From About Beagle:
Beagle is beautifully documented, among the best I’ve ever seen, I strongly suggest reading it in its entirety before proceeding here. Well done, on so many fronts, to Omer Yampel, @yampelo, the project lead. I’m going to limit reprinting that content here and focus almost exclusively on specific use cases.
Omer’s Beagle Roadmap:
Thanks to Omer for all the detail!
A quick primer on creating the Beagle Docker instance and running it with ease. I found it easiest to operate from a regular command prompt.
Browse to http://localhost:8000 and your Beagle instance will be waiting for, ready to ingest and graph. In terms of said ingestion, Beagle stands ready to consume:
Uploading is as simple as select and click. Choose the file type, select the appropriately matched file, click Upload Data, wait a few seconds, then Submit as seen in Figure 1.
Figure 1: Beagle upload
For purposes of our toolsmith experiments I choose some select samples from my historical archive, including a number of memory images, Procmon files, and an exemplary Windows security event log. Don’t let the fact that I didn’t test VT, FireEye, or Sysmon samples limit you from doing so. Ease of use is substantial here, the only limiting factor is the potential for bloating your Docker image with logs or images ingested. If you’re going to turn Beagle into to a workhorse for your blue team and DFIR workloads, I strongly recommend you make use of Beagle’s Python API on a dedicated system with fairly beefy resources to enhance performance.
Figure 2: Procmon menu
I’ll begin with a graph for the Procmon CSV file generated from the above mentioned invoice malware analysis. Beagle includes an excellent Node Search feature; given that the opening graph is not particularly revealing I thought it best to try to zoom to a featured node. Rather than create static images that fail to do Beagle justice representing its usefullness, I’ve created video captures of the click-throughs, similar to those on Omer’s GitHub site. I first searched the keyword invoice and immediately landed on the sample execution process labeled Invoice_Trailcore_100355038.exe. I selected that process entity in the search results, then double-clicked the node in the graph. Figure 3 speaks for itself at this point.
Figure 3: Malicious invoice process actions
Don’t forget to drill in to Tree, Timeline, and Table views in addition to the Graph. They’re equally revealing dependent on your goals. Forensicators will truly appreciate Timeline in particular.
Figure 4: Invoice malware similarities across Procmon and memory samples
I noted that in the Procman graph we also see clear evidence of the sample execution causing a crash dump. You can see the node representing drwtsn32.exe, for the Dr. Watson crash dump service. There are also related calls to dwwin.exe, the Watson client.
One of favorite memory images is one from years ago taken from a system compromised with Trojan.APT.9002, the last version before the adversary went completely diskless, one of the very first to run in memory only. As we observe the graph in its opening state, we don’t note much of merit. I used Beagle’s node search to hunt for a known suspect process in this memory image, specifically 3176. It was an extremely easy pivot from there. In this case, 3176 turned out to be associated with rundll32.exe, which, in addition to reading nework connection-related registry keys, immediately tapped Internet history via index.dat. Experience the drill-down in Figure 5.
Figure 5: PID 3176 up to no good
My final example is another of my favorites, exhibiting malicious behavior captured in a Windows security event log from a log grabbed via a Red Team server as they were popping off with some of their typical mayhem. The Red Team we partner with, and I do mean partner (Go, Purple Team, go!), are artistic and advanced the majority of the time. I condsider them 2-3 years ahead of the industry and peer teams in mindset and tactics. Given the Purple Team approach we embraced years ago, their methods and approach serve both their customer and the Blue Team. We are definitely better for their efforts. But once in awhile they’ll go loud, and it visulizes gloriously. This graph represents one of the Red Team’s special multifunction payloads, I’ll give you no more detail than that, but enjoy the ride in Figure 6.
Figure 6: Red Team goes loud
Seems to me that said payload gets busy quickly, you likely noted that it launched numerous processes instantly, including cmd.exe, schtasks.exe, and taskeng.exe. Clearly, they’re establishing a swift foothold and seeking to persist. I love how Beagle represents processes in red, and the Red Team’s payload spawns yet more processes. Visual Red Team love for sure. Want a quick, dynamic graph, or would you prefer to hunt out those same spawned processes in a security event log without visual aid?
This is another case where I have not given an excellent tool its full due, but it’s so robustly documented, and well received by the community via significant infosec social media presences, as well as the likes Black Hat Asia Arsenal (shout out to @ToolsWatch. Again, please read the GitHub documentation in its entirety and consider leveraging the Python library after experimenting with the Docker image.
Apr 5th 2019
1 month ago