If you are a security researcher or even a passionate about how attackers implement phishing you will find yourself to look for phishing kits. A phishing kit is not a phishing builder, but a real implementation (actually re-implementation) of a third party website built to lure your victim. Initially attackers use a phishing builder to “clone” the original web site but after that they introduce – in the fresh re-generate website – interesting ad-dons such as for example: evasion techniques (in order to evade to phishing detectors), targeted elements (in order to targetize the victims), fast re-directors ( to follows the attack chain into the original web-site or to a relay to try to infect you) and sometimes exploit-kits to try to exploit your browser before letting you go.

Motivation

There are places where you can buy PhishingKits, for example BleepingComputer wrote a great article on that here, but if you want to get them for free in order to study attack schema and Kit-composition you don’t’ find collections for free. So I decided to share my PhishingKit Tracker, updated automatically by my backend engine every day for study and research purposes.

You can find it HERE (PhishingKitTracker github repo)

Disclaimer

This repository holds a collection of Phishing Kits used by criminals to steal user information. Almost every file into the raw folder is malicious so I strongly recommend you to neither open these files, nor misuse the code to prank your friends. Playing with these kits may lead to irreversible consequences which may affect anything from personal data to passwords and banking information.

I am not responsible for any damage caused by the malware inside my repository and your negligence in general.

NB: Large File System Hahead

PhishingKitTracker is stored into Git Large File System (git-lfs) due to the big amount of data tracked. You should install git-lfs before cloning this repository.

RAW Data

In raw folder are tracked the Phishing Kits in the original format. No manipulation are involved in that data. A backend script goes over malicious harvested websites (harvesting from common sources) and checks if Phishing Kits are in there. In a positive case (if a PhishingKit is found) the resulting file is downloaded and instantly added to that folder. This folder is tracked by using Git Large File System since many files are bigger than 100MB. The “RAW Data” is a quite unexplored land, you would find many interesting topics with high probability. Please remember to cite that work if you find something from here, it would be very appreciated.

STATS

In stats folder are maintained two up-to-date files:

1. files_name it holds the frequency of the found file-names associate with kits. In other words every phishing kit is saved on the phishing host with a name. filke_name keeps track about every file names and its frequency. If you are wondering why am I not tracking hashes, is because phishing kits are big compressed archives, so it would make no sense at this stage since they always differ each other (but check in src folder for additional information)
2. sites hols the frequency of the hosting domain names. In other words where the phishing kit was found. No duplicates are tracked by meaning that the frequency and the file names are unique. So for example if you see something like: 3 li.humanbiomics-project.org it means that in li.humanbiomics-project.org have been found three different Phishing Kits over time.

Both of these files have been generate by simple bash scripts like:

• ls raw/ | cut -d'_' -f1 | uniq -c | sort -bgr > stats/sites.txt
• ls raw/ | cut -d'_' -f2 | uniq -c | sort -bgr > stats/files_name.txt

these scripts are run on every commit making files inline with the raw folder.

On the other side a file called similarity.csv is provided with a tremendous delay due to the vast amount of time in generating it. That file provides the similarity between the tracked Phishing Kits. It’s a simple CSV file so that you can import it on your favorite spreadsheet and make graphs, statistics or manipulate it in the way you prefer.

SIMILARITY.CSV structure

The similarity structure is like the following one: FileA,FileB,SimilarityAVG,SimilarityMin,SimilarityMax where:

• FileA is PhishingKit which is considered in that analysis.
• FileB is the PhishingKit to be compared to PhishingKit FileA
• SimilarityAVG is the Average in similarity. That average is calculated by computing the similarity check to every single (interesting) file in the PhishingKit archive (FileA) to every single (interesting) file in the PhishingKit archive to be compared (FileB)
• SimilarityMin is the lowest similarity value found between PhishingKitA and PhishingKitB
• SimilarityMax is the highest similarity value found between PhishingKitA and PhishingKitB

If you want to generate similarity.csv by your own I provide a simple and dirty script into the src folder. So far it has several limitations (for example it computes ZIP only files). please make pull requests for improving and empower it. Each contribute would be very helpful.

SRC

Please check those variables (compute_similarity.py) and change them at your will.

EXTENSION_FOR_ANALYSIS = ['.html','.js','.vbs','.xls','.xlsm','.doc','.docm', '.ps1']
OUTPUT_FILE =  'similarity.csv'                                                 
RAW_FOLDER = '/tmp/raw/'                                                        
TEMP_FOLDER = '/tmp/tt'     

Once you’ve changed them you can run the script and take a long rest. It will navigate through the RAW_FOLDER, grab the .zip files and tries to compute code similarity between them. At the very end it will save results into OUTPUT_FILE. From now you can import such a a file into your favorite spreadsheet processor and elaborate the code similarity.

So far the python script is able to only compare zip tracked phishingkit, for different compressed format it’s still work in progress.

NB: The Python script is in a super early stage of development. Please help to improve it.

How to contribute

Introducing the walking script for different compression formats. In other words if you want to contribute you can write a new section such as the following one (code_similarity.py) but for different compression extensions such as: .tar.gz, .tar, .rar. /7z and so on and so forth.

# Extracts Zip files based on EXTENSION_FOR_ANALYSIS. It returns the etire file
# path for future works
def extractZipAndReturnsIntereistingFiles(file_to_extract):
    interesting_files = []
    n_interesting_files = []
    try:
        with ZipFile(file_to_extract, 'r') as zipObj:
            listOfFileNames = zipObj.namelist()
            for fileName in listOfFileNames:
                for ext in EXTENSION_FOR_ANALYSIS:
                    if fileName.endswith(ext):
                        try:
                            zipObj.extract(fileName, TEMP_FOLDER)
                            interesting_files.append(os.path.join(TEMP_FOLDER, fileName))
                        except Exception as e:
                            continue
                    else:
                        n_interesting_files.append(os.path.join(TEMP_FOLDER, fileName))
    except Exception as e :
        return interesting_files
    return interesting_files

One more way to contribute is to make the comparison loop smarter and quicker. You might decide to parallelized task by forking and spawning more process or by changing the way I use multi-threading in this quick and dirty statistic script. In conclusion every working pull is welcomed.

Cite the Phishing Kit

@misc{ MR,
       author = "Marco Ramilli",
       title = "Phishing Kits Tracker",
       year = "2020",
       url = "https://marcoramilli.com/2020/07/13/introducing-phishingkittracker/",
       note = "[Online; July 2020]"
     }