Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
branch: master
Fetching contributors…

Cannot retrieve contributors at this time

193 lines (188 sloc) 9.953 kb

Securing the Rails

Securing the Rails

What we’re going to cover

  • Cross Site Scripting (XSS)
  • Cross Site Request Forgery (CSRF)
  • SQL injection
  • Session management
  • Mass assignment
  • Exception handling/notification
  • Static analysis tools
  • Where can you learn more?

What we’re not going to cover

  • Physical security of your servers
  • Operating System security
  • Configuration management
  • Compliance concerns

This sounds like a lot, but you get a lot for free!

Cross Site Scripting (XSS) Basics

  • A User injects executable code into your application
curl -d "user[name]=<script>alert('xss');</script>" \
  • Your application saves it
  • When your application renders that information the code is executed

Possible outcomes of a successful XSS attack

  • Cookie theft
  • Session hijacking
  • Redirection of user data to an untrusted source
  • Defacement of your application
  • Denial of service

What should you do?

  • Kick back and enjoy a frosty beverage!
  • Rails 3+ automatically escapes output for you
  • Unless you tell it not to
  • But be careful of 3rd party libraries!
  • Always audit your inputs for potential holes
  • If you are on Rails 2, install the rails_xss gem
  • You can accomplish the same thing by adding h() calls to all of your templates
    • Pro Tip: Don’t do this, you will fail.

Cross Site Request Forgery (CSRF)

  • Let’s say you are a bank, and a customer is logged in
  • When that customer visits an embedded tag makes a request to your application on behalf of the user
<script type="text/javascript">
          type: 'POST',
          url: '',
          crossDomain: true,
          data: '{"from":"youraccountnumber",

Possible outcomes of a successful CSRF attack

  • Any action an authenticated user can take is able to be performed on behalf of that user by a malicious agent
  • Example: Working with Rails recommendations circa 2008

What should you do?

  • Again, relax, knowing that Rails has your back on this one
  • Make sure the option is set in the the application configuration that enables this
  • Check again that this option is set
protect_from_forgery :secret => "123456789012345678901234567890..."
  • Flag any changes to the file that contains this option and audit the changes to ensure it doesn’t get turned off

SQL Injection

  • These attacks try to manipulate the your applications data or expose data that a user wouldn’t otherwise have access to.
  • It is normally a pretty easy attack to pull off, but can get complicated in certain situations
SELECT * FROM users WHERE login = '' OR '1'='1'
  AND password = '' OR '2'>'1' LIMIT 1
  • This could be introduced by doing the following inside of a rails application
User.first("login = '#{params[:name]}'
           AND password = '#{params[:password]}'")

Possible outcomes of a successful SQL injection attack

  • Unauthorized access to data
  • Tampering with data
  • Defacement of your system

What should you do?

  • This type of attack is pretty much covered by Rails
  • If you use standard Active Record calls, everything is properly escaped for you
  • Normally, params are also escaped by default
  • You have to work pretty hard to introduce a SQL injection hole in your application

Session Hijacking/Replay

  • Once a user is authenticated against your system, the session is their stamp of entry. If another person gets a copy of the session details (the cookie), they can present this to the sytem to gain the same access that the real user has to the system
  • Unfortunately, this is really easy to do
  • If your cookies travel over plain text (HTTP vs HTTPS), then all it takes is for a user to be on a public network while they log into your system
  • This is also possible via an XSS attack

Possible outcomes of a successful Session hijack

  • The attacker will have access to anything a the hijacked users session permits
  • If this is an admin account, the impact can be huge

What should you do?

  • Protect your login and cookie transfer by forcing it to travel over SSL
    • This will require you to purchase an SSL certificate and set it up on your server(s)
MyApp::Application.config.session_store :cookie_store,
  :key => '_my_app_session',
  :secure => Rails.env == 'production',
  :httponly => true,
  :expire_after => 60.minutes
  • See XSS section

A note about cookie based session storage

  • As a general rule of thumb, you should only store data that is absolutely critical to maintain the state of your application
  • In other words, don’t put anything but a user id in your session data
  • Rails cookie store data might look encrypted, but it is only base64 encoded, making it very easy to decode the information once it is stolen

Mass Assignment

  • This is what happened to Github recently
  • Given a model with unrestricted attribute accessibility, it is possible for a user to modify attributes that should not be accessible.
  • Example
  • While this seems harmless, think of the follwing post data
curl -d "user[admin]=true" http://localhost:3000/user

Possible outcomes of a successful mass assignment attack

  • An attacker can disrupt data integrity
  • An attacker can escalate permissions and control the system
  • If you user model is tied directly to a directory platform (think Active Directory), they can gain access to your network and servers as well

What should I do?

  • Fortunately this is quite easy to protect against
class User
  attr_accessible :name, :email, :phone

Exception Handling and Notification

  • When mistakes surface as exceptions, it is important to record them
  • If a user is able to trigger an exception by acting on your system, it says a lot
  • If an attacker sees this behavior, they will focus harder on that area of your system looking for a way in
  • When you record exceptions, it is usually done via a system that will notify you that the exception occurred
    • Examples: Airbrake, Exception Notifier
  • While these tools offer advantages, they also present additional potential security holes
  • If exception data is sent in plain text, it can be stolen and used against you
  • If you are using a 3rd party service, your exception data is only as safe as their systems

Possible outcomes of improper exception handling/notification

  • Attackers gain information about your system and use it against you as they form more focused attacks
  • Users perform denial of service (DoS) attacks against your system by triggering floods of exceptions, which are expensive to process

What should you do?

  • Test! I’m not talking about unit testing here, I mean get people to actually click around any try to produce exceptions
  • Monitor your logs for exceptional situations and fix them immediately, no matter how insignificant they seem
  • If you are using a third party system, ensure that your data is travelling over SSL
  • If you are using a third party system, determine the risk of access to your data via a breach in their system
    • Essentially, is it ok if someone else owns that data?

Static analysis tools

  • These tools can look at your code and check for potential vulnerabilities
  • file:brakeman-all.png
  • file:brakeman-mass.png
  • file:brakeman-xss.png
  • file:brakeman-misc.png

Where can I learn more?

  • OWASP WebGoat project
  • Web Application Hackers Handbook
  • Practical Software Security
  • Practice Practice Pratice!


Jump to Line
Something went wrong with that request. Please try again.