<?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 Issra Akrout  on Medium]]></title>
        <description><![CDATA[Stories by Issra Akrout  on Medium]]></description>
        <link>https://medium.com/@issra.akrout?source=rss-235424ee9931------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*obgkNVBppcDa_ygAsYe8aQ.jpeg</url>
            <title>Stories by Issra Akrout  on Medium</title>
            <link>https://medium.com/@issra.akrout?source=rss-235424ee9931------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 08:51:28 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@issra.akrout/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[Building an AI-Powered Cyber Defence System with Reinforcement Learning]]></title>
            <link>https://medium.com/@issra.akrout/building-an-ai-powered-cyber-defence-system-with-reinforcement-learning-87bf4baf5ae5?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/87bf4baf5ae5</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[autonomous-cyber-defence]]></category>
            <category><![CDATA[multi-agent-systems]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[reinforcement-learning]]></category>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Sat, 21 Nov 2026 19:07:54 GMT</pubDate>
            <atom:updated>2026-11-21T19:07:54.460Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Designing a simulated network environment and training a PPO agent to learn defensive decisions.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/760/1*oIOciiMAzFgYvDh3cEJXfA.png" /></figure><p>Cybersecurity is often described as a race between attackers and defenders. As networks become more dynamic, relying exclusively on static rules and predefined responses can make defence difficult. This raises an interesting question: <strong>could an AI agent learn how to make defensive decisions by interacting with a simulated network?</strong></p><p>To explore this question, I built a Python-based cyber-defence simulation and trained a Reinforcement Learning agent using Proximal Policy Optimization (PPO).</p><p>This project was a hands-on exploration of the intersection between <strong>Artificial Intelligence, Reinforcement Learning, and Cybersecurity</strong>. In this article, I explain the environment I designed, the decisions available to the agent, the reward function, and how I structured the training and evaluation process.</p><blockquote><strong>Note on reproducibility:</strong> The GitHub repository accompanying this article contains a reconstructed implementation of the project, prepared to make the approach and experiments reproducible. The code can be used to retrain the PPO agent and generate evaluation results.</blockquote><h3>1. The idea: learning to defend instead of following only fixed rules</h3><p>A conventional defensive workflow often relies on alerts, signatures, thresholds, and analyst-defined playbooks. These are valuable, but they do not necessarily provide an adaptive decision-making strategy.</p><p>Reinforcement Learning offers a different perspective. An agent interacts with an environment, observes the current state, selects an action, and receives a reward that reflects the outcome. Over repeated interactions, it learns a policy intended to maximize cumulative reward.</p><p>For a simplified cyber-defence scenario, the agent might need to decide whether to:</p><ul><li>Continue monitoring suspicious activity.</li><li>Block a suspicious connection.</li><li>Isolate a potentially compromised host.</li><li>Investigate an alert to obtain more confidence.</li></ul><p>These decisions involve trade-offs. Blocking too aggressively can interrupt legitimate traffic. Waiting too long can allow compromise to increase. Isolating a critical host may contain an attack but reduce service availability.</p><p>The goal of the simulation was to represent these trade-offs in a controlled environment.</p><h3>2. Designing the simulated environment</h3><p>I implemented a custom environment using Python and the Gymnasium interface. The environment models an abstract network state rather than real packets or a full enterprise topology.</p><p>The observation is represented by seven normalized values:</p><p>FeatureWhat it representsAttack statusWhether an attack is activeCompromise levelEstimated level of compromiseSuspicious trafficSuspicion score associated with trafficService availabilityRemaining service availabilityAlert confidenceConfidence in the current alertHost criticalityImportance of the affected hostPrevious disruptionWhether the previous action disrupted service</p><p>This compact representation gives the agent enough information to make decisions while keeping the initial problem manageable.</p><h3>Environment architecture</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AKC6yFI4UZS_c95dqj6Dgw.png" /></figure><p>The environment is stochastic: attack pressure and some state transitions include randomness. This is important because a policy should not simply memorize one fixed sequence of events.</p><h3>3. Defining the action space</h3><p>The agent has four discrete actions.</p><p><strong>Monitor</strong> allows the agent to continue observing the environment. It is a low-disruption action, but it may allow an active attack to progress.</p><p><strong>Block</strong> attempts to stop suspicious traffic. When an attack is active, this can contain it. When no attack is present, an unnecessary block is penalized.</p><p><strong>Isolate</strong> attempts to contain a compromised host. It can be effective against an active attack, but it also reduces service availability. The cost is higher for more critical hosts.</p><p><strong>Investigate</strong> increases alert confidence. It can help the agent gather evidence before taking a more disruptive action.</p><p>The action space deliberately includes both preventive and disruptive choices. This makes the reward design an important part of the experiment.</p><h3>4. Reward engineering: security is not the only objective</h3><p>One of the most interesting parts of a Reinforcement Learning project is designing the reward function.</p><p>If the agent receives a reward only for stopping attacks, it may learn to isolate everything. That could improve containment while making the network unusable.</p><p>For this reason, the reward combines several considerations:</p><ul><li>Reward successful containment.</li><li>Penalize unnecessary blocking.</li><li>Penalize unnecessary isolation.</li><li>Penalize residual compromise.</li><li>Reward service availability.</li><li>Apply a small time cost.</li></ul><p>The objective is not simply to maximize the number of blocked attacks. It is to encourage a balance between security and operational continuity.</p><p>This also introduces a limitation: a reward function is only an approximation of the real objective. Different weights can produce different defensive behaviours, so reward sensitivity should be evaluated rather than assumed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xu_Cf1haQiEcmAlXQBd_XA.png" /></figure><h3>5. Training the agent with PPO</h3><p>I used Proximal Policy Optimization as the learning algorithm.</p><p>PPO is a policy-gradient method that constrains policy updates to avoid excessively large changes. It is widely used as a practical baseline because it is relatively stable and straightforward to experiment with.</p><p>The reference implementation uses Stable-Baselines3 with an MLP policy. The model interacts with the custom Gymnasium environment and learns from the rewards returned after each action.</p><p>The training configuration includes:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X4BGNUc_wNxNs5j1Or485w.png" /></figure><p>These are baseline hyperparameters, not a claim that they are optimal. The training budget and random seed are configurable from the command line.