<?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 Omar Ahmed on Medium]]></title>
        <description><![CDATA[Stories by Omar Ahmed on Medium]]></description>
        <link>https://medium.com/@omarahmed_13016?source=rss-c71f2cf0306f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*ZRtt5gTY_o4ngIeD</url>
            <title>Stories by Omar Ahmed on Medium</title>
            <link>https://medium.com/@omarahmed_13016?source=rss-c71f2cf0306f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:01:24 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@omarahmed_13016/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[Leaking PII at Scale: How Third Parties Can Unintentionally Put Your Data at Risk.]]></title>
            <link>https://medium.com/@omarahmed_13016/leaking-pii-at-scale-how-third-parties-can-unintentionally-put-your-data-at-risk-6101fcb3d5e0?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/6101fcb3d5e0</guid>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[writeup]]></category>
            <category><![CDATA[vulnerability]]></category>
            <category><![CDATA[hackerone]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Fri, 17 Jan 2025 13:57:41 GMT</pubDate>
            <atom:updated>2025-01-17T14:02:53.845Z</atom:updated>
            <content:encoded><![CDATA[<p>Hello Everyone, I wanted to share with you how I discovered a misconfiguration on a website offering software services and was able to enumerate all organizations relying on it that had that misconfiguration.</p><p>The target offers you the ability to build custom apps with “ no code “ meaning it’s an easy way for you to get an application mainly for forms/user requests/ etc.</p><p>To further clarify this situation, it’s important to note that while the platform in question wasn’t a traditional third-party service, it acts as a foundational tool for building custom applications. The vulnerability here could be classified as a <strong>supply chain vulnerability</strong> because the platform’s internal misconfiguration affects all websites and applications built on it. This highlights a critical risk in relying on third-party platforms or services for web development — misconfigurations in these platforms can introduce vulnerabilities across a wide range of sites and applications.</p><p>It all started with a subdomain on HackerOne, which was leveraging this no-code platform. I noticed a request being sent in the background that looked something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lnKI7Z4yDnvvgfSHv-42Ug.png" /></figure><p>At first glance, the request seemed to contain nothing but sheet and table names along with some data that didn’t appear to be of much interest. But curiosity got the better of me, and I decided to dig deeper into the behavior behind it. I wanted to understand what these tables were and, more importantly, how I could access the native tables they referred to.</p><p>After some investigation, I found that there was another background request being sent to load example data from Firestore. This was when things started to click.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*skE6yc2w3duNyvUCKK4yoQ.png" /></figure><p>I analyzed the request and realized that it was structured like this:</p><blockquote><a href="https://firestore.googleapis.com/v1/projects/glide-prod/databases/(default)/documents/glide-apps-v4-data/w3F1fSiO0tERgXUJLDgY/tables/_native-table-oSUHfdYNOb4wzpZJiC8i/rows">https://firestore.googleapis.com/v1/projects/TargetName/databases/(default)/documents/Target-apps-v4-data/subdomainId/tables/_nativeTableId/rows</a></blockquote><p>Now that I had figured out the process, I understood that all I needed to do was obtain the table IDs, and I could start collecting information from these tables. However, things weren’t as simple as I initially thought. Not all of the tables were publicly accessible.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e6NUAimIHTInbT7-Hdnteg.png" /></figure><p>At this point, I was puzzled. This behavior looked like a classic misconfiguration. What was causing this inconsistency? To investigate further, I decided to go directly to the target website. Until now, I had only been interacting with a subdomain that was part of the application I was testing on HackerOne.</p><p>I logged in and created a new application to see how the system worked from the inside…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/710/1*e6Jw9ylMIASY5DZE2d9VbA.png" /></figure><p>analysing those settings and playing with them I discovered that if you are setting you application to be public, the tables would in return become public accessible. And this is only the case if you are using the tables offered by the target it self, if you are creating your own tables, they don’t comply to these privacy settings and they are private by default.</p><p>Great now that I now have this information, how can we actually know the amount of affected websites? I started analysing javascript but found nothing so I started fuzzing on the firestore link we found intially through different combinations till I found the following</p><blockquote><a href="https://firestore.googleapis.com/v1/projects/glide-prod/databases/(default)/documents/app-logins/e6vKplbKmdQz3fjGVSBS?pageToken=AFTOeJwzJnzh8Az8o4ZgF1rJ72ySworCmXNSUZp7r_ZArx3xnmJm57Kx5VfOf4-IQkbHaS3HC3v9Ug5kB4UKxLj0I-ZoaX08BxDyoIlYC1PNX9311bSm5BmXBsox82B0gJed3us">https://firestore.googleapis.com/v1/projects/TargetName/databases/(default)/documents/app-logins</a></blockquote><p>This endpoint listed 20 applications and their subdomains on the platform, with a token to continue listing more applications.