In 2011, I was invited to Austin, TX by the local ISSA and OWASP chapters to teach a class on Armitage and the Metasploit Framework. I think we had 90 students. I remember the pain of burning DVDs in preparation for this class. Myself and two of the organizers agreed to split the DVD burning load equally. Fun times.
This workshop also had the first version of my Penetration Testing Lab DVD. It came with a Xubuntu VM for an attack platform, Metasploitable for a remote exploit lab, and a Mint Linux VM for a client-side attack lab.
This half-day workshop was my first time giving a course on hacking. If you want to see it, I also made a screen recording from the slides:
October 2011, I came to LAS CON and taught a one-day Advanced Threat Tactics course (slides). The subtitle of this 2011 course? From penetration testing to threat emulation. This course was a lot of fun to teach. I taught my students how to execute targeted phishing attacks against then-modern enterprises. We covered how to build a system profiler, cobble together user-driven attacks, and how to send a phishing email by hand.
The timing of this course was good. Shortly after that course, I set to work on a functional specification for what would become Cobalt Strike. It’s hard to pick an initial feature set for a product. You don’t want to pick 10 features and take each of them to the 10% point. The trick is to find 1 or 2 things that are important and take each of them to the 80-90% point. I decided to use my 2011 course as a guide. Cobalt Strike would provide features for actions that 2011’s Advanced Threat Tactics students had to do by hand.
Cobalt Strike’s 2012 release consisted of Armitage, a targeted attack process, and a reporting engine. To me, Cobalt Strike’s initial release was a big blank canvas. I perceived a lot of gaps in the Metasploit Framework and other tools when applied to the red team problem set. Cobalt Strike was my opportunity to work full-time on these and see what I could come up with. Cobalt Strike’s 2012 initial release also came with a new course: Penetration Testing with Cobalt Strike.
In the 12 months after Cobalt Strike’s first release, I got a lot done. I introduced Beacon, Cobalt Strike’s Applet Kit, and a model for distributed operations. I also added DNS communication and SOCKS pivoting to Beacon. The DNS piece was significant. I now had a desirable communication option that most didn’t have access to. The SOCKS pivoting capability led to the ‘meterpreter’ command. This command would seamlessly tunnel Meterpreter through Beacon. I now had an alternate communication layer for Metasploit, without sacrificing features or ease of use!
By mid-2013, it was clear that Cobalt Strike’s online material was out of step with what the product had become. I had to redo my online course. I wanted to do it “right” and I wanted it to be as engaging as possible. I bought a microphone so I could make the audio better. I also bought a tablet to whiteboard different concepts during the course. It took about a month to update my slides and record the new course. The result was 2013’s Tradecraft:
As an aside, the whole experience of putting this course together was a nightmare. I had trouble getting used to the tablet and I couldn’t find software to seamlessly change colors without interrupting my presentation. I also had struggles with audio. My office space at the time had too much echo. The best location was the carpeted bedroom in my apartment. To record Tradecraft, I sat cross-legged on my bedroom floor with my laptop propped up on books. If you watch this course, try to imagine this scene.
I was very happy with the final result of the 2013 Tradecraft course. I saw Tradecraft as the culmination of my initial vision for Cobalt Strike. By this point, I had a very solid platform for red team operations, built on top of the Metasploit Framework.
I didn’t feel my work was done yet. While I had a solid process based on the Metasploit Framework, I had run into situations where asynchronous communication with Beacon was my only option. I wanted to know that it was possible to fall back to “just Beacon” and operate in these situations.
I began to work on adding Beacon features to complement the things a skilled operator could do from a command shell. For example, I didn’t add automation to run code on a remote host for lateral movement. An operator could do this from a shell. But, I did add token stealing and privilege escalation features. These features combined with what the operator could do from a shell made for a fairly flexible lateral movement capability.
The biggest shift came with Cobalt Strike 2.1. This is the release where I added PowerShell support to Beacon. This release felt like Christmas. I was able to import PowerSploit and PowerTools scripts and use them as-is. Over night, Beacon gained an amazing amount of post-exploitation capability and automation.
My use of Cobalt Strike 2.1 and feedback from users pointed the same way: The combination of Beacon and PowerShell were enough for the red team problem set. Cobalt Strike’s fallback way to operate had become the preferred way for power users. This realization is when Cobalt Strike 3.0 was born.
It was time to evolve and imagine Cobalt Strike without the Metasploit Framework. I sat down, examined Cobalt Strike’s process, and noted which techniques it relies on. I then mapped out how I would do these things without the Metasploit Framework. In some cases, functionality existed elsewhere. Pass-the-hash with Mimikatz is a good example of this. In other cases, it made sense to build new features into Beacon. The screenshot tool and port scanner are good examples of this. This thought exercise became my functional specification and roadmap for Cobalt Strike 3.0.
Cobalt Strike 3.0 shipped in September 2015. This effort re-aligned Cobalt Strike’s features and workflows around Beacon. I also released 2015’s Advanced Threat Tactics course to cover the modern red teaming process Cobalt Strike 3.0 was built to support.
I’ll end this post with one last thought. Cobalt Strike 3.0’s offensive process is not Cobalt Strike specific. It’s recognition of this fact: a lightweight payload, mimikatz, and PowerShell are the foundations of a modern offensive process. The lightweight payload can be anything, so long as it provides the communication options and flexibility needed to support your operation. 2015’s Advanced Threat Tactics course is significant because it documents this modern process and shows what’s possible with this foundation.