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Umk-PEjE7T2YPSykRiwwhQ.png" /></figure><h3>6. Evaluation: compare behaviour, not just reward</h3><p>A trained agent should not be judged by a single reward value.</p><p>The repository includes an evaluation script that runs the PPO policy and a simple heuristic baseline over multiple episodes. It exports episode-level data so that performance can be inspected rather than reduced to one number.</p><p>The evaluation metrics include:</p><ul><li>Mean episode return.</li><li>Attack containment rate.</li><li>Final compromise level.</li><li>Service availability.</li><li>Episode length.</li></ul><p>The evaluation script also creates a CSV file and a JSON summary. Plotting utilities generate figures for episode returns, compromise, and availability.</p><p><strong>Important:</strong> no numerical performance claim is included here because the reconstructed repository has not been run in your environment yet. After running training and evaluation, insert your actual values and plots in this section. Do not report a performance improvement unless the measured results support it.</p><h3>7. What this experiment taught me</h3><p>This project highlights several important lessons about applying Reinforcement Learning to cybersecurity.</p><p><strong>Reward design changes the agent’s behaviour.</strong> A model can optimize the reward it receives without necessarily learning the defensive behaviour we intended. The reward must represent the trade-offs carefully.</p><p><strong>Availability matters.</strong> A defensive action that stops an attack but disables a critical service is not automatically a good decision.</p><p><strong>Stochastic environments require repeated evaluation.</strong> One successful episode is not enough to demonstrate robust performance. Results should be measured across multiple seeds and scenarios.</p><p><strong>A simulation is a starting point, not a real network.</strong> The environment provides a controlled way to study decision-making, but it does not reproduce all the complexity of real network telemetry, attacker adaptation, or operational constraints.</p><h3>8. From a single PPO agent to collaborative cyber defence</h3><p>The current reference implementation focuses on a single defensive agent. A natural next step is to extend the environment to multiple security zones and specialized agents.</p><p>For example, one agent could monitor network traffic, another could observe endpoint behaviour, and a third could coordinate incident response.</p><p>This leads to Multi-Agent Reinforcement Learning (MARL). One algorithm worth investigating is Multi-Agent PPO (MAPPO), commonly studied under the centralized-training/decentralized-execution paradigm.</p><p>The challenge is no longer only learning a good individual policy. It is learning how agents can coordinate, share information, and avoid conflicting defensive actions.</p><p>That is a promising direction for extending this work toward more complex autonomous cyber-defence research.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NnuUNlTDNwZ6aSyUJFU04A.jpeg" /></figure><h3>9. Reproducibility and source code</h3><p>I prepared a reproducible repository containing:</p><ul><li>The custom Gymnasium environment.</li><li>PPO training code.</li><li>Evaluation against a heuristic baseline.</li><li>Metrics export and plotting scripts.</li><li>Unit tests.</li><li>Installation and execution instructions.</li></ul><p><strong>GitHub repository:</strong> <a href="https://github.com/issraakrout/ai-cyber-defence-ppo">https://github.com/issraakrout/ai-cyber-defence-ppo</a></p><p>Create this repository under your GitHub account and push the files before sharing the URL publicly.</p><p>To reproduce the experiment:</p><pre>git clone https://github.com/issraakrout/ai-cyber-defence-ppo.git<br>cd ai-cyber-defence-ppo<br>pip install -e .<br>python scripts/train.py --timesteps 100000 --seed 42<br>python scripts/evaluate.py --episodes 100<br>python scripts/plot_results.py<br>pytest</pre><h3>Conclusion</h3><p>Building this project helped me explore how Reinforcement Learning can be applied to a simplified cyber-defence problem.</p><p>The most interesting part was not simply training an agent. It was modelling the decisions a defender faces: when to observe, when to investigate, when to block, and when the cost of isolation may be too high.</p><p>PPO provides a useful starting point for this exploration, while multi-agent approaches open the door to more collaborative defence strategies.</p><p>This is one step in a broader journey into <strong>AI-driven cybersecurity, autonomous defence, and intelligent security systems</strong>.</p><p>The next challenge is to move from one agent making decisions to multiple agents learning how to defend a network together.</p><h3>About the author</h3><p><strong>Issra Akrout</strong> is a computer engineering student specializing in cybersecurity, with interests in AI security, security governance, risk and compliance, and autonomous cyber defence. She enjoys building practical projects at the intersection of cybersecurity and Artificial Intelligence.</p><ul><li>LinkedIn: <a href="https://www.linkedin.com/in/issra-akrout/">https://www.linkedin.com/in/issra-akrout/</a></li><li>GitHub: <a href="https://github.com/issraakrout">https://github.com/issraakrout</a></li><li>Medium: <a href="https://medium.com/@issra.akout">https://medium.com/@issra.akrout</a></li><li>TryHackMe: <a href="https://tryhackme.com/p/Issra.Akrout">https://tryhackme.com/p/Issra.Akrout</a></li></ul><p><em>#Artificial_Intelligence #Cybersecurity #Reinforcement_Learning #Python #Machine_Learning</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=87bf4baf5ae5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I built a personal cyber lab — here’s exactly how]]></title>
            <link>https://medium.com/@issra.akrout/i-built-a-personal-cyber-lab-heres-exactly-how-e204c95de010?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/e204c95de010</guid>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Tue, 24 Mar 2026 15:24:15 GMT</pubDate>
            <atom:updated>2026-03-24T15:24:15.906Z</atom:updated>
            <content:encoded><![CDATA[<h3>Setting up the battlefield</h3><p>Before launching any attack, I verified that all machines were on the same isolated network (cyber_lab NAT Network in VirtualBox)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/556/1*SB7mP1vsxLg5PhUcDZru0g.png" /><figcaption>kali configuration</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/749/1*XzdZyrFPtR9LynJQpJY4AQ.png" /><figcaption>windows 10 configuration</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/613/1*NkkAuEoLqCFOKM3pSVVUuQ.png" /><figcaption>metasploitable 3 configuration</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/351/1*DgqhCWoJpgX6x5xBifbd4w.png" /><figcaption>all we need</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/372/1*m3Qmkqw9eMiWdl8GuwlkoA.png" /><figcaption>Kali is now on 10.0.1.7 on the cyber_lab network</figcaption></figure><h3>Reconnaissance — Mapping the Attack Surface</h3><blockquote>Every attack starts with reconnaissance. Before touching anything, I used Nmap to silently map the entire lab network — discovering live hosts, open ports, running services, and operating system versions. This is exactly what an attacker does in the first minutes of a real intrusion.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/692/1*KFfcQFXP01IIn8mauSO-Ug.png" /><figcaption>Network discovery — 3 targets found on the isolated lab network</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g0_9gOqBDC6_OJyEZWl0tQ.png" /><figcaption>Metasploitable 3 exposes 7 vulnerable services — each one a potential entry point</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/678/1*ulWEfFyxZqDHWXHhL5KUhw.png" /></figure><p>Nmap revealed that the Windows 10 machine exposes SMB on port 445 the same protocol exploited by WannaCry in 2017. We also identified the exact OS version, which narrows down potential vulnerabilities significantly.