</p><p>I quickly wrote a simple script to collect all the subdomains, a process that took about 3–4 hours to complete, gathering data on 1.2 million subdomains.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2wTogw_Yi6WIbubb71s89A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cvntLntRH0uVbDA4LjhlRA.png" /><figcaption>1.2 million subdomains</figcaption></figure><p>I then created another script to automate the initial request that would give me the native table IDs, preparing them for the next step.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BlX-iRdZ4cKrxghmm7a_yA.png" /></figure><p>The rest of the script was focused on checking whether these tables were publicly accessible or not. Initially, I discovered thousands of exposed tables, but I considered the possibility that many of them were simply test applications or small apps being built. So, I added an additional condition: for a table to be eligible to be considered as leaked PII, it had to contain at least 30 entries.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e2eiqkSM-Kv-pFixnU0_zA.png" /></figure><p>With this criteria in place, I continued my enumeration and identified the websites affected by the vulnerability. Out of the 1.2 million subdomains I scanned, I found 720 hosting data leaks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i-hZO6H8c2hMbj4Umdtj5w.png" /><figcaption>720 subdomain hosting dataleaks out of 1.2 million</figcaption></figure><p>Of those, only one program had a bug bounty program, which was the one that had initially drawn me into this process. For my efforts, I received an award for two high-severity vulnerabilities: one for the widespread exposure (since it affected more than one subdomain, it was flagged as a duplicate), and another for the same mistake being repeated once again, which I later reported again.</p><p>PII example -&gt;</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*u_LNvDGs_84Rm_IzzTDv0A.png" /><figcaption>hundreds of PII of a film production company</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wQnbipefPeb7X3Clu-m9FA.png" /></figure><p>Contacting most of these organizations didn’t yield any results, as many of them didn’t even respond. This highlights the urgent need for companies to establish clear policies for reporting security issues that affect their websites even tho I was lucky enough to get 4 digits bounty twice for this finding.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6101fcb3d5e0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Exposing hidden DOS techniques laying in plain sight.]]></title>
            <link>https://medium.com/@omarahmed_13016/exposing-hidden-dos-techniques-laying-in-plain-sight-29965ab1d53b?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/29965ab1d53b</guid>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[vulnerability]]></category>
            <category><![CDATA[bug-bounty-writeup]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[bug-bounty-tips]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Thu, 07 Nov 2024 16:41:09 GMT</pubDate>
            <atom:updated>2024-11-07T16:41:09.368Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*EAPocb2hiKMv_rCn.jpg" /></figure><p>Hello everyone! Since I always love giving back to the community, in this post, I’ll be sharing a few techniques I’ve leveraged to identify DoS vulnerabilities in bug bounty programs. There’s a common misconception around DoS and DDoS, especially as some programs mistakenly list ‘DoS’ as out of scope when they actually mean ‘DDoS.’</p><p>The distinction is important: <strong>DoS (Denial of Service)</strong> refers to a single source attempting to overwhelm a target system to make it unavailable for legitimate users, often by exploiting flaws that exhaust resources or cause the application to crash. <strong>DDoS (Distributed Denial of Service)</strong>, on the other hand, involves multiple sources — often a network of compromised devices (botnet) — coordinating to flood the target with traffic.</p><p>While DDoS attacks typically require substantial resources and are frequently categorized as out of scope in bug bounty programs due to the network-level impact and potential legal concerns, DoS vulnerabilities often arise from specific weaknesses in the application layer that can be exploited by a single user. Many programs consider these in-scope, as addressing these vulnerabilities is crucial for ensuring application stability and security.</p><p>So lets go forward exploring those techniques.</p><p>— — — — — —</p><p>1- missing checks on rendering images’ parameters ( height and length )</p><p>sometimes you will upload a profile picture or a picutre of a specific product, check the rendering page as it might be taking parameters responsible for the resolution of that image. In my case this led to a DOS on the customer’s website even so fast that this was fixed before traigging its report. ( led to closure as an info, they discovered it before they looked at my report, bb is a scam :P )</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3dcgna93rHXW6KWP-IvDqw.png" /></figure><p>2- Found a url/parameter that was taking a url and based on that value, the server would go fetch it for you based on that data you eneter? you might’ve tried getting SSRF and it didn’t work!