In a previous post, we discussed the “@” symbol used to separate an apparent legitimate URL from the real target. In this case, there has been a small flood using the URL of “http://jmcglone.com@” with many different URLs or IP addresses after the “@” symbol. If we look at the VirusTotal information for this page, we see the online scan says it is clean and that it has also been around for ten years.
This web page is never actually called, it is the URL after the “@” symbol that is called!
Running a search in InQuest Labs for that domain name we find there are currently 116 hits for It.
We usually find that these documents have little or no content in them and have an external relationship or external link. File found here.
Here we see how they are sent, and only the data after the “@” is actually called. Again these will download an RTF document and using CVE-2017-1182 it will decode the embedded shellcode and then call out to download the final malware.
We find that this site has an open directory. Pretty much every one checked had an open directory.
As you can see, the file extension says “doc," but it is in fact an obfuscated RTF. The obfuscation is to split up the hex string of the shellcode using various spaces, tabs, newlines, and vertical tabs. For some reason, Office will ignore those characters and run the shellcode.
Using a highly experimental tool I wrote to extract the decoded shellcode we can see that this will call out to.
We can see that this file is in an open directory too.
Running this sample thru a sandbox ID’s it as GuLoader. You can find the sample Here on MalwareBazaar.
Using the Indicator lookup on the IP for the downloaded file we can see there were several hits for it and the IP is detected as malicious.
Pivoting on the first link from the Lookup window to UrlHaus we can see that the same IP is used for more than one path. We can also see that it will contain many different malware families with the same File name. Looking at a second file, we can see that the URL in front of the “@” symbol is different.
Using the Indicator Lookup we find that we have 28 files currently with that URL.
Going back and looking at the IP with the indicator lookup we see we still only have 28 files of various sizes.
Here we can see that there are multiple directory’s using that IP address.
Clicking the First link we can go to URLhaus, and we can see an indicator that they are already offline and it is an open Directory. I captured this sample before this screenshot above so what does this open directory look like.
Here we see it has multiple sub-directories.
Opening the first one we see that in this case there is only one file per subdirectory. They all had the same filename name as well.
Going thru each subfolder and decoding the shellcode gives us this list of links to the final malware files. At the time of writing this not all links were still active so the final malware was not Downloaded.
Here is a unique list of hashes for those files that could be downloaded. The detected families varied and it even contained a putty.exe. The next thing we find is a series of Link shorteners used.
If we look at this one posted on Twitter by InQuest here .
Here we can see it expands to the original style of link we first seen.
This one uses two different link shortners.
As you can see from the screenshot above it follows the same pattern. This screenshot was just taken so the file may be down now noting the 404. As we have seen here they are reusing the IPs and URLs leaving an open directory. They just keep adding and removing sub-directories for the same IPs.
They are also experimenting with various Link shortening services to hide further the URLs they are calling out to. Lastly, we also see multiple malware families being dropped as the final malware.
So you never know what you are going to end up with these.
Previous Post Link:
Link to the first sample on InQuest Labs:
Link to sample on MalwareBazaar:
Link to the Second file on InQuest Labs:
Link to Twitter post for first shortened Link:
Link to m4n0w4r Twitter post:
Link to InQuest labs for sample: