Sunday, August 28, 2016

Links and Updates

Corporate Blogs
Two cool things about my day job is that I see cool things, and get to share some of what is seen through the SecureWorks corporate blog.  Most of my day job can be described as DFIR and threat hunting, and all of the stuff that goes into doing those things.  We see some pretty fascinating things and it's really awesome that we get to share them.

Some really good examples of stuff that our team has seen can be found here, thanks to Phil. Now and again, we see stuff and someone will write up a corporate blog post to share what we saw.  For example, here's an instance where we saw an adversary create and attempt to access a new virtual machine.  Fortunately, the new VM was created on a system that was itself a the new VM couldn't be launched.

In another example, we saw an adversary launch an encoded and compressed PowerShell script via a web shell, in order to collect SQL system identifiers and credentials.  The adversary had limited privileges and access via the web shell (it wasn't running with System level privileges), but may have been able to use native tools to run commands at elevated privileges on the database servers.

Some other really good blog posts include (but are not limited to):
A Novel WMI Persistence Implementation
The Continuing Evolution of Samas Ransomware (I really like this one...)
Ransomware Deployed by Adversary with Established Foothold
Ransomware as a Distraction

I watched Ryan Nolette's BSidesBoston2016 presentation recently, in part because the title and description caught my attention.  However, at the end of the presentation, I was mystified by a couple of things, but some research and asking some questions cleared it up.  Ryan's presentation was based on a ransomware sample that had been discussed on the Cb blog on 3 Aug by the time the BSides presentation went on, the blog post was almost a year old.

During the presentation, Ryan talked about bad guys using vshadow.exe (I found binaries here) to create a persistent shadow copy, mounting that copy (via mklink.exe), copying malware to the mounted VSC and executing it, and then deleting all VSCs.  Ryan said that after all of that, the malware was still running.  However, the process discussed in the presentation wasn't quite right...if you want the real process, you need to look at this Cb blog post from 5 Aug 2015.

This is a pretty interesting technique, and given that it was discussed last year (it was likely utilized and observed prior to that) it makes me wonder if perhaps I've missed it in my own investigations...which then got me to thinking, how would I find this during a DFIR investigation?  Ryan was pretty clear as to how he uses Cb to detect this sort of activity, but not all endpoint tools have the same capabilities as Cb.  I'll have to look into some further testing to see about how to detect this sort of activity through analysis of an acquired image.

Saturday, August 13, 2016

LANDesk in the Registry

LANDesk in the Registry
Some of my co-workers recently became aware of information maintained in the Windows Registry by the LANDesk softmon utility, which is pretty fascinating when you look at it.  The previously-linked post states that, "LANDesk Softmon.exe monitors application execution..." not just installed applications, or just services, but application execution.  The post goes on to state:

Unfortunately, if an application is no longer available the usage information still lives on in the registry.

This goes back to what I've said before about indicators on Windows systems, particularly within the Registry, persisting beyond the deletion or removal of the application, which is pretty awesome.

The softmon utility maintains some basic information about the executed apps within the Software hive, with subkeys named for the path to the app.  The path to the keys in question is:

      MonitorLog\<path to executed app>

Information maintained within the keys includes the following values:
  • Current User
  • First Started
  • Last Started
  • Last Duration
  • Total Duration
  • Total Runs
This site provides information regarding how to decode some of the data associated with the above values; the "First Started" and "Last Started" values are FILETIME objects, and therefore pretty trivial to parse.

This information isn't nearly as comprehensive as something like Sysmon, of course, but it's much better than nothing.

Sysforensics posted a LANDesk Registry Entry Parser script on GitHub, about 2 yrs ago.  Don Weber wrote the original RegRipper plugin back in 2009, and I made some updates to it in 2013.  There's also a plugin that incorporates the data into a timeline.

Wednesday, August 10, 2016


Data Exfil
A question that analysts get from time to time is "was any data exfiltrated from this system?"  Sometimes, this can be easy to determine; for example, if the compromised system had a web server running (and was accessible from the Interwebs), you might find indications of GET requests for unusual files in the web server logs.  You would usually expect to find something like this if the bad guy archived whatever they'd collected, moved the archive(s) into a web folder, and then issued a GET request to download the archive to their local system.  In many cases, the bad guy has then deleted the archive.  With no instrumentation on the system, the only place you will find any indication of this activity is in the web server logs.

However, for the most part, definitive determination of data exfiltration is almost impossible without the appropriate instrumentation; either having a packet sniffer on the wire at the time of the transfer, or appropriate endpoint agent monitoring process creation events (see the Endpoints section below) in order to catch/record command lines.  In the case of endpoint monitoring, you'd likely see the use of an archiving tool, and little else until you moved to the web server logs (given the above example).

Another area to look is the Background Intelligent Transfer Service, or "BITS".  SecureWorks has a very interesting blog post that illustrates one way that this native Windows service has been abused.  I highly suggest that if you're doing any kind of DFIR or threat hunting work, you do a good, solid read of the SecureWorks blog post.

I am not aware of any publicly-available tools for parsing the BITS qmgr0.dat or qmgr1.dat files, but you can use 'strings' to locate information of interest, and then use a hex editor from that point in order to get more specific information about the type of transfer (upload, download) that may have taken place, and it's status.  Also, be sure to keep an eye on those Windows Event Logs, as well.

Finding Bad
Jack Crook recently started a new blog, Finding Bad, covering DFIR and threat hunting topics.  As

Jack's most recent post on hunting for lateral movement is a good start, but IMHO, the difference in artifacts on the source vs the destination system during lateral movement needs to be clearly delineated.  Yeah, I may be pedantic, but from my perspective, there is actually a pretty huge difference, and that difference needs to be understood, for no other reason that because the artifacts on each system are different.

Adam recently posted a spreadsheet of various endpoint solutions that are's interesting to see the comparison.  Having detailed knowledge of one of the listed solutions does a level set with respect to my expectations regarding the others.

MAC Addresses in the Registry
I recently received a question from a friend regarding MAC addresses being stored in the Registry.  It turns out, there are places where the MAC address of a system is "stored" in the Registry, just not in the way you might think.  For example, running the RegRipper plugin, we see (at the bottom of the output) something like this:

Unique MAC Addresses:

I should also point out that the plugin, which is about 8 yrs old at this point, also might provide some information.

Registry Findings - 2012
The MAC Daddy - circa 2007

I ran across EventMonkey (wiki here) recently, which is a Python-based event processing utility.  What does that mean?  Well, from the wiki, that means "A multiprocessing utility that processes Windows event logs and stores into SQLite database with an option to send records to Elastic for indexing."  Cool.

This definitely seems like an interesting tool for DFIR analysts.  Something else that the tool reportedly does is process the JSON output from Willi's EVTXtract.

As I've mentioned before, later this month I'll be presenting at ArchC0N, discussing some of the misconceptions of ransomware.  I ran across an interesting blog post recently regarding, Fixing the Culture of Infosec Presentations.  I can't say that I fully agree with the concept behind the post, nor with the contents of blog post.  IMHO, the identified "culture" is from too narrow a sample of's unclear as to which "infosec conferences" this discussion applies.

I will say this...I stopped attending some conferences a while back because of the nature of how they were populated with speakers.  Big-named, headliner speakers were simply given a time slot, and in some cases, did not even prepare a talk.  I remember one such speaker using their time slot to talk about wicca.

At one point in the blog post, the author refers to a particular presenter who simply reads an essay that they've written; the author goes on to say that they prefer that.  What's the point?  If you write an essay, and it's available online, why spend your time reading it to the audience, when the audience is (or should be) fully capable of reading it themselves?

There's another statement made in the blog post that I wanted to comment on...

We have a force field up that only allows like .1% of our community to get on the stage, and that’s hurting all of us. It’s hurting the people who are too afraid to present. It’s hurting the conference attendees. And it’s hurting the conferences themselves because they’re only seeing a fraction of the great content that’s out there.

I completely agree that we're missing a lot of great content, but I do not agree that "we have put up a force field"; or, perhaps the way to look at it is that the force field is self-inflicted.  I have seen some really good presentations out there, and the one thing that they all have in common is that the speaker is comfortable with public speaking.  I say "self-inflicted" because there are also a lot of people in this field who are not only afraid to create a presentation and speak, they're also afraid to ask questions, or offer their own opinion on a topic.

What I've tried to do when presenting is to interact with the audience, to solicit their input and engage them.  After all, being the speaker doesn't mean that I know everything...I can't possibly know or have seen as much as all of us put together.  Rather than assume the preconceptions of the audience, why not ask them?  Admittedly, I will sometimes ask a question, and the only sound in the room is me breathing into the mic...but after a few minutes folks tend to start loosing up a bit, and in many cases,  a pretty good interactive discussion ensues.  In the end, we all walk away with something.

I also do not believe that "infosec presentations" need to be limited to the rather narrow spectrum described in the blog post.  Attack, defend, or a tool for doing either one.  There is so much more than that out there and available.  How about something new that you've seen (okay, maybe that would part of "defend"), or a new way of looking at something you've seen?  Want a good example?  Take a look at Kevin Strickland's blog post on the Evolution of the Samas Ransomware.  At the point where he wrote that blog post, SecureWorks (full disclosure, Kevin and I are both employed by SecureWorks) had seen several Samas ransomware cases; Kevin chose to look at what we'd all seen from a different perspective.

There are conferences that have already gone to taking at least some of the advice in the blog post. For example, the last time I attended a SANS360 presentation, there were 10 speakers, each with 6 minutes to present.  Some timed their presentations down to the second, while others seemed to completely ignore the 6 minute limit on presentations.  Even so, it was great to see a speaker literally remove all of the fluff and focus on one specific topic, and get that across.

Sunday, July 17, 2016


Book Update
I recently and quite literally stumbled across an image online that was provided as part of a challenge, involving a compromised system.  I contacted the author, as well as discussed the idea of updating the book with my tech editor, and the result of both conversations is that I will be extending the book with an additional chapter.  This image is from a Windows 2008 web server that had been "compromised", and I thought it would be great to add this to the book, given the types of "cases" that were already being covered in the book.

RDP Bitmap Cache Parser
I ran across this French web site recently, and after having Google translate it for me, got a bit better view into what went into the RDP bitmap cache parsing tool. This is a pretty fascinating idea; I know that I've run across a number of cases involving the use of RDP, either by an intruder or a malicious insider, and there could have been valuable clues left behind in the bitmap cache file.

Pancake Viewer
I learned from David Dym recently that Matt (works with David Cowen) is working on something called the "pancake viewer", which is as DFVFS-backed image viewer using wxPython for the GUI.  This looks like a fascinating project, and something that will likely have considerable use, particularly as the "future functionality" is added.

Web Shells
Web shells are nothing new; here is a Security Disclosures blog post regarding web shells from 3 years ago.  What's "new" is what is has recently been shared on the topic of web shells.

This DFIR.IT blog post provides some references (at the bottom of the post) where other firms have discussed the use of web shells, and this post on the SecureWorks site provides insight as to how a web shell was used as a mechanism to deploy ransomware.

This DFIR.IT blog post (#3 in the series) provides a review of tools used to detect web shells.

A useful resource for Yara rules used to detect web shells includes this one by 1aNOrmus.

Wednesday, July 06, 2016

Updates and Stuff

Registry Analysis
I received a question recently, asking if it were possible to manipulate Registry key LastWrite time stamps in a manner similar to file system time stomping.  Page 30 of Windows Registry Forensics addresses this; there's a Warning sidebar on that page that refers to SetRegTime, which directly answers this question.

The second part of the question was asking if I'd ever seen this in the wild, and to this point, I haven't.  Then I thought to myself, how would I identify or confirm this?   I think one way would be to make use of historical data within the acquired image; let's say that the key in question is available within Software hive, with a LastWrite time of 6 months prior to the image being acquired.  I'd start by examining the Software hive within the RegBack folder, as well as the Software hives within any available VSCs.  So, if the key has a LastWrite time of 6 months ago, and the Software hive from the RegBack folder was created 4 days ago and does NOT include any indication of the key existing, then you might have an issue where the key was created and time stomped.

Powershell Logging
I personally don't use Powershell (much, yet), but there have been times when I've been investigating a Windows system and found some entries in the Powershell Event Log that do little more than tell me that something happened.  While conducting analysis, I have no idea if the use of Powershell is pervasive throughout the infrastructure, if it's something one or two admins prefer to use, or if it was used by an attacker...the limited information available by default doesn't give me much of a view into what happened on the system.

The good news is that we can fix this; FireEye provides some valuable instructions for enabling a much greater level of visibility within the Windows Event Log regarding activity that may have occurred via Powershell.  This Powershell Logging Cheat Sheet provides an easy reference for enabling a lot of FireEye's recommendations.

So, what does this mean to an incident responder?