</p><p>Now what you are going to do is simply go to <a href="https://ash-speed.hetzner.com/">https://ash-speed.hetzner.com/</a> and grep a test file ( around 1–10 GB) and if the server waits for that url to finish, you can flood the server with such requests that would take it done eventually :)</p><p>In my case this happened in a different way, there was a biling function and that billing function was taking a payment uuid and giving back its PDF uploaded to AWS. Neither IDORs nor SSRF worked. but I noticed that it was taking a list instead of a url, thoughts? send a list containing 1000 uuid! result? -&gt; time out after 60 seconds with a lag on the website.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DAbcrPv1qfdHAdMpmAM0kA.png" /></figure><p>few requests and the site was down 😮.</p><p>3- Identify services fetching data from external resources.</p><p>In my case I found that fetching products were done from a certain service that took strange parameters. I looked up where would those parameters get passed to, did some recon and stumbled upon a 3rd party documentation stating how to use those parameters, they are harmless by design. but I noticed something strange, the 3rd party website had a rate limit and the website I was testing was not since it wasn’t actually a sensetive action. what did I do? used FFUF with 1000 thread in order to exahust the 3rd party application’s rate limit and afterwards the site I was testing on went out of service as the 3rd party was not actually responding to it since it passed the rate limit. :)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2pHjJMdHNGCOeWhDQW2fRg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*8iGvYUzkIQgZsPvN0dkZAQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/602/1*AvwhpzjbBjQU-avIkuoVFg.png" /></figure><p>In conclusion, I want to highlight that you need to push your mind to think of invisible attack vectors that might be hidden in front of you. all of those functions were not vulnerable to common vulnerabilities ( idors/ ssrf / sqli ) but if you don’t set limits to your mind, you certainly get few nice bugs.</p><p>till next time!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=29965ab1d53b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[IIS welcome page to source code review to LFI!]]></title>
            <link>https://medium.com/@omarahmed_13016/iis-welcome-page-to-source-code-review-to-lfi-23ec581049f5?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/23ec581049f5</guid>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[hackerone]]></category>
            <category><![CDATA[application-security]]></category>
            <category><![CDATA[bug-bounty-tips]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Sat, 31 Aug 2024 15:57:45 GMT</pubDate>
            <atom:updated>2024-09-01T10:37:24.680Z</atom:updated>
            <content:encoded><![CDATA[<p>Hi, in this writeup I’ll walk you through how I managed to get a limited Local file disclousre (LFI) / Blind SSRF.</p><h3>TL;DR:</h3><p>found an IIS welcome page, enumerated directories&amp; files using IIS Short name scanner and FFUF, found an open source webchat software, source code review led to LFI and blind SSRF.</p><p>So, I had this subdomain that returned the IIS welcome page -&gt;</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/796/1*WwB0QCizvKJx0bHKBnqKCQ.png" /></figure><p>then I used <a href="https://github.com/irsdl/IIS-ShortName-Scanner">https://github.com/irsdl/IIS-ShortName-Scanner</a> before fuzzing and It showed that this target was vulnerable to <a href="https://github.com/irsdl/IIS-ShortName-Scanner/blob/master/presentation/Steelcon-2023-Beyond_Microsoft_IIS_Short_File_Name_Disclosure.pdf">IIS Short File Name Disclosure</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RwS3_vVQhXkxuMCJhUnUKA.png" /></figure><p>after finding those, use ffuf to try to find those files, I found that one of the folders is EVOLUTION! then I ran the tool again &amp; so on and so forth till I reached finally to this path</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bcwKrNiJV7faWRCbddCYEg.png" /></figure><p>but before going forward I thought about maybe this web chat is open source, is it? turns out it is!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d7Uu3Y6xsd-UJdRvH9aFNw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AeAvnIbih3dufUAdjzlg7w.png" /></figure><p>My first impression was if it’s created 10 years ago, they didn’t know what security means back then, right? ¯\_(ツ)_/¯</p><p>digging through the source code I found the following snippet:</p><p><a href="https://github.com/eStream/eStreamChat/blob/2417a12c0999609e3feedbf5281d07393b126c23/eStreamChat/Thumbnail.ashx.cs">https://github.com/eStream/eStreamChat/blob/2417a12c0999609e3feedbf5281d07393b126c23/eStreamChat/Thumbnail.ashx.cs</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QGhTwuqiRGxnTusm1MrD6A.png" /></figure><p>Looking at the source which is in this case the img parameter which took a parameter “img” and based on its value it interpreted it in 2 different ways, if it found UserFiles in the value of that parameter, it would try to query it from the server directly (LFI) which can be trivially bypassed using path traversal, UserFiles/../wheredoyouwannago</p><p>and in the second case it would do a request on the server’s behalf to the url you specify :). However, through analysis we can find that the response isn’t directly returned but in fact it goes into ResizeImage function which sadly wouldn’t accept the type of data we are trying to return unlessi it’s an image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O4yYC07q7d-pPtkbysJSLw.png" /></figure><p>so we could exfilterate images on the server or perform Blind SSRF by simply hitting that endpoint</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IAZy7FBYrhHUIGNt17ix0g.png" /></figure><p>when trying the LFI on our server, it turns out it didn’t need to specify the UserData path:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F0dUnqO1y5RgJf0VowXr4Q.