I know I have been fairly inactive on here lately, so I figured I would start the new year off on the right foot by posting an article even if it is rather non technical.
Clients looking to have an assessment of their IT infrastructure and personnel via a penetration test always have to spell out what is in scope so nobody gets into legal issues or crashes production systems. Unfortunately when you are the good guy you have pesky things to worry about that Sergey in his Adidas tracksuit never has to think about. “What if my banking RAT spreads onto the HR systems and accidentally steals PII?” – said no attacker ever. In the long run clients usually want a test to be as real world as possible, but they certainly don’t want to have downtime or lose money. It is a reasonable accommodation because as Wu-Tang elegantly puts it Cash Rules Everything Around Me (CREAM).
Things can get a bit more ambiguous with scope if you are using enterprise services from third parties (Google EC2, Dropbox Enterprise, etc) but most of them do have fine print and proper forms which can be filled out which would allow the penetration test to include your assets hosted on their servers. Let’s face it, the day and age of a static IP range of company owned hardware is behind us for the most part. On top of this modern organizations have a complex and sometimes undocumented attack surface which includes servers and services co-located, leased, virtualized, and on-demand which can encompass multiple countries with their varying cyber laws.
What a client tells you is their IT assets rarely lines up with all the potential ways they can be digitally compromised.
On a recent assessment I came across some vulnerabilities unknowingly (to the client) created by the company they contracted to create the website and mobile app. This got me thinking about how rarely anyone includes third party developers in the scope of a penetration test even though they often times still maintain and keep clones or testing versions of your target.
You come in as a pen tester to look at the finished or nearly complete website/app that is provided to you by the client. That’s great and all, but don’t forget that at some point there was likely dozens of versions created and tested back at the web/app developers network. Unless you make an effort to uncover where those assets are you might be missing out on an expanded attack surface in your assessment.
What I am not saying is that without clearly defined guidelines can you can compromise an outside developers network in order to gain access to your client. What I am saying is that at a minimum you should be looking at them for data which can be leveraged in a pen test. If your client paid to have a website created, and a clone of it is sitting on a separate subdomain you should attempt to have that included in the scope. If that is not doable, then there is plenty that you can do as a sort of pentest light which keeps you legally cleared. Below are the very well known tools that can be done passively (ish) and still give you good results. Notice I am not mentioning using burp for sql injection or anything agressive.
Passive Recon of Developer Networks:
Robtex.com (the old site is still more functional) – look to see what lives in the IP range of your client AND the company that made the website/app. Look at the reverse lookups to determine if the company had something live dev.yourclient.webcompany.com. This can also be a place to find developer companies with duplicates of your clients SSL certificates…..
shodan.io – Not even sure I have to say this. What can not be overstated is that if there is a glaring vulnerability documented by shodan that you are certainly not the first to find this. In addition to the dangers of leaking data, database structures of the testing db (can help with sql injection on the production network), don’t underestimate the dangers of giving access to a test database. How is the testing/dev backend being brought online for production? As is the case currently, thousands of MongoDB’s have anonymous read/write access. What’s to stop me from putting in an account on the testing database if I know it is going to be replicated over once it goes into production?
Case in point, here is a common sight if you do db searches.
Google.com – Don’t laugh. Use special searches to look for juicy info that might be unlinked, or intended to be internal. I have found dozens of sensitive documents, PII spills, and change request docs using this method. Example: inurl:dev.company.gov Internal Use Only Google has a knack for indexing documents which companies forget about because they are not linked to websites anymore. Even if it is gone now you might be able to get the text of the document cached if you are lucky. On a recent assessment I found the developers decided to put all their internal Change docs off of an indexed directory for some odd reason. In one of those documents they duly noted that they Backdoored the website with the account “Developer:1234”. To think i was wasting time reversing an Android apk file to look for credentials….
Maybe more to come… Have fun.