</p><h3>Exploitation — Attacking the Targets</h3><blockquote><strong>SSH brute force</strong> is one of the most common attacks in the wild. Using Hydra with the rockyou.txt wordlist — containing 14 million real passwords leaked from data breaches — I attempted to crack the SSH credentials on Metasploitable 3.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YTtfu7WotGN6LJZdm352JA.png" /><figcaption>The command I ran — Hydra successfully cracked SSH credentials in seconds</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/828/1*vMFp3gN8qtxVoXjnMdujfw.png" /><figcaption>Hydra cracked SSH credentials — valid login found on Metasploitable 3</figcaption></figure><blockquote>Once the credentials were cracked, I used them to open a real SSH session into the target machine. This is the moment an attacker goes from observing a system to fully controlling it — able to read files, install malware, create backdoors, or pivot to other machines on the network</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/803/1*kzXEAbh5VL6JN1AyVzI-pw.png" /><figcaption>Successfully logged into Metasploitable 3 via SSH — full shell access achieved</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZOl1cwBhK7vm90NRiw3q6Q.png" /><figcaption>Post-exploitation recon — system users and kernel version exposed. A 2014 kernel on a production server is a critical finding</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*C_HaDTmstSPzIsTpNDKtaA.png" /><figcaption>Windows 10 initially blocked the SMB scan — its firewall was active</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/577/1*-pxprghn71yuZ7vLaAHGNA.png" /><figcaption>Apache web server exposes 4 web applications — each a potential attack vector</figcaption></figure><blockquote>SQL injection is consistently ranked in the OWASP Top 10 — the most critical web application vulnerabilities. By inserting malicious SQL code into a login form, I tricked the database into ignoring authentication entirely and returning all its records. No username, no password — just one crafted input and the entire payroll database was exposed.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/537/1*1r4U7Q0VAmBlROxC7NdaQw.png" /><figcaption>Payroll application login page — our SQL injection target</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/504/1*UCD_iBd_jyJWAeuK21iWuA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/879/1*0XVD7OHzMa_j5enUjBYoYA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/722/1*tGS6cri8Jy10EsyEv1k57A.png" /><figcaption>SQL injection dumped the entire payroll database — 14 employees, their full names and salaries exposed with a malicious input</figcaption></figure><p>With one line of SQL injected into a login form, I extracted the entire company payroll database — no password needed. This is exactly how real data breaches happen. The fix is simple: use prepared statements. The consequence of not fixing it: every employee’s salary is public</p><h3>Detection Stack</h3><h4>Suricata IDS</h4><p>Setting up the attack was the easy part. The real challenge was building the defense layer — and then watching it catch my own attacks live.</p><p>I installed Suricata IDS (Intrusion Detection System) directly on my Kali machine and pointed it at eth0 — the same interface carrying all my attack traffic. Then I ran the exact same Nmap scan from the recon phase.</p><p>knowing what your attacks look like from the other side</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UWDPSo5jKHZgx3QydnWXWg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/939/1*ZWBWNWdOmRWK6v_T4cQXnA.png" /></figure><p>The moment I started scanning — Suricata fired instantly. Every packet logged: timestamp, protocol, source IP, destination IP. 10.0.1.7 (me, the attacker) caught red-handed by my own defense system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Tr11ESRoN0sH1X4r_7AsZw.png" /><figcaption>Suricata alerts catching Nmap scan</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/932/1*74-nTslcjU-23QjkB5zqgg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vpYERHK6D3juLFYdJDD8uQ.png" /></figure><h3>Dashboard</h3><p><em>Note: I chose a custom Python dashboard over ELK stack <br>deliberately — building your own log parser shows deeper <br>understanding than configuring an existing tool. ELK is <br>a great production choice; Python is a better learning choice.</em></p><p>50 alerts. One scan. Every single probe logged. The dashboard auto-refreshes every 5 seconds — a live feed of my own attack traffic. TLS probes, HTTP anomalies, protocol violations — Suricata caught every technique my Nmap scan used.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fYnSNOzlZ9PfMGqEd_8Hpg.png" /><figcaption>INSERT siem-dashboard.png</figcaption></figure><p>This is what a SOC analyst sees when you’re attacking their network. Now I’ve seen it from both sides.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/696/1*Z4q1MEc1r26hgn38axnYYg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2LZoO85dD0Y8SPykv4jymQ.png" /><figcaption>siem-387-alerts.png</figcaption></figure><p><strong>Target 2 — Windows 10 (10.0.1.2)</strong></p><p>Windows 10 was included as a second target in the lab. The machine was reachable (ICMP ping successful), but SMB port 445 <br>refused connections — even after firewall adjustments.</p><p>This highlights a key reality in cybersecurity: Not all targets are vulnerable — and part of the job is identifying when systems are properly secured.</p><p>Even without exploitation, the scan results provided valuable insight <br>into the system’s exposure and defensive posture.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*94-U4OKCBrBe3o8I3DDuFQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VFyhOPQ-S63LMqiVjYP_wA.png" /><figcaption>automation-engine.png</figcaption></figure><p>This project was not about using tools — it was about understanding both sides of cybersecurity.</p><p><strong>I acted as the attacker:<br></strong>- Scanning the network<br>- Exploiting weak credentials<br>- Extracting sensitive data</p><p><strong>And I acted as the defender:<br></strong>- Deploying an IDS (Suricata)<br>- Monitoring attack traffic in real time<br>- Building a custom dashboard to visualize threats</p><p>The most important lesson: Security is not just about breaking systems — <br>it’s about understanding how attacks look from the inside.</p><p>And once you see both sides, you think differently.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e204c95de010" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Projects : generative ai for professionals]]></title>
            <link>https://medium.com/@issra.akrout/ptojects-generative-ai-for-professionals-e4f057f5c030?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/e4f057f5c030</guid>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[make]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[zapier]]></category>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Tue, 17 Mar 2026 14:45:21 GMT</pubDate>