Well, if you're investigating Powershell-based attacks, it can be extremely valuable, IF Powershell has been updated and the logging configured. If you're a responder in an FTE position, then definitely push for the appropriate updates to systems, be it upgrading the Powershell installed, or upgrading the systems themselves.  When dealing with an attack, visibility is don't want the attacker moving through an obvious blind spot.

This is also really valuable for testing purposes; set up your VM, upgrade the Powershell and configure logging, and then go about testing various attacks; the FireEye instructions mentioned above refer to Mimikatz.

Powershell PSReadLine Module
One of the things that's "new" to Windows 10 is the incorporation of the PSReadLine module into Powershell v5.0, which provides a configurable command history (similar to a bash_history file).  Once everything's up and running correctly, the history file is found in the user profile at C:\Users\user\AppData\Roaming\Microsoft\Windows\PowerShell\PSReadline\ConsoleHost_history.txt.

For versions of Windows prior to 10, you can update Powershell and install the module.

This is something else I'd definitely upgrade, configure and enable in a testing environment.

CLI Tools
As someone who's written a number of CLI tools in Perl, I found this presentation by Brad Lhotsky to be pretty interesting.

Tool Listing
Speaking of tools, I recently ran across this listing of tools, and what caught my attention wasn't the listing itself as much as the format.  I like the Windows Explorer-style listing that lets you click to see what tools are listed...pretty slick.

I was doing some preparation for a talking I'm giving at ArchC0N in August, and was checking out the MMPC site.  The first thing I noticed was that the listing on the right-hand side of the page had 10 new malware families identified, 6 of which were ransomware.

My talk at ArchC0N is on the misconceptions of ransomware, something we saw early on in Feb and March of this year, but continue to see even now.  In an attempt to get something interesting out on the topic, a number of media sites were either adding content to their articles that had little to do with what was actually happening at a site infected with ransomware, or they were simply parroting what someone else had said.  This has let to a great deal of confusion as infection vectors for ransomware in general, as well as what occurred during specific incidents.

More on the Registry...
With respect to the Registry, there're more misconceptions that continue unabated, particularly from Microsoft.  For example, take a look at the write-up for the ThreatFin ransomware, which states:

It can create the following registry entries to ensure it runs each time you start your PC:

The write-up then goes on to identify values added to the Run key within the HKCU hive.  However, this doesn't cause the malware to run each time the system is started, but rather when that user logs into the system.

Monday, June 20, 2016

New Book

So, yeah...I'm working on another book (go figure, right?).  This one is different than the previous books I've written; rather than listing the various artifacts available within an acquired image, the purpose of this book is to provide a walk-through of the investigative process, illustrate how the various artifacts can be used to complete analysis, and more importantly, illustrate and describe the various decisions made throughout the course of the examination.  The focus of this book is the process, and all along the way, various "analysis decisions" will be highlighted and detailed.

The current table of contents, with a short description of each chapter, is as follows:

Chapter 1 - Introduction
Introduction to the core concepts that I'll be reinforcing throughout the remaining chapters of the book, including documentation (eeewww, I know, right?).

Chapter 2 - Malware Detection Scenarios
In Ch 2, there are two malware detection scenarios.  Again, these are detection scenarios, not analysis scenarios.  I will discuss somethings that an analyst can do in order to move the analysis along, documenting and confirming the malware that they found, but there are plenty of resources available that discuss malware analysis in much greater detail. One of the analysis scenarios that I've seen a great deal of during my time as a DFIR analyst has been, "...we don't for sure, but we think that this system may have malware on it..."; as such I thought that this would be a great scenario to present.

In this chapter, I will be walking through the analysis process for two scenarios, one using a WinXP image that's available online, the other using a Win7 image that's available online.  That way, when reading the book, you can download a copy of the image (if you choose to do so) and follow along with the analysis process.  However, the process will be detailed enough that you won't have to have the image available to follow along.

Chapter 3 - User Activity Scenarios
This chapter addresses tracking user activity during an examination, determining/following the actions that a user took while logged into the system.  Of course, a "user" can also be an intruder who has either compromised an account, or created one that they're now using.

As with chapter 2, I'll be walking through two scenarios, one using a WinXP image, the other using a Win7 image, both of which are available online.

The purpose of chapters 2 and 3 is to illustrate the end-to-end analysis process; its not about this tool or that tool, its about the overall process.  Throughout the scenarios, I will be presenting analysis decisions that are made, describing why I decided to go a certain direction, and illustrating what the various findings mean to the overall analysis.

