Testing of Web Security

Question

In your experience, what have you found, worked on, or encountered in terms of site vulnerabilities? And what actions did you take to mitigate these issues?

This may include XSS (cross site scripting), SQL Injection attacks, plain old DDOS or phishing attempts on your site's customers. Only yesterday I came across an entire section of Firefox tools for auditing sites and their potential for various vulnerabilities.

Looking to expand my knowledge in this area for a role, so more information to read or learn is always good - solid links appreciated too! And war stories of the worst you've found or scariest hole you've seen - learning from experience is the best way sometimes!

Answer 1

I've done security review, white-box and black-box, for dozens (hundreds?) of applications and sites.

1. XSS and SQL injection get a lot of press, but know what I find the most common security flaw to be? Leaving debug and test functionality in production code. Either by tampering with POST parameters (isDebug=True) or via spidering a site and finding leftover pages, these are the worst mistakes I see regarding security. If you're including test/debug code, put it in a separate code branch, or at least prepare a checklist for removal prior to launch.

2. The next most common vulnerability I've seen is simply the ability to bypass security mechanisms by grabbing a URL from the page source. The technical name is 'Forceful Navigation' or 'Forced Browsing' This is something anyone who can read HTML can do, yet I'm surprised by the variety of applications vulnerable. Reviewing a ticket-purchasing site yesterday, I was able to buy tickets for sold-out shows using this method. On previous sites, I was able to skip paying altogether (many, many Paypal sites pass the "purchase complete" URL to paypal via POST parameters - yoink!). You need some sort of back-end statefulness or check to ensure completion, payment, availability, accuracy, etc.

3. To be frank, I usually let tools like AppScan, BURP proxy, WebScarab, Fortify, FindBugs, or YASCA (depending on budget and source code accessibility) find XSS and SQL injection attacks for me. I'll try the simple stuff on my own, look for obvious holes, but there's too many known combinations to try yourself. I keep a small collection of scripts and test cases for more advanced or recently discovered flaws.

I'm going to stop at 3, because I really could go on all day, I'm losing focus from your question, and nobody wants to read a wall of text.

Some resources for new and seasoned web security gurus: (ARGH. I can't officially post links yet. Copy/paste. Sorry)

The Open Web Application Security Project (OWASP)

http://www.owasp.org/

Web Security Testing Cookbook

This book is written for auditors, testers, and less for developers. Which is pretty unusual for an O'Reilly book.

websecuritytesting.com

Vulnerability Categorization by Fortify

www.fortify.com/vulncat/

Common Weakness Enumeration (warning: extensive)

nvd.nist.gov/cwe.cfm

Common Attack Pattern Enumeration and Classification (warning: even more extensive)

capec.mitre.org/

Google's Web Security Tutorials

(rather weak)

code.google.com/edu/security/index.html

Answer 2

I joined a web app project that included a document library. The way it referenced documents was something like http://example.com/getdocument?file=somefile.pdf. Of course I just had to try file=/etc/passwd, and of course it worked.

Solution: Perform user input sanitization and/or use some level of abstraction between resources requested in the URL and actual filesystem resources.

This is the cousin of SQL injection attacks. Examine any allowed request that suspiciously looks like it gives the client too much control.