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*x6cidjXMH3KwV_PwQRkDQg.png" /></figure><p>I hope this writeup added some fresh ways of thinking &amp; valuable information to you!</p><p>till we meet again :)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=23ec581049f5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The great chain of duplicates. Utilizing Javascript and Logic in the bug bounty field.]]></title>
            <link>https://medium.com/@omarahmed_13016/the-great-chain-of-duplicates-utilizing-javascript-and-logic-in-the-bug-bounty-field-4758b6977fb3?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/4758b6977fb3</guid>
            <category><![CDATA[bug-bounty-tips]]></category>
            <category><![CDATA[hackerone]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Sat, 04 Nov 2023 21:17:55 GMT</pubDate>
            <atom:updated>2023-11-04T21:17:55.267Z</atom:updated>
            <content:encoded><![CDATA[<p>Hi, in this blog I’ll walk you through how analysing javascript files and chaining bugs made me so close to get some bounties. This might be a long blog but i’ll try to split it into several points/ reports. I’ll also share my methodolgy in analysing minified JS files.</p><p>— —</p><p>let’s start with setting up the environment. So first thing I do in a website is simply browsing, keep poking around and touching functions, check wappalyzer to know which stack it’s using while burp is already in background. after a bit, I take some random paths from burp history that were interacting with the website API to get the js files that might be actually interesting</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hfiETmLVFxp4jDK4vxiSJw.png" /></figure><p>Certainly looking at this</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/772/1*k1yYBwbrJ8nKcwuJHI0MdQ.png" /></figure><p>might feel overwhelming so we try a couple of things.</p><p>first we try to add .map to the end of the js file url to see if we get anything back and last we look in the file for .map as it might be embedded in some comments. If we manage to get the .map files our life will be easier but if we didn’t that’s not the end of the road :) we will go to <a href="https://beautifier.io/">https://beautifier.io/</a> and insert our ugly js file and this site will try to beautify it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*brbq5VtfH6oDt4NgzD5QTg.png" /><figcaption>before</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QB2sYpnfepHTB6bYVAh3WA.png" /><figcaption>after</figcaption></figure><p>then we take the result in put in Visual studio code and set language to typescript and yes, this is a lot easier to read and take notes rather than tracking it on Burp suit.</p><p>— — — — —</p><p>Great! with this in hand let’s see I was able to dumb 500K PII including emails/names/phonenumbers and addresses .. it’ll be 3 parts</p><p>The first part is about understanding the website functionalities, second and third are bugs that i chained with previous knowledge based on part 1</p><p>— — —</p><p>So, visiting my target I found a request appeared in my burp and at first glance i thought it was responsible for searching</p><p>It was a post request with the following body</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/764/1*76b61b8gtOUd0MOVtOL3Sg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/453/1*seHybvkJy0ou_1kZJEQZYg.png" /></figure><p>this is some of the parameters it took. Result was some made activity by me. Analysing them was pretty straight forward. resultDefn was needed for us to determine how the result attributes should be like and it also took attributes parameter and it was responsible for taking the search parameters. One important thing also you need to look at is entityName. but why? after inspecting some other requests to that search function I found something odd, some requests to the same endpoint had different entity names, different resultDefn and certainly different attributes. So my first reaction to this — Looking at different names with ( activityUser, activities, products ) — I guessed this model is created to talk to the database directly, i was enlightened to be honest and I said to my self I guess this is a very beautiful feature, it’s fuzzing time :O so I kept fuzzing for different entity names or lets say ( Table names ) but came to the sad truth with tons of 404. So what other place is great for finding different entity names? JAVASCRIPT! by searching for an already in hand entity name</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/699/1*g-Ccg7WAcZuD-bOXd9XpRA.png" /></figure><p>Okay now we know entity names exist in javascript but is there’s a sufficent way to extract all of them ? in my case yes but it’s not always the case, I simply read the above code and found that it first checks if a user is authorized to access the ( Activity user save ) model then checks if r = activity user so I said ok let’s look for r=” in the file we had and the result was astonishing -&gt;</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/244/1*-2EW2tZUpPzoGDUy07kKtA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/187/1*bXYDRWup5An2kN1EvtKUBQ.png" /></figure><p>so after we harvested a large set of entity names, we needed to know which entities were valid so we ran burp intruder with the list of entity names we gathered.