            <atom:updated>2026-03-17T15:27:52.805Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Automating Professional Tasks-</strong><a href="https://eu1.make.com/public/shared-scenario/Njvn2tXBp02/weekly-progress-email">Make</a></h3><p>Sending weekly progress summary emails to a manager.</p><p>Every Friday, compiling updates from multiple sources (task tracker, notes, calendar) and writing a summary email is time-consuming. It follows the same structure every week and is an ideal candidate for automation.</p><p>Why this task?</p><p>• It happens on a fixed schedule (every Friday).</p><p>• The format is always the same (tasks completed, blockers, plans for next week).</p><p>• It pulls data from predictable sources (e.g., Google Sheets task list).</p><ul><li>Manual execution takes 20–30 minutes each week</li></ul><p>The automation workflow is broken down into the following sequential <strong>steps</strong>:</p><p>1. Trigger Every Friday at 4:00 PM (scheduled time trigger)</p><p>2. Fetch Data Read completed tasks from a Google Sheet</p><p>3. Generate Summary Format task data into a structured email body</p><p>4. Send Email Log Send the summary email via Gmail to the manager</p><p>5. Append a record to a ‘Sent Summaries’ sheet for tracking</p><p>After evaluating the main platforms, <strong>Make (formerly Integromat)</strong> was selected for this task. It offers a generous free tier, a visual drag-and-drop scenario builder, and excellent integrations with both Google Sheets and Gmail</p><p>The <strong>workflow </strong>was built with the following modules in <strong>Make</strong>:</p><p><strong>1. Trigger — Schedule module: </strong>Set to run every Friday at 4:00 PM. No external app connection needed for the trigger itself.</p><p><strong>2. Google Sheets — Search Rows: </strong>Connects to the task tracker spreadsheet. Filters rows where the ‘Status’ column equals ‘Done’ and ‘Week’ matches the current week number.</p><p><strong>3. Tools — Text Aggregator: </strong>Loops through the fetched rows and builds a formatted list of completed tasks as plain text.</p><p><strong>4. Gmail — Send an Email:</strong> Sends the assembled email to the manager’s address. The subject line includes the current date dynamically: e.g., ‘Weekly Update — March 14, 2026’. 5. Google Sheets — Add a Row: Logs the send event (timestamp + number of tasks) in a ‘Sent Log’ sheet for record-keeping</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DJP6Dkggg7hBgprrwIozrQ.png" /><figcaption>google sheet</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dAOebV_L3IOYQaQji5mxrQ.png" /><figcaption>Scenario MAKE</figcaption></figure><p>The scenario was tested using Make’s built-in ‘Run Once’ function, which executes the workflow immediately with live data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s1HnjAwRZCM-GZUZQCcPrQ.png" /><figcaption>test done with success</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/736/1*qJ4eXFSGoqJdbW_VhWhJLA.png" /><figcaption>email bien envoyé au manager</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/570/1*wzrWqP8lEejeeUYeAQH0Sg.png" /><figcaption>une nouvelle date ajouté dans google sheet</figcaption></figure><h3><strong>Your AI Email Assistant</strong></h3><p>This project demonstrates a fully functional AI Email Assistant built entirely with no-code tools. The assistant automatically detects incoming emails, uses AI to generate professional reply drafts, saves them to Gmail for review, and creates follow-up tasks in Asana — all without writing a single line of code.</p><p>Repetitive Task Chosen: <a href="https://zapier.com/shared/76a4af113a1d9d14162eb027f9736ae42fa47848"><strong>Following up on unanswered emails</strong></a></p><p>In a professional academic and administrative context, a large volume of emails require follow-up — such as requests for information, project coordination, and meeting confirmations. Many of these go unanswered, leading to delays and missed tasks.</p><p>The core workflow chosen for this assistant is: <strong>automatically generating a professional reply draft and logging a follow-up task</strong> every time an important email is received and labeled</p><ul><li>Platform chosen: Zapier Zapier was selected as the automation platform because it offers a free tier, supports thousands of app integrations, and provides a visual drag-and-drop interface that requires no coding. It connects Gmail, AI services, and Asana seamlessly in a single automated workflow called a ‘Zap’</li><li>Email provider: <strong>Gmail</strong></li></ul><p>Gmail was connected to Zapier using OAuth authentication. A dedicated Gmail label called<strong> ‘AI-Assistant’</strong> was created. The Zap is configured to trigger whenever an email receives this label, giving full manual control over which emails are processed.</p><p>Integration steps completed:</p><p>* Created label ‘AI-Assistant’ in Gmail settings</p><p>* Connected Gmail account to Zapier via secure OAuth</p><p>* Configured trigger: ‘New Labeled Email’ using the AI-Assistant label</p><p>* Tested trigger successfully — Zapier detected the labeled test email</p><ul><li>AI service used: <strong>AI by Zapier (powered by GPT-4o mini)</strong></li></ul><p>The built-in ‘AI by Zapier’ action was used — this is a free AI tool inside Zapier powered by OpenAI’s GPT-4o mini model. It was configured with a custom prompt to generate professional email replies based on the incoming email’s subject and body.</p><p>Prompt used:</p><p>“You are a professional email assistant. Read the following email and write a polite, concise, and professional reply. Email Subject: {Subject} — Email Body: {Body Plain}. Write only the reply email body. Do not include a subject line.”</p><p>The AI successfully generated a personalized, professional reply during testing, addressing the sender by name and responding appropriately to the email content.</p><h4>Triggers and Actions Setup</h4><ul><li>TRIGGER Gmail — New Labeled Email</li></ul><p>The Zap is initiated when a new email in Gmail is labeled with ‘AI-Assistant’. This gives the user control — only emails manually tagged (or auto-tagged by a Gmail filter) will trigger the workflow. The trigger captures the email’s subject, body, sender, and date.</p><ul><li>DRAFT ACTION AI by Zapier — Generate Reply Draft</li></ul><p>The email subject and body are passed to GPT-4o mini via the ‘AI by Zapier’ action using a custom prompt. The AI generates a complete professional reply. This reply is then saved as a Gmail Draft addressed to the original sender, with ‘Re:’ prepended to the subject line. The user reviews and sends the draft manually — nothing is sent automatically.</p><ul><li>FOLLOW-UP Asana Task with Due Date (3 days)</li></ul><p>A follow-up mechanism is built in through Asana: each triggered email creates a task with a due date set 3 days from the trigger date. This ensures the user is reminded to follow up if no response has been received. The task appears in the ‘New Emails’ section of the ‘Email Follow-ups’ project in Asana.</p><ul><li>TASK LOG Asana — Create Task in ‘Email Follow-ups’ Project</li></ul><p>Every triggered email is automatically logged as a task in Asana with:</p><p>* Task Name: ‘Follow up: {email subject}’</p><p>* Description: AI-generated summary of the email content</p><p>* Project: Email Follow-ups</p><p>* Section: New Emails</p><p>* Due Date: 3 days from email receipt</p><p>* Workspace: Email Assistant</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*80w5YE1jdd1VBr_er3poLw.png" /><figcaption>Triggers and actions</figcaption></figure><h4><strong>Test and Refine</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g2ZkbMpHavwWWQa48iNoMw.png" /><figcaption>mail de test</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8IBblaF4CUMf7hUEZtbDqA.png" /><figcaption>draft généré</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nFW-mVcenIP5kZxJmC-u-Q.png" /><figcaption>Nouvelle tache ajouté automatiquement dans asana</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e4f057f5c030" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mini Projet — Images and Prompts]]></title>
            <link>https://medium.com/@issra.akrout/mini-projet-images-and-prompts-c23e631bfe1d?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/c23e631bfe1d</guid>
            <category><![CDATA[prompt-engineering]]></category>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Wed, 11 Mar 2026 21:04:13 GMT</pubDate>