Chapter 4 - Setting up and using a test environment
Many times, an analyst may need to test a hypothesis in order to confirm (or deny) the creation of an artifact or indicator.  Or, the analyst may opted to test malware or malicious documents to determine what occurred on the system, and to illustrate what the user saw, and what actions the user had to have taken.  In this chapter, we'll walk through setting up a virtual environment that would allow the analyst to test such things.

This may seem like a pretty obvious chapter to many...hey, this sort of thing is covered in a lot of other resources, right?  Well, something I see a great deal of, even today, is that these virtual testing environments are not instrumented in a way that provides sufficient detail to allow the analyst to then collect intelligence, or propagate protection mechanisms through their environment.

This chapter is not about booting an image.  There are plenty of resources out there that address this topic, covering a variety of formats (i.e., "...what if I have an *.E01 image, not a raw/dd image...?").

Chapter 5 - RTFM for DFIR
If you're familiar with the Red Team Field Manual, chapter 5 will be a DFIR version of this manual.  Like RTFM, there will not be detailed explanations of the various tools; the assumption is made that you (a) already know about the tool, or (b) will put in the effort to go find out about the tool.  In fact, (b) is relatively easy...sometimes just typing the name of the CLI tool at the prompt, or typing the name followed by "/?", "-h", or "--help" is all you really need to do to get a description of the tool, its syntax, and maybe even example command lines illustrating how to use the tool.

Okay, so, yeah...I know that this is a bit different from the way I've done things in the past...most often I've just posted that the book was available.  With my last book, I had a "contest" to get submissions for the book...ultimately, I just got one single submission.

The reason I am posting this is due to this post from the ThisWeekIn4n6 blog, specifically this statement...

My only comment on this article is that maybe he could be slightly more transparent with how he’s going in the book writing process. I recall seeing a couple of posts about the competition, and then the next one was that he had completed the book. Unfortunately I missed the boat in passing on some research into the SAM file (by several months) however Harlan posted about it here.
With that in mind, I imagine he will be working on an update to Windows Forensic Analysis to cover some additional Windows 10 artifacts (and potentially further updates to other versions). Maybe a call out (yes, I know these haven’t been super successful in the past; maybe a call out to specific people? Or universities?)....

With respect to the Windows Registry Forensics book, I thought I was entirely "transparent"...I asked for assistance, and in the course of time...not just to the "end" of the contest time limit, but throughout the rest of the time I was writing the book...I received a single submission.

The "Ask"
Throughout the entire time that I've written books, the one recurring question that comes up over and over again is, "...does it cover Windows ?"  Ever time the question is asked, I have the same, because I don't have access to that version of Windows.

This time, in an attempt to head off those questions, I'm putting out a request to the DFIR community at large.  Specifically, if you have access to an image of a Windows 10 system (or to an image of any of the server versions of Windows after 2003) that have been compromised in some manner (i.e., malware, unauthorized access, etc.), and are worth of investigation, can you share them?  The images I'm using in this book are already available online, and I'm not asking that these images also be available online; if you don't mind sharing a copy of the images with me, I will walk through the analysis and include it in the book, and I will destroy/return the images after I'm done with them, whichever you would like.  

Anyone who shares an image of a Windows server version beyond (not including) Windows 2003, or an image of a Windows 10 system, for which I can include the analysis of that image in my book will receive a free, signed (yes, by me...) copy of the book once it comes out.

Addendum: Something that I wanted to add for clarity...I do not have, nor do I have access to, any system (or an image thereof) running Cortana, or anything special.  The laptop that I write the books (and blog posts) from is a Dell Latitude E6510.  My point is that if you have questions such as, "what are the artifacts of someone using Cortana?" or of any other application specific to Windows 10, please understand that I do not have unlimited access to all types of equipment.  This is why I made the request I did in this blog post.


Data Hiding Techniques
Something I caught over on Weare4n6 recently was that there's a new book on it's way out, due to be available in Oct, 2016, entitled, "Data Hiding Techniques in Windows OS".  The description of the book on the blog is kind of vague, with references to steganography, but I'm hoping that it will also include discussions of things like ADSs, and using Unicode to "hide" files and Registry keys/values in plain sight.