</p><p>Results were divided into 3 categories</p><p>1- 404 in case an entity name is wrong</p><p>2- 403 if we are not allowed to interact with such entity, ( so there’s an actual protection layer)</p><p>3- 200 but no data was found</p><p>So I first decided that we know how to eliminate the third category so how should we tweak our search so it always yields our data at least?</p><p>I noticed this depended on attributes and also Resultattribute params that we menitoned above. So I first omitted the resultAttribute in a valid request that already yielded data before and the result was a so huge, the normal request returned firstname and name of the owner, removing the result param resulted in showing every single piece of data related to this data of ours including admins changed it/updated it etc. So okay now we try this in category 3 we already got but still we recieved no data was found, so I tried doing the same thing with attributes parameter, at first I tried to set SEARCH_Text to * in order to match all data</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/356/1*wuA7xMKCv5yBGUb1yHpMeg.png" /></figure><p>still no data was found. so I tried removing the whole attributes param but I was introduced to a new error that stated that I needed to have a valid attribute parameter. A random thought came to me, what about sending empty {} in the attributes param?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/232/1*Z4hAQqOPepxIm-3G6SDHQg.png" /></figure><p>Worked as CHARM, so now our categories got deduced to be either we have access to a specific entity or not. with that in mind I went to try all entites I gathered, found one and managed to get a valid POC to leak firstName and last name but we are after a longer chain aren’t we? So now we know the structure of how requests are sent to the api that deals with database.</p><p>— — — -</p><p>Part 2: ALL YOUR S3 FILES ARE MINE, BUT PLEASE TELL ME WHAT ARE YOUR FILE NAMES</p><p>inspecting and analysing JS lead me to this great snippet</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/735/1*Y1AAa9q1Ug5JNI8fcrL2pg.png" /></figure><p>So i wondered if this is what it looks to be :O</p><p>Say no more</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OKU9FGcUkx8b2C85fT3lAA.png" /></figure><p>So we are able to create signed urls for any files on a bucket however I wasn’t able to actually list those files :/</p><p>pathtraversal is not working here but we don’t need it as path parameter already is starting from root directory ( I browsed the app and found that uploading any file gets a generated filename which is filename_timestamp so it was infeasiable to target all users )</p><p>more info about signed urls here -&gt; <a href="https://www.youtube.com/watch?v=MBQJJ3jfJ8k&amp;ab_channel=BugBountyReportsExplained">https://www.youtube.com/watch?v=MBQJJ3jfJ8k&amp;ab_channel=BugBountyReportsExplained</a></p><p>So now we got 2 pieces :)</p><p>— — —</p><p>PART 3, We might need Backups but what might go wrong?</p><p>Here comes the missing plot</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/764/1*dGp1QYX1cRhodq6GwuUmag.png" /></figure><p>remember the entityName? So what’s the difference? here I found out that instead of outputting the data we use in searching, we export it based on the exportFormat. so now we perform the same thing we made when we were analysing the search function</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WGDiFZR4ThDnKWC_6ElCRg.png" /></figure><p>Oh? Why are we getting results not found even tho if we send it in search function it works? what did we do wrong? back to basics, why did we send entityName and attributes only? because we don’t know what attributes might be and because we want to retrieve all the result data not a specific piece of the result. hmmm that’s alright, could it be the case that because we are not entrying the format of our result we are not recieveing any result? yes that was the case, fixing the issue was by simply sending anything in the result and backend neglected it anyways and sent the whole result D’: So I simply went to the original request for the search function</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/781/1*QORzjtZQv7MfPuui--0HCw.png" /></figure><p>that resulted a third category having 200 but no result and tried it here and guess what?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pFaRipqdwCQVsIUbg4i7gg.png" /></figure><p>Using bug 2 that we already had</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I0NbuKUI7zO-QaqhXscL2w.png" /></figure><p>GREAT! we managed to export our data can but how can we use this against other users? now the haunt begins again with a goal of finding some parameters to put in the attribute as in the original search for example it had this but wasn’t vulnerable to idor</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/390/1*f6CcHz-t-8ScCeRuwApYtA.png" /></figure><p>but in this export url, this was vulnerable and gave us different user activity -&gt;</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/1*FrQZRlkbnFbFq72M6fAw2g.png" /></figure><p>but as you see, there’s no confidential data :( ?