            <atom:updated>2026-03-11T21:04:13.803Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/685/1*HOvbsPtUScdp7oHna27RoQ.png" /></figure><h3>An isolated image of anything in the room</h3><p>An isolated photorealistic image of over-ear headphones placed neatly, centered in the frame.</p><p>Background: solid dark gray matte background.<br>Lighting: studio lighting highlighting the texture and curves.<br>Style: premium product photography, minimal aesthetic.<br>Composition: centered object with soft shadow.<br>Quality: ultra realistic, 4K.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IXQJDYvGE1MD24sjzRgvGA.png" /><figcaption>qwen</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/672/0*BZUKPjQdb5nRlC6t.jpg" /><figcaption>claude</figcaption></figure><h3>A character set featuring characters from your country or culture</h3><p>A character set featuring a historical Tunisian figure inspired by Mohamed Ali El Hammi, early 20th-century labor leader from Tunisia. <br>The character is an elderly North African man with a traditional Tunisian appearance, wearing early 1900s clothing (simple suit, traditional hat or chechia), dignified expression, intellectual look.</p><p>Style: vintage historical illustration.<br>Color palette: sepia and yellowed paper tones to resemble an old photograph from the 1920s.<br>Texture: aged paper, film grain, subtle scratches and faded ink.<br>Background: minimal antique parchment or archival paper texture.<br>Composition: character sheet showing multiple poses (front view, side profile, seated portrait, standing pose).<br>Lighting: soft warm lighting like an old studio portrait.<br>Mood: historical, respectful, documentary style.<br>Quality: highly detailed, realistic illustration, museum archive aesthetic, 4K.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5qzDoXGGY0G_lPLP50F0-g.png" /><figcaption>qwen</figcaption></figure><h3>A realistic sports image (Swimming)</h3><p>Ultra-realistic professional sports photography of a competitive swimmer performing freestyle in an Olympic swimming pool, powerful stroke, water splashing dynamically around the athlete, muscles defined and wet skin reflecting light, intense focus on the swimmer’s face with goggles and swim cap. Environment: Olympic competition pool with clear blue water and lane lines. Lighting: dramatic stadium lighting reflecting on water droplets and ripples. Camera: high-speed sports photography, 85mm lens, shallow depth of field, frozen motion. Composition: low water-level angle capturing the swimmer cutting through the water. Details: crystal clear water droplets, motion splash, realistic skin texture, detailed muscles, reflections on goggles. Style: ESPN / professional sports magazine photography. Quality: hyper-realistic, ultra-detailed, 8K resolution. Mood: powerful, energetic, athletic performance moment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Xpptpie8h3kgy023eKdsSQ.png" /><figcaption>qwen</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c23e631bfe1d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Week 21/30 : Network & Wi‑Fi Exploitation Techniques]]></title>
            <link>https://medium.com/@issra.akrout/week-21-30-network-wi-fi-exploitation-techniques-3791ba54da55?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/3791ba54da55</guid>
            <category><![CDATA[network]]></category>
            <category><![CDATA[wifi]]></category>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Sun, 22 Feb 2026 14:41:17 GMT</pubDate>
            <atom:updated>2026-02-22T14:41:17.020Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/812/1*dVx551LMhoUEv3dqFO4w5w.png" /></figure><h3>Bases Wi-Fi &amp; sécurité</h3><ul><li>WEP (Wired Equivalent Privacy ) : Ancien protocole, utilise un chiffrement facile à casser</li><li>WPA (Wi-Fi Protected Access) : Améliore la sécurité, mais reste vulnérable aujourd’hui</li><li>WPA2 : Utilise <strong>AES</strong> (Advanced Encryption Standard : chiffrement fort) pour sécuriser les données</li><li>WPA3 : Utilise une méthode d’authentification plus moderne (<strong>SAE</strong>)</li></ul><h3>Capture handshake et attaque dictionnaire WPA</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/998/1*l2YXM4mJy4_V9yRNjEafXA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/556/1*FxnbNihwIGH1JQ6_5bJC0Q.png" /></figure><ul><li>Le fichier est <strong>exploitable</strong></li></ul><p>Le WPA handshake a été capturé et analysé à partir d’un fichier .cap à l’aide de l’outil aircrack-ng. La capture contient un handshake valide avec PMKID, confirmant l’interception de l’échange d’authentification WPA.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/822/1*4fOQ81nNTBwAFWRj0s9CJA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/802/1*n5gVxErR6tEKkc-oyXdTdg.png" /><figcaption>attaque dictionnaire WPA.</figcaption></figure><p>Une attaque par dictionnaire a été menée hors ligne à l’aide de l’outil aircrack-ng sur un fichier contenant un handshake WPA valide. Les mots du dictionnaire ont été testés avec succès, cependant aucune clé valide n’a été trouvée, indiquant que le mot de passe ne figurait pas dans la liste utilisée.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/1*kR411LEpxOXaYjjIH8KPLg.png" /><figcaption>Convertir le fichier .cap pour hashcat</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/1*YQ800heaw5VE8bZvV0EGmA.png" /><figcaption>Lancer une attaque dictionnaire WPA</figcaption></figure><p>La tentative de conversion du fichier de capture au format hashcat (.hc22000) à l’aide de hcxpcapngtool n’a pas généré de hash exploitable. Cela est dû à la qualité limitée du fichier .cap, à l’absence de certaines trames EAPOL nécessaires, ainsi qu’au format de capture basique. Par conséquent, aucune attaque par dictionnaire n’a pu être menée avec hashcat.</p><h4><strong>****changement de fichier.cap****</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/522/1*iKMl8LeOqUkp1wf5aXtdQQ.png" /></figure><p>Le fichier http-download-good.pcapng ne contient pas de trafic Wi-Fi 802.11 et ne peut pas être utilisé pour l’analyse WPA. Un fichier pcapng contenant un handshake WPA/PMKID valide a donc été utilisé pour les attaques par dictionnaire.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/852/1*HgLYwgca8uLNoSK4wl0ZLA.png" /><figcaption>Mauvais choix : fichier ne contient pas ESSID</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/832/1*s5iH5vMy4mWrQ8LigwJYBw.png" /><figcaption>Bon choix du fichier .pcap</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/977/1*jGI6b2Rx6D615Q2dS6O4SA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/972/1*FVXSuwnLs7z45GTfWpjhwA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/833/1*pJzJGoFZomK__NFsPkTzbA.png" /><figcaption>Problème avec le wordlist utilisé(ne contient que 4 mots)</figcaption></figure><p>Le handshake est valide et exploitable.<br> L’attaque dictionnaire a échoué car le mot de passe ne figure pas dans la wordlist utilisée.</p><p>** Les deux objectifs ont été atteints : la capture et l’exploitation d’un handshake WPA ainsi que la réalisation d’attaques par dictionnaire à l’aide de aircrack-ng et hashcat.<br> L’absence de récupération de clé indique un mot de passe robuste, ce qui constitue un résultat valide en sécurité informatique.</p><h3>Attaque WPS — Reaver, Bully</h3><p><strong>WPS = Wi‑Fi Protected Setup</strong></p><p>Pourquoi WPS est vulnérable ?