My initial reaction to the book, however, had to do with the cover.  To the right is a copy of the cover of my latest book, Windows Registry Forensics.  Compare that to the book cover at the web site.  In short, it looks as if Elsevier is doing it again...previous editions of my books not only had the same color scheme amongst the books, but shared that color scheme with books from other authors.  This led to massive confusion; I once received a box of books just before I left for a conference, and I took a couple of copies to give away while I was there.  When I tried to give the books away, people told me that they "...already had the book...", which I thought was odd because I had just received the box.  It turned out that they were looking at the color scheme and not actually reading the title of the book.

Right now I have nine books on my shelf that have the same or very similar color and green.  I am the sole author, or co-author, of four of them.  The other five have an identical color scheme...a slightly different shade of green...but I am the author of only one of them.

Mari's got another great post up, this one about using a Linux distro to image a Mac.  Yeah, I know, this blog isn't about Macs, but this same method could potentially be used to image a Windows server that you're having trouble with.  Further, Mari is one of the very few people within the community who develops and shares original material, something the community needs to encourage with more analysts.

Carberp Updates
A couple of articles appeared recently regarding changes to Carberp (PaloAltoNetworks, TrendMicro), specifically with respect to the persistence mechanism.  The PAN article mentions that this may be used to make sandbox analysis much more difficult, in that it requires user interaction to launch the malware.

The PAN article ends with a bunch of "indicators", which consist of file hashes and domain names.  I'd monitor for the creation of the key/value instead.

Some online research indicated that this persistence mechanism had been discussed on the Hexacorn blog over 2 years ago.  According to the blog post, the persistence mechanism causes the malware to be launched when the user launches an Office application, or IE with Office plugins installed.  This can make IIV determination difficult if the analyst isn't aware of it.

I updated the RegRipper plugin accordingly.

Wednesday, June 08, 2016


RegRipper Plugin Updates
I recently released some updates to RegRipper plugins, fixing code (IAW Eric's recent post) and releasing a couple of new plugins.  As such, I thought it might be helpful to share a bit about how the plugins might be used during an exam. - honestly, I'm not entirely sure how I'd use this plugin during analysis, beyond using it to document basic information about the system. - I can see using this plugin for just about any engagement where the program execution category of artifacts is of interest.  Remember the blog post about creating an analysis matrix?  If not, think malware detection, data exfil, etc.  As Eric mentioned in his recent webcast, you could use indicators you find in the ShimCache data to pivot to other data sources, such as the AmCache.hve file, Prefetch files, Windows Event Log records with sources such as "Service Control Manager" in a timeline, etc.  However, the important thing to keep in mind is the context of the time stamps associated with each file entry...follow the data, don't force it to fit your theory.

Specific things I'd look for when parsing the ShimCache data include entries in the root of the Recycle Bin, the root of the user's profile, the root of the user's AppData/Roaming and Desktop folders, in C:\ProgramData, and with a UNC path (i.e., UNC\\tsclient\...).  Clearly, that's not all I'd look for, but those are definitely things I'd look for and be very interested in.  At one point, I'd included "alerts" in the output of some plugins that would automatically look for this sort of thing and alert the analyst to their presence, but there didn't seem to be a great deal of interest in this sort of thing. 

Win10 Notification DB
During his recent webcast regarding the AmCache.hve file, Eric mentioned the SwiftForensics site a couple of times.  Well, it turns out that Yogesh has been on to other things since he posted about the AmCache.hve file...he recently posted a structure description for the Windows 10 Notification database.  Yogesh also included a Python script for parsing the notification database...if you're examining Windows 10 systems, you might want to check it out.

I don't have enough experience yet examining Windows 10 systems to know what sorts of things would be of value, but I can imagine that there would be considerable value in this data, in a case where the user claimed to not have received an email, only to have an examiner pull a snippet of that email from the notification database, for example.

Speaking of Yogesh's comments regarding the AmCache.hve file, one of his posts indicates that it would be a goldmine for malware hunters.  As I mentioned in my previous post on the subject, in the past two months, I've examined two Windows 7 systems that were known to be infected with malware, and while I found references to the malware files in the AppCompatCache, I did not find references to the files in the AmCache.hve file.

To be clear, I'm not saying that either Yogesh's or Eric's comments are incorrect...not at all.  I'm not saying that, suggesting that, or trying to imply that.  What I am saying is that I haven't seen it yet...but also like I said before, that doesn't mean that I'm going to stop looking.

I haven't mentioned Prefetch artifacts in this blog for a while, as I really haven't had any reason to do so.  However, I recently ran across Luis Rocha's Prefetch Artifacts post over on the CountUponSecurity blog, and I found it to be a pretty valuable reference.