</p><p>Now let’s go back for category 2 entity names that gave us access denied</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*50Z9QNtjyvQ3kvYV6e_h9w.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EefK2K5AlIvPVjOJIZzPTw.png" /></figure><p>RIP MR developer :(</p><p>every entity that gave me access denied was here giving me much data and joy</p><p>here’s for instance logs of loggin into account ( MAC — IP — OS etc)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/873/1*fSvQcv_qlIuyjZEQXK2V8w.png" /></figure><p>So now we got a very nice chain of bugs that resulted in leakage of around 500K records of customers and much more info from all entites in the back end. I hope this enriches your knowledge and let me know what you think about it :D</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4758b6977fb3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Graphql path traversal lead to disclosure of PII.]]></title>
            <link>https://medium.com/@omarahmed_13016/graphql-path-traversal-lead-to-disclosure-of-pii-38597b8446d4?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/38597b8446d4</guid>
            <category><![CDATA[bug-bounty-tips]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[hackerone]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Fri, 03 Nov 2023 12:57:24 GMT</pubDate>
            <atom:updated>2023-11-03T12:58:38.554Z</atom:updated>
            <content:encoded><![CDATA[<p>Hi,<br>I’m here to share a path traversal bug I found on a private program on h1 which allowed me to access restricted internal API with unrestricted privileges.</p><h3>TL;DR:</h3><p>found a graphql endpoint that takes ID as a parameter and using path traversal I was able to find other confidential data of other people.</p><p>Now i’ll be sharing how I stumbled across this request.</p><p>At first I was exploring the application and found that it uses both GraphQL/rest api endpoints.</p><p>after going through some of the requests, i found a request that was responsible for editing customers’ address using customer ID so I tried inserting ../ after the id and I recieved the following error message</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bxaqTyRdjt4tHUwFk9luBg.png" /></figure><p>So as I said in my tweet that I actually tried to simply insert other user’s address id and I received 404 not found. how this vulnerability occur and how is the payload constructed ? internal api queries the recipetent data based on the recipientID but how is it constructed?</p><p>When I tried to simply put the hash of a victim, I recieved a 404 not found with the following path</p><p><a href="http://recipient-service/v1/customer/e2efd53c83740ba367a52c1d94810e7d/recipient/victimrecipienthash">http://recipient-service/v1/customer/e2efd53c83740ba367a52c1d94810e7d/recipient/victimrecipienthash</a></p><p>Then I thought if the website don’t trim the ../ I might be able to use path traversal to query other internal paths/ or go to other customers data based on customer hash. So we inserted test/../../../ and the endpoint responeded with path not found with the following path</p><p><a href="http://recipient-service/v2/customer/">http://recipient-service/v2/customer/</a> So we know that we can manipulate the URL and hopefully to get a neat way to hit other internal endpoints:)</p><p>afterwards i got a hash of my victim account + hash of reciptient in the victim account and inserted it in this way</p><p><a href="http://recipient-service/v1/customer/victimaccount/recipient/victimrecipienthash">http://recipient-service/v1/customer/victimaccount/recipient/victimrecipienthash</a> with the followin payload test/../../../victimaccount/recipient/victimrecipienthash</p><p>And i was able to query the data/modify it of that recipient. I 100% believe that what’s important that the ability for the attacker to go to any internal path in that service meaning it’s an ssrf but It’s limited to the reciptent micro service they have. I tried to esclate it by trying to fuzz for different internal urls and I was able to send POST requests to other similar endpoints and I shared this info with team and they accepted it as a high finding based on how this would actually affect customers’ data.</p><p>Here’s a diagram of how we can use this attack in attacking different users and trying to access confidential data on this service</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/656/1*bVmndMZt1vaEqA-Ya14vIA.png" /></figure><p>I hope this added some knowledge to you, thanks for reading and wait for more writeups soon.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=38597b8446d4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ِAccount takeover hidden in Javascript files plus some extra work? my type.]]></title>
            <link>https://medium.com/@omarahmed_13016/%D9%90account-takeover-hidden-in-javascript-files-plus-some-extra-work-my-type-a2eb8208f049?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/a2eb8208f049</guid>
            <category><![CDATA[hackerone]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[bugbounty-writeup]]></category>
            <category><![CDATA[bug-bounty-tips]]></category>
            <category><![CDATA[penetration-testing]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Tue, 29 Aug 2023 09:59:23 GMT</pubDate>
            <atom:updated>2023-08-29T09:59:23.325Z</atom:updated>
