If attackers can deploy a remote administration tool (RAT) on your network, it makes it so much easier for them. RATs make it luxurious for bad guys; it’s like being right there on your network. RATs can log keystrokes, capture screens, provide RDP-like remote control, steal password hashes, scan networks, scan for files and upload them back to home. So if you can deny attackers the use of RATs, you’ve just made life a lot harder for them.

We are getting better at catching so-called advanced persistent threats by detecting the malware they deploy on compromised systems. We can say this because experts are seeing more attackers “living off the land.” Living off the land means an attacker goes malware-free and instead relies on the utilities, scripting engines, command shells and other native resources available on systems where they gain an entry point.

By living off the land, they keep a much lower profile. They aren’t stopped as much by application control and whitelisting controls. There’s no malware for antivirus to detect.

And Windows provides plenty of native resources for this kind of attacker. (Linux and UNIX do too, but I’m focusing on Windows since client endpoints initially targeted by today’s attackers mostly run Windows.) You might be surprised how much you can do with just simple batch files, let alone PowerShell. And then there’s WMI. Both PowerShell and WMI provide a crazy amount of functionality. You can access remote systems and basically interface with any API of the operating system. You can open up network connections for “phoning home” to command and control servers, and more. This is all stuff that in years past required an EXE or DLL. Now you can basically do anything that a custom built EXE can do but without touching the file system which so much of our current security technology is based on.

How do you prevent attacks like this? PowerShell has optional security restrictions you can implement for preventing API access and limiting script execution to signed script files. With WMI it’s not as clear. Obviously, all the normal endpoint security technologies have a part to play.

But let’s focus on detection. It’s impossible to prevent everything and mitigate every vulnerability. So we can’t neglect detection. The challenge with detecting attackers that are living off the land is twofold. The activities you need to monitor:

1. Aren’t found in logs
2. Are happening on client endpoints

Both of these create big challenges. Let’s talk about #1 first. A.N. Ananth and I describe the types of activities that are clues to possible attacker living off the land in 5 Indicators of Evil on Windows Hosts using Endpoint Threat Detection and Response and I encourage you to watch that session which is full of good technical tips. But the point is that what you need to watch for isn’t in the Windows security log or other logs. Instead, detection requires a combination of file scanning, configuration checks, querying of running processes and so on — all stuff that requires code running on the local system or very powerful and complex remote access. If we were only talking about servers, we could consider deploying an agent. But to catch today’s threats, you need to be monitoring where they begin, which is on client endpoints — the desktops and laptops of your employees. And there’s no way to remotely reach into that many systems in real time, even if you overcame the technical hurdles of that kind of remote access. So that leaves agents, which always cause a degree of pushback.

But it’s time to stop calling them agents. Today what we need on endpoints are sensors. It’s a subtle but important shift in mindset. In the physical world, everyone understand the need for sensors, and that sensors have to be deployed where the condition is being monitored. If you want to know when someone enters your building at night, you need a sensor on every door. Likewise, if you want the earliest possible warning that your organizations have been compromised, you need a sensor on every endpoint.

So I encourage you to start thinking and speaking in terms of leveraging your endpoints as a sensor rather than yet another system that requires an agent. And look for security vendors that get this. EventTracker has done a great job of evolving their agent into a powerful and irreplaceable endpoint security agent that “sees” things that are just impossible to see any other way.