I am preparing myself for the Hacktivity conference in Budapest, where I am gonna talk about the security of the Cross-Origin Resource Sharing (CORS). As part of the preparation I will summarise my thoughts in a couple of blog posts.
To start off with I will describe the potential attackers who could try to use CORS in their attacks and I will build an attacker model.
First let’s look at the architecture where CORS is relevant.
It can be seen on the picture that the attacker has control of at least one server. Of course this server could be in the internal network, however, this way the model is more general. The target can be either in the intranet or in the Internet, which brings us to the first differentiation point: the attacker’s knowledge about the internal network.
Here the Internal attacker means that he has knowledge about the internal network and services, but it doesn’t mean necessarily that he is in the internal network. A good example is an ex-employee, who knows how to interact with the internal service and has great chances to do social engineering, however, he has no access to the internal network anymore.
The attacker has no knowledge about the internal network. In this case he could either attack services on the Internet, to which he has access, and he is able to create attacks. He can also create attacks to get to know the internal network to find well known software (i.e., open source project used by the company) which he can analyse off-line.
Although the attacker could be local, but he would have better options then using CORS, so I would generally consider a remote attacker. As shown on the architecture the attacker has control at least over one server on the Internet. This server can be his own, then he needs to trick the user to visit it, or it can be a compromised server, which he could use to inject his own code for instance through an XSS. There are enough vulnerable servers on the Internet so this is a good option as well.
The goal of the attacker is either to steal information from the target servers, to which he doesn’t have access, or manipulate these applications in a way that can help him in further attacks. When attacking a service on the Internet his goal might be to use the target user’s authenticated session to steal data. In case of the internal target the most important goal is to get access to the target services at all.
To finish the analysis, using the above described attributes a potential attack could be for example the following:
- Well informed about the target service.
- Remote attacker.
- Goal: access protected content or services.