            <content:encoded><![CDATA[<p>Hey guys, after my latest account takeover I decided to collaborate with one of my friends on the same program I got the first ATO on. Was it useful? will see.</p><p>— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —</p><h3>TL;DR:</h3><p>I found an endpoint that leaks the id of the user then through other endpoint I was able to grab the reset password token that enabled me to takeover any user’s account using Email only.</p><p>— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —</p><p>Now, through these words i’ll be describing the methodology or how we think of certain things rather than just giving you the proof of concept</p><p>So, Lets start!</p><p>At first, after having the other ATO my friend decided to collab with me, he took a second look on the js files and grabbed all the endpoints and it was for us to try every and each endpoint manually and they were about 200 endpoint :D. ( Why did we do such thing manually? most of the endpoints needed specific parameters and to check whether its a GET or POST request so it needed lots of js reading!)</p><p>And while checking the endpoints that we collected manually, we found this in the js files</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/788/1*_6b1egU0i3M4lXbBqls6jQ.png" /></figure><p>It was an endpoint that’s used to define the type of advertisements to be sent to the user on his email and it’s only used for the accounts that were registered with SSO. However, the endpoint itself didn’t validate whether it’s sso or normal email and if you send a request with any email you have it would simply response to you with the location the user selected on that email so it was actually leaking PII + we were modifying the keywords of that person to customize the type of ads that were sent to him.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QpGme8mK0uY0bh1Z-Tf3Cg.png" /></figure><p>and below that it included some internal ips ;). it was 3 AM so I reported it ( It was closed as infomrative, don’t ask me why ) and sent my finding to my friend so he can see if we can esclate it further.</p><p>On the morning we went out for endpoints that required that “id” as we had an idea in mind that this looks like the unique id for the user and we can find idors with it or even worse .. ATO? so up till now is a flow (Email -&gt; unique uuid )</p><p>we used gau, waybackurls and crawling the website looking for more endpoints that are using this ID and one endpoint appeared from nowhere</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/685/1*J6W6utHMBHsNdeE51Mrufw.png" /><figcaption>that was the endpoint we found</figcaption></figure><p>as we can see it took user_id and when we open this endpoint, we will see that this actually sent us an advertisement to our email so we thought of email bombing as we can use any email on the website and send it many advertisements mails leading to unsubscriping users from it or, we can actually unsubscribe the user !</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LJKWJDb-fB3PSU7VfvJ8Tw.png" /></figure><p>At this point we kept searching for other endpoints while we sent the above link to sqlmap to check if id is vulnerable for sql injection, however my friend checked what if we actually set id=2 ?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/513/1*jer_kAA6k7_2XfAVxEy1fg.png" /></figure><p>At this moment he was shocked but when he tried to press on the reset password word, it told him token is invalid. However, everything made sense now.</p><p>emailPreview endpoint was meant for developing/testing and it was simply testing the functionality of sending emails but forgotten there so I said, if this is testing emails sent to mail, shouldn’t we send a mail first before we try to use this endpoint? and here was the last piece of the puzzle</p><p>Go to login page -&gt; press reset password -&gt; now enter victim email -&gt; go to our first vulnerable request that gave us user_id -&gt; take that user_id and go to the emailPreview endpoint and</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/293/1*VmmlHPghHACBhrMJCVQVKA.png" /><figcaption>bingo</figcaption></figure><p>takeaways</p><p>if you find user uuid leaked anywhere, go fuzz for endpoints using it and use your burp search for easing the process.</p><p>Thanks for reading and wait for more writeups.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a2eb8208f049" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Account takeover hidden in Javascript files.]]></title>
            <link>https://medium.com/@omarahmed_13016/account-takeover-hidden-in-javascript-files-3a7e3773304?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/3a7e3773304</guid>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[bug-bounty-tips]]></category>
            <category><![CDATA[hackerone]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Tue, 04 Jul 2023 17:23:16 GMT</pubDate>
            <atom:updated>2023-07-04T17:26:20.150Z</atom:updated>