</p><p>Le PIN WPS :</p><ul><li>fait <strong>8 chiffres</strong></li><li>MAIS vérifié en <strong>2 parties</strong> :</li></ul><p>— 1ère partie : 4 chiffres</p><p>— 2ème partie : 3 chiffres (+ checksum)</p><p>Complexité réelle : (Beaucoup plus faible que WPA2)</p><pre>10⁴ + 10³ ≈ 11 000 tentatives</pre><p>Workflow complet (logique d’attaque)-Étapes générales</p><ul><li>Scanner les réseaux</li><li>Identifier WPS activé</li><li>Vérifier si WPS est locké</li><li>Lancer l’attaque WPS</li><li>Récupérer le PIN</li><li>Récupérer la clé WPA</li></ul><p><strong>1- Scanner les réseaux</strong></p><pre>wash -i wlan0mon</pre><p><strong>2- Identifier WPS activé</strong></p><ul><li>Colonne <strong>WPS = Yes</strong></li></ul><p><strong>3- Vérifier si WPS est verrouillé</strong></p><ul><li>Locked = No → attaquable</li><li>Locked = Yes → attaque bloquée</li></ul><p><strong>4- Lancer l’attaque WPS</strong></p><pre>reaver -i wlan0mon -b &lt;BSSID&gt; -c &lt;canal&gt; -vv</pre><p><strong>5- Récupérer le PIN</strong></p><ul><li>Attaque par brute force du PIN WPS (8 chiffres)</li></ul><p><strong>6- Récupérer la clé WPA</strong></p><ul><li>Une fois le PIN trouvé, la clé WPA est affichée automatiquement</li></ul><p><strong>7- Limitation rencontrée</strong></p><p>L’attaque WPS n’a pas pu être réalisée en conditions réelles à cause de :</p><ul><li>Absence de carte Wi‑Fi USB compatible monitor mode</li><li>Interfaces réseau limitées à Ethernet (eth0, eth1)</li><li>Impossibilité d’activer wlan0mon</li></ul><p>Justification technique:</p><blockquote><em>Les outils d’attaque Wi‑Fi nécessitent une carte capable de fonctionner en mode monitor.<br> Dans un environnement virtualisé sans adaptateur Wi‑Fi USB compatible, les attaques WPS ne peuvent être exécutées que de manière théorique ou simulée.</em></blockquote><p><strong>8- Analyse et discussion</strong></p><ul><li>WPA/WPA2 est robuste si le mot de passe est fort</li><li>Les attaques par dictionnaire dépendent fortement de la qualité de la wordlist</li><li>WPS est une <strong>faille critique</strong>, mais souvent :</li></ul><p>Désactivée / Verrouillée après plusieurs tentatives / Les réseaux modernes (WPA3) réduisent fortement ces risques</p><p><strong>C’est quoi Bully ?</strong></p><p><strong>Bully</strong> est un outil d’attaque <strong>WPS</strong> (comme Reaver, mais souvent plus rapide).<br> Il exploite une faiblesse du <strong>PIN WPS (8 chiffres)</strong> pour récupérer ensuite la <strong>clé WPA/WPA2</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/608/1*I4Vo3uDBBW6Xoo36bi1BsA.png" /><figcaption>resultat de : Scanner les réseaux WPS</figcaption></figure><pre>***Attaque WPS avec Bully***<br>sudo bully -i wlan0mon -b 50:0F:80:70:18:D0 -c 6</pre><p><strong>Brute force du PIN WPS</strong></p><p>Bully teste les PINs :</p><pre>12345670<br>12345671<br>12345672<br>...</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/580/1*kScmjl4q9Bw6h1uta9nTjw.png" /><figcaption>resultat de brute force</figcaption></figure><h3>Analyse Wi-Fi &amp; détection — Kismet, Wireshark</h3><p>Kismet est un <strong>sniffer Wi‑Fi passif</strong> :</p><ul><li>Il <strong>n’envoie rien</strong></li><li>Il <strong>écoute tout</strong></li><li>Invisible pour le réseau</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/997/1*MrSbvBnt_o_p1jh5KMTOuw.png" /><figcaption>generé par commande : sudo kismet</figcaption></figure><p>Kismet permet de détecter passivement les réseaux Wi‑Fi, d’identifier leur sécurité, les clients connectés et de capturer des handshakes WPA sans interagir activement avec le réseau.</p><p>Wireshark est un <strong>analyseur de paquets</strong> :</p><ul><li>Analyse <strong>frame par frame</strong></li><li>Décodage des protocoles</li><li>Très utilisé en <strong>forensic réseau</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/877/1*JCpyAqx8O1RpHykbSIntnA.png" /><figcaption>Handshake WPA</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/867/1*vCX_pc2m0Olr1MD_zG-pvw.png" /><figcaption>Management frames</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/847/1*OzdjeeXKF1Af0FGPW7aggA.png" /><figcaption>Beacon frames</figcaption></figure><p>L’analyse Wireshark permet d’inspecter les trames 802.11, de vérifier la présence d’un handshake WPA valide (EAPOL), d’identifier les mécanismes de chiffrement et de détecter des attaques telles que la déauthentication.</p><h3>FREE Rooms on Tryhackme:</h3><ul><li><a href="https://tryhackme.com/room/wifihacking101?utm_campaign=social_share&amp;utm_medium=social&amp;utm_content=share-completed-room&amp;utm_source=copy&amp;sharerId=6846ad24bceb8a846ffa25ca">Wifi Hacking 101</a></li><li><a href="https://tryhackme.com/room/networkingconcepts?utm_campaign=social_share&amp;utm_medium=social&amp;utm_content=share-completed-room&amp;utm_source=copy&amp;sharerId=6846ad24bceb8a846ffa25ca">Networking Concepts</a></li><li><a href="https://tryhackme.com/room/adventofcyberpreptrack?utm_campaign=social_share&amp;utm_medium=social&amp;utm_content=share-completed-room&amp;utm_source=copy&amp;sharerId=6846ad24bceb8a846ffa25ca">Advent of Cyber Prep Track</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3791ba54da55" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Week 20/30 : Building a Stronger ISMS Foundation]]></title>
            <link>https://medium.com/@issra.akrout/week-20-30-building-a-stronger-isms-foundation-21e65453bb6c?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/21e65453bb6c</guid>
            <category><![CDATA[mi]]></category>
            <category><![CDATA[iso-27001-isms]]></category>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Sun, 15 Feb 2026 17:26:50 GMT</pubDate>
            <atom:updated>2026-02-15T18:38:31.198Z</atom:updated>
            <content:encoded><![CDATA[<p>Week 20/30 : Building a Stronger ISMS Foundation</p><p><a href="https://www.eascertification.com/wp-content/uploads/2024/01/ISO-IEC-27001-2022.pdf">ISO/IEC 27001:2022</a></p><p><a href="https://dn721901.ca.archive.org/0/items/iso-27001-2013/ISO_27001_2013.pdf">Technologies de l’information — Techniques de sécurité — Systèmes de management de la sécurité de l’information — Exigences</a></p><p><a href="https://tryhackme.com/room/cybergovernanceregulation">TryHackMe | Governance &amp; Regulation</a></p><p><a href="https://youriso.co.uk/iso-27001-requirements/">ISO 27001 Requirements: Full Guide Including FREE Checklist</a></p><p><a href="https://catalyicsecurity.com/a-complete-guide-for-iso-iec-27001-lead-implementer-what-it-is-and-how-you-become/">Complete Guide for ISO 27001: 2022 Lead Implementer</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2p4gsGgnQDCLZqX_PTnSEQ.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=21e65453bb6c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[week 18/30 — From Play To Purpose]]></title>
            <link>https://medium.com/@issra.akrout/week-18-30-from-play-to-purpose-1b2b46e7f09e?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b2b46e7f09e</guid>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Sun, 25 Jan 2026 17:12:16 GMT</pubDate>
            <atom:updated>2026-01-25T17:12:16.943Z</atom:updated>
            <content:encoded><![CDATA[<h3>week 18/30 — From Play To Purpose</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/757/1*B3xkyEqqIUy4GaK55kgYHQ.png" /></figure><h3>Accessing Another Process’s Memory (Windows API)</h3><p>To manipulate a game’s memory from an external process, you use the Windows API:</p><h4>Step 1: Find the Process ID</h4><pre>DWORD pid = GetProcessIdFromName(&quot;Game.exe&quot;);</pre><h4>Step 2: Open the Process</h4><pre>HANDLE hProc = OpenProcess(PROCESS_ALL_ACCESS, FALSE, pid);</pre><h4>Step 3: Read or Write Memory</h4><pre>int value;<br>ReadProcessMemory(hProc, (LPCVOID)address, &amp;value, sizeof(value), 0);<br>int newValue = 999;<br>WriteProcessMemory(hProc, (LPVOID)address, &amp;newValue, sizeof(newValue), 0);</pre><h3>First Memory Scan (e.g., Health)</h3><p>Objective: Find and freeze player health in AssaultCube.</p><ol><li>In AssaultCube, note your current health (e.g., 100).</li><li>In Cheat Engine:</li></ol><ul><li>Value: 100</li><li>Scan Type: Exact Value</li><li>Value Type: 4 Bytes</li><li>Click First Scan</li></ul><p>3. Take damage in-game. Note new health (e.g., 92).</p><p>4. In CE, enter 92 → click Next Scan.</p><p>Repeat until only a few results remain.</p><ol><li>Double-click the correct address in the results list.