group acquires a tool, especially static analysis, and a few months later they realize it is just sitting on the shelf. Like all of us, we are excited when we get a new toy. We use it, learn it. Then, the newness wears off. Without the proper processes in place, this can happen with security tools as well. We don’t want this to happen to you.

The first step in determining which tools make sense is to understand the development process now and where it is going. Lets talk about static analysis for a moment. If your development process doesn’t make use of continuous integration and doesn’t plan to then that is not a high priority feature of the tool you receive. Have you considered what IDE the developers use?

I once had a situation where static analysis was about to be rolled out. The security team worked to pick a vendor and get the ball rolling. It was later in the process that the development team was brought on board with the conversation. They asked questions that were less of a priority to the security team, but more of a priority to them. For example, The tool provided a plugin for their IDE. However, when digging deeper, the plugin ran a few versions behind. So what had appeared to be a good setup, now looked a little less efficient since the plugin may not work. This isn’t a deal breaker, but depending on how you were planning on this solution working, it adds friction to the process. The more friction, the more pain.

The next step is understanding who will ultimately be using this tool? For those that listen to the podcast on a regular basis and follow my blogs, you know I am a huge fan of the application teams having direct access to these tools. In my opinion, static analysis is a developers tool. It is there to evaluate the source code and identify policy violations that only the developer can resolve. Giving them access and control over that function and embedding it into their process reduces that friction. Don’t confuse this with the idea that security is not getting the results. Security still needs to have insight into what is going on with security flaws within the application. But to be efficient, the results of these tools need to fit into the flow of how development works. Not be one-off reports under different processes.

This is no different, in my opinion, than dynamic analysis or interactive analysis. These are tools that should be used by the QA group. The group that is responsible for testing and most likely already has automated testing capabilities. They are trained in identifying bugs, reporting, and tracking them. Building these types of tools into their process just makes sense. How many listeners have their own stories of the security team exporting a thousand page report out of that dynamic scanner and sending it to the application team as a pdf? I have been there. Even I won’t read through that report.

The moral of the story here is that if we don’t understand how development works, what type of tools they already use, and what they can handle, we will probably pick a solution that will be ineffective, or at the very least, cause us a lot more work. The goal is not to just shove tools onto the development teams and say do this. As application security representatives, our goal is to help build better applications. That doesn’t mean that our job is to run all the tools and hound the application teams to fix items. By understanding the environments and processes we can pick tools that will fit much better, allowing us to focus our time on building out other solutions or processes for the organization. We are still relied upon to provide the expert advice and guidance to the issues that are identified.

If you are considering implementing these types of tools, take a moment to sit down with all involved parties to get everyone on board. A well laid out plan will go much further than a shotgun approach.