            <content:encoded><![CDATA[<p>Hey guys,<br>I’m here to share a broken access control bug I found on a private program on h1 which enabled me to take over any account with knowing its email only.</p><h3>Methodology :</h3><p>I noted down some things which I saw in the website, how it register users and login them in and etc. Register flow was Simple OTP that you receive on the email you registered with. So at this point, without having many functionalities to poke with, I decided to go to javascript files to analyse the source code.</p><p>I saw this method on the code</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/926/1*eJZSDV0qtlLcxLwALOWbKg.png" /></figure><p>The thing that caught my mind is, there’s no mention for this method anywhere else so I guessed that this is either under development or from past login ways. Anyway, tracking the sendOTPCodeEmail, it was valid and I couldn’t find something interesting in it. However, saveJobseekerPasswordInCache function was so INTERESTING!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/767/1*DQcrSqyJeFjFZg0b450x9Q.png" /></figure><p>So this endpoint took 3 parameters. So I thought there’s no actuall checks there because the website actually thinks that the user still registering on the website so I decided to try to find out how’s the request sent by replicating the flow that happens in doFetchRequest and it was a simple one. I successfully entered my other mail I use for testing and a password that I created but nothing changed and I couldn’t just figure out what did I lack or maybe the actual code is redundant as I thought. But here comes the life saver, saveJobseekerPasswordInDatabase!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/772/1*ero_dVpQk7VfleCfiBY6Ew.png" /></figure><p>seems our scenario has got to an end! Now I tried to send the email I used in the first request and I received a very nice ( PASSWORD HAS CHANGED SUCESSFULLY ) and I was able to login to that account with the new password I put.</p><p>So it needs the two requests to be fired with sequence as we obviously need to store the password in the cache of the app — in this case we can consider it as a different table in the database that has email and password — then it actually store the password we entered the password we sent on the users’ table.</p><p>Takeaways:</p><p>1- Always try to read javascript files and methods, not only to search for secrets but it will reveal the application for u.</p><p>2- Enjoy finding bugs rather than what would a program rate you because this bug was out of scope :D ! I didn’t notice that Authentication testing is out of scope but I enjoyed reporting my first ATO.</p><p>Thanks for reading!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3a7e3773304" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[IDOR, unpin posts for fun.]]></title>
            <link>https://medium.com/@omarahmed_13016/idor-unpin-posts-for-fun-18f628eaef24?source=rss-c71f2cf0306f------2</link>
            <guid isPermaLink="false">https://medium.com/p/18f628eaef24</guid>
            <category><![CDATA[bug-bounty-tips]]></category>
            <category><![CDATA[bug-bounty-writeup]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <dc:creator><![CDATA[Omar Ahmed]]></dc:creator>
            <pubDate>Tue, 13 Jun 2023 18:06:49 GMT</pubDate>
            <atom:updated>2023-06-13T18:06:49.507Z</atom:updated>
            <content:encoded><![CDATA[<p>Hey guys,<br>I’m here to share my recent IDOR on LinkedIn bug bounty program on h1 which enabled me to unpin any pages/companies’ posts without any permissions.</p><h3>Prerequisites</h3><p>What is IDOR?<br><em>Insecure direct object references (IDOR) are a type of access control vulnerability that arises when an application uses user-supplied input to access objects directly.</em></p><p>So, I was hunting on LinkedIn for more than 3 weeks but most of the findings were actually closed as duplicates and info so I knew that easy peasy IDORs aren’t the way we needed to find something that simple browsing with AUTHORIZE plugin wouldn’t find ;)</p><p>I stumbled on a request that seemed to be odd as it gave me an odd response code</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mAk-EKpdui0NJ8MPZ2a4RQ.png" /></figure><p>So the request is simply taking company ID in the URL and an order to delete that pinned post, having this weird response code I decided to go deeper. So I started looking for what is that version tag parameter and how can I get it.</p><p>I made another company page for the attacker and found out that the version tag exists on it, just a different sequence</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0FevVY0SFkDWWTQNQeVxKw.png" /><figcaption>version tag for attacker company</figcaption></figure><p>and when I tried that version tag it actually gave me the same error code. So I thought like, this must be belonging to each page and there’s no way to get the version tag if we can’t access that URL above that gives us the version tag because .. wait, we can actually try that! And YES when I send a request to <a href="https://www.linkedin.com/voyager/api/voyagerOrganizationDashCompanies/90912089?decorationId=com.linkedin.voyager.dash.deco.organization.AdminCompany-67">https://www.linkedin.com/voyager/api/voyagerOrganizationDashCompanies/ID-OF-Company?decorationId=com.linkedin.voyager.dash.deco.organization.AdminCompany-67</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GIsmqWl2Relc8XXyQTqb5w.png" /><figcaption>version tag of victim company</figcaption></figure><p>it gives you the version tag that you would use in such case! and when I tried to unpin the post again, 200 was there saying: you say I obey :D</p><p>thus, I had the ability to unpin all company posts because company IDs are in a numerical order.</p><p>Bug was triaged shortly after reported and was considered LOW due to the limited business impact, they said, but It is a responsive team and I recommend you to go deep dive on this application if you want to sharpen your skills in finding IDORS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=18f628eaef24" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>