</li><li>Change the value back to 100 and lock it (freeze).</li></ol><p>Result: You now have infinite health.</p><h3>Writing Your First DLL Hack</h3><pre>// DLLMain.cpp<br>#include &lt;Windows.h&gt;<br>DWORD WINAPI HackThread(HMODULE hModule)<br>{<br>    MessageBox(NULL, &quot;DLL Injected Successfully!&quot;, &quot;GameHack&quot;, MB_OK);<br>    // Your hack code goes here<br>    FreeLibraryAndExitThread(hModule, 0);<br>    return 0;<br>}<br>BOOL APIENTRY DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved)<br>{<br>    if (ul_reason_for_call == DLL_PROCESS_ATTACH)<br>        CreateThread(nullptr, 0, (LPTHREAD_START_ROUTINE)HackThread, hModule, 0, nullptr);<br>    return TRUE;<br>}</pre><h3>Injecting the DLL</h3><p>You can inject the DLL into a game like AssaultCube using tools such as:</p><h4>Extreme Injector</h4><p>Steps:</p><ol><li>Launch the target game (AssaultCube).</li><li>Open Extreme Injector.</li><li>Select the game process.</li><li>Add your compiled DLL.</li><li>Click “Inject”.</li></ol><p>You should see your MessageBox pop up if successful.</p><p><strong>Use Memory Viewer to:</strong></p><ul><li>View nearby memory values</li><li>Analyze memory regions</li><li>Inject assembly code</li></ul><p><strong>Use Structure Dissector to:</strong></p><ul><li>Analyze full objects (players, enemies)</li><li>Map fields (e.g., X/Y/Z, ammo)</li></ul><p><strong>CPU architecture:</strong></p><ul><li>x86 (32-bit)</li><li>x64 (64-bit, also referred to as x86_64 or AMD64)</li></ul><h3>Identifying Architecture of a Game</h3><p>To determine if a game is 32-bit or 64-bit:</p><ul><li>Open the executable in PE-bear, CFF Explorer, or IDA Free.</li><li>Check the machine type field in the PE header:</li><li>0x14C = x86 (32-bit)</li><li>0x8664 = x64 (64-bit)</li></ul><p>In Windows Task Manager:</p><ul><li>32-bit programs will show as game.exe *32</li></ul><h3>How to Patch in Practice (Cheat Engine)</h3><ol><li>Attach Cheat Engine to the game process.</li><li>Find the target instruction using the debugger or AOB scan.</li><li>Right-click the instruction → Replace with NOPs.</li><li>Alternatively, use “Auto Assemble” to write a script:</li></ol><pre>[ENABLE]<br>00401090:<br>nop<br>nop<br><br>[DISABLE]<br>00401090:<br>sub [esi+0F8], eax</pre><h3>What Is a Function Call?</h3><p>A function call is an instruction that transfers execution to a block of code performing a specific task. In assembly, it appears as:</p><pre>call 0x00401000</pre><p>When call is executed:</p><ul><li>The address of the next instruction is pushed to the stack (return address).</li><li>The CPU jumps to the target address (0x00401000).</li><li>After the function executes, it returns via the ret instruction.</li></ul><p><strong>— Reverse Engineer Health Logic in AssaultCube</strong></p><ul><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-1-launch-the-game-and-attach-cheat-engine">Step 1: Launch the Game and Attach Cheat Engine</a></li><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-2-scan-for-player-health">Step 2: Scan for Player Health</a></li><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-3-identify-the-health-address">Step 3: Identify the Health Address</a></li><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-4-trigger-a-health-change">Step 4: Trigger a Health Change</a></li><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-5-disassemble-the-instruction">Step 5: Disassemble the Instruction</a></li><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-6-patch-the-instruction">Step 6: Patch the Instruction</a></li><li><a href="https://courses.redteamleaders.com/courses/c0f369dc-356d-4362-8942-b5446e02164b/take/29-hands-on-reverse-engineer-health-logic-in-assaultcube#user-content-step-7-verify-behavior">Step 7: Verify Behavior</a></li></ul><h3>Spoofing IsDebuggerPresent() Return</h3><p>You can hook IsDebuggerPresent() using a DLL injection technique and force it to always return FALSE:</p><pre>BOOL WINAPI HookedIsDebuggerPresent() {<br>    return FALSE;<br>}</pre><p>(Implementation of the hook with Detours, MinHook, or manual patching is discussed in later modules.)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/846/1*XgbgZ43RNrXHWX6TLxAwgQ.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b2b46e7f09e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[    ()]]></title>
            <link>https://medium.com/@issra.akrout/-1edc1921f310?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/1edc1921f310</guid>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Sun, 28 Dec 2025 19:12:09 GMT</pubDate>
            <atom:updated>2025-12-28T19:12:09.698Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/888/1*jirdMXsae9JkdcpUlZ3f0w.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1edc1921f310" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Certified Phishing Prevention Specialist (CPPS)]]></title>
            <link>https://medium.com/@issra.akrout/certified-phishing-prevention-specialist-cpps-70ebeb11dc08?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/70ebeb11dc08</guid>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Thu, 25 Dec 2025 23:01:44 GMT</pubDate>
            <atom:updated>2025-12-28T19:15:55.615Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k5fhKIyTYW4H3NKE_nY47g.jpeg" /></figure><h3>Introduction to Phishing</h3><p>Phishing = “fishing” + “phreaking”</p><h4>Types of Phishing</h4><p>1. Email Phishing ( In 2021, a phishing campaign targeted Microsoft users by spoofing Office 365 login pages )</p><p>2. Spear Phishing ( In 2016, spear-phishing emails targeting the Democratic National Committee resulted in a high-profile data breach. )</p><p>3. Smishing (SMS Phishing) ( In 2020, a widespread smishing campaign impersonated Netflix, tricking users into entering credentials on a fake site. )</p><p>4. Vishing (Voice Phishing) ( A 2021 scam impersonated Amazon customer service, claiming unauthorized purchases and asking for account details. )</p><p>5. Clone Phishing ( Cybercriminals cloned emails from a trusted supplier in 2022, tricking businesses into wiring payments to fraudulent accounts. )</p><p>6. Quishing (QR Code Phishing) ( A campaign in 2021 swapped restaurant QR codes with fake ones that stole diners’ payment information. )</p><p>7. Social Media Phishing ( In 2019, attackers used Facebook Messenger to send phishing links disguised as “shared videos” from trusted contacts. )</p><h4>Impact of Phishing Attacks</h4><p>1. Impact on Organizations(<strong>Financial Losses, Reputation Damage, Operational Disruption)</strong></p><p>2. Impact on Individuals(<strong>Financial Fraud, Identity Theft, Emotional and Psychological Effects)</strong></p><p>IRS Data Breach (2016)/Twitter Spear Phishing Attack (2020)/Sony Pictures Hack (2014)/2023 Phishing Campaigns Using AI</p><h3>Anatomy of a Phishing Attack</h3><h4>Social engineering techniques :</h4><ol><li>Urgency and Pressure Tactics / 2. Authority and Trust / 3. Fear and Alarm / 4. Curiosity / 5. Familiarity and Mimicry / 6. Reciprocity / 7. Social Proof</li></ol><h4>Key Elements Used to Deceive Victims :</h4><ol><li>Spoofed Addresses / 2. Fake Links (URL Spoofing) / 3. Malicious Attachments /</li></ol><h4>Examples of Phishing Emails</h4><ol><li>Google Docs Phishing Attack (2017)</li><li>Sony PlayStation Phishing Scam (2021)</li></ol><h3>Recognizing the Red Flags</h3><p>Common Signs of Phishing Emails : Suspicious or Unfamiliar Sender Addresses / Generic Greetings / Urgent and Threatening Language / Unusual Attachments or Links / Spelling and Grammar Mistakes /</p><h4>Phishing Email Checklist: How to Spot Red Flags</h4><ol><li>Check the Sender’s Email Address</li><li>Look for Generic Greetings</li><li>Examine the Language and Tone</li><li>Hover Over Links and Inspect URLs</li><li>Be Careful with Attachments</li><li>Verify the Message’s Authenticity</li><li>Look for Inconsistent or Suspicious Formatting</li><li>Check for Spoofed URLs in Email Signature</li><li>Use Security Tools and Verify</li></ol><h3>Tools and Techniques to Stay Protected</h3><p><strong>Spam Filters, Antivirus Software, and Firewalls</strong></p><p><strong>Password Management Best Practices : </strong>Use Strong and Unique Passwords / Change Passwords Regularly / <strong>Best Practice</strong>: Always use a unique password for each account, and consider using a password manager to enhance security.</p><p><strong>Multi-Factor Authentication (MFA) and Its Importance</strong></p><p><strong>Reporting Phishing Attempts to IT and Cybersecurity Teams</strong></p><h3>Organizational Policies and Compliance</h3><h4>Acceptable Use Policies (AUP) :</h4><p>AUPs might include rules such as:</p><ul><li>Prohibiting the sharing of login credentials.</li><li>Mandating the use of organization-approved email platforms.</li><li>Reporting suspicious emails immediately to the IT department.</li></ul><h4>Regulatory Requirements</h4><p>General Data Protection Regulation (GDPR) (EU)</p><p>Health Insurance Portability and Accountability Act (HIPAA) (USA)</p><p>California Consumer Privacy Act (CCPA) (USA)</p><p>ISO/IEC 27001 (International Standard)</p><h4>Continuous Training and Awareness</h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=70ebeb11dc08" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Week 13/30 — Réseaux avancés & sécurité]]></title>
            <link>https://medium.com/@issra.akrout/week-13-30-r%C3%A9seaux-avanc%C3%A9s-s%C3%A9curit%C3%A9-2a1a3d69f0b0?source=rss-235424ee9931------2</link>
            <guid isPermaLink="false">https://medium.com/p/2a1a3d69f0b0</guid>
            <dc:creator><![CDATA[Issra Akrout ]]></dc:creator>
            <pubDate>Sun, 14 Dec 2025 18:09:52 GMT</pubDate>
            <atom:updated>2025-12-14T18:09:52.852Z</atom:updated>
            <content:encoded><![CDATA[<h3>Week 13/30 — <strong>Advanced Networks &amp; Security</strong></h3><h3>Protocoles avancés : SMBv2, SMBv3, NTLMv1/v2</h3><p><strong>SMB (Server Message Block)</strong> est un protocole utilisé pour le <strong>partage de fichiers et d’imprimantes</strong> sur un réseau.</p><p><strong>SMBv2 et SMBv3</strong> sont des versions améliorées par rapport à SMBv1, avec plus de sécurité et de performance.</p><p><strong>NTLM (v1/v2)</strong> sont des protocoles d’<strong>authentification Windows</strong> souvent utilisés avec SMB. NTLMv2 est beaucoup plus sûr que NTLMv1 qui est vulnérable à certains types d’attaques.</p><h3>Sécurité réseau : IDS, IPS, Firewall, VPN</h3><ul><li><strong>Firewall</strong> : Filtre le trafic entrant/sortant selon des règles (pare-feu).</li><li><strong>IDS (Intrusion Detection System)</strong> : <strong>Détecte</strong> les activités malveillantes sur le réseau.</li><li><strong>IPS (Intrusion Prevention System)</strong> : <strong>Bloque automatiquement</strong> les activités suspectes.</li><li><strong>VPN (Virtual Private Network)</strong> : Crée une <strong>connexion chiffrée sécurisée</strong> entre deux réseaux ou entre un utilisateur et un réseau.</li></ul><h3>Intrusion Detection(TryHackMe)</h3><p>Learn cyber evasion techniques and put them to the test against two IDS</p><h4><strong>Network-based IDS (NIDS)</strong></h4><p><strong>Avantages : </strong>Une seule installation peut surveiller tout le réseau. Déploiement plus simple que d’autres types d’IDS.</p><p><strong>Limites : </strong>Taux élevé de <strong>faux positifs</strong> à cause du volume de trafic et de règles difficiles à ajuster. / Nécessite un <strong>temps d’apprentissage</strong> pour distinguer le trafic normal de l’anormal. / Moins efficace face au <strong>chiffrement</strong> du trafic (TLS, etc.).</p><p><strong>Déploiement : </strong>Généralement placé <strong>côté LAN</strong>, parfois des deux côtés du firewall. / Détecter les attaques externes a peu de valeur car elles sont constantes. / Peut inclure des fonctions <strong>IPS</strong> (blocage automatique), souvent désactivées pour éviter les blocages erronés.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/792/1*BQ0I2bgOEvXpJfgX8da0Yw.png" /></figure><h3>Secure Network Architecture(TryHackMe)</h3><h4>Network Segmentation</h4><p>With subnetting in place, there is no restriction in place as to where the infected device could connect as long as the proper routes are in place, leaving sensitive information and servers open to the unknown device. How do we fix this problem? VLANs! → <strong>VLANs</strong> (<strong>V</strong>irtual <strong>LAN</strong>) are used to segment portions of a network at layer two and differentiate devices</p><h4>Common Secure Network Architecture</h4><p>how do we properly implement VLANs as a security boundary? Security zones! <strong>Security zones</strong> define <strong>what</strong> or <strong>who</strong> is in a VLAN and how traffic can travel <strong>in</strong> and <strong>out</strong>.</p><p><strong>commonly standardized zones: </strong>External , DMZ (demilitarized zone) , Trusted , Restricted , Management , Audit</p><h4>Network Security Policies and Controls</h4><p>traffic filtering provides network security, validation, and segmentation by filtering network traffic based on pre-defined criteria.</p><p>how create an ACL and defines the policy:</p><ul><li>Create the new access-list policy, defining the ACL number (1–2699)</li></ul><p>set policy access-list &lt;acl_number&gt;</p><ul><li>Set the description of the access list.</li></ul><p>set policy access-list &lt;acl_number&gt; description &lt;text&gt;</p><ul><li>Create a new rule (or ACE) under the ACL and define the action of the rule.</li></ul><p>set policy access-list &lt;acl_number&gt; rule &lt;1-65535&gt; action &lt;permit|deny&gt;</p><ul><li>Define criteria or parameters for the rule to enforce/match.</li></ul><p>set policy access-list &lt;acl_number&gt; rule &lt;1-65535&gt; &lt;destination|source&gt; &lt;any|host|inverse-mask|network&gt;</p><p>At this point, we have an ACL and an ACE that is actively being enforced by the router.</p><h4>Zone-Pair Policies and Filtering</h4><p>Traffic correlation is standardized as the state of a packet, e.g. (protocol, process, direction, etc.) We must employ a firewall to parse the state of a packet and enforce policies based on that state.</p><p><strong>Zone-pairs</strong> are a direction-based and stateful policy that will enforce the traffic in single directions per each VLAN</p><ul><li>the generic syntax for adding a zone-pair.</li></ul><p>set zone-policy zone &lt;zone A&gt; from &lt;zone B&gt; firewall &lt;name&gt; &lt;ruleset name&gt;</p><ul><li>we will set the zone-pair for both the LAN → WAN and WAN → LAN pairs.</li></ul><ol><li>set zone-policy zone LAN from WAN firewall name lan-wan</li><li>set zone-policy zone WAN from LAN firewall name wan-lan</li></ol><h4>Validating Network Traffic</h4><p>we must use SSL/TLS inspection to intercept HTTPS connections.</p><p>SSL/TLS inspection uses an <strong>SSL proxy</strong> to intercept protocols, including HTTP, POP3, SMTP, or other SSL/TLS encrypted traffic. Once intercepted, the proxy will decrypt the traffic and send it to be processed by a <strong>UTM</strong> (<strong>U</strong>nified <strong>T</strong>hreat <strong>M</strong>anagement) platform. UTM solutions will employ deep SSL inspection, feeding the decrypted traffic from the proxy into other UTM services, including but not limited to web filters or <strong>IPS</strong> (<strong>I</strong>ntrusion <strong>P</strong>revention <strong>S</strong>ystem), to process the information.</p><h4>Addressing Common Attacks</h4><p><strong>DHCP Snooping : </strong><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst6500/ios/12-2SXF/native/configuration/guide/swcg/snoodhcp.pdf">Cisco</a> defines <strong>DHCP snooping</strong> as “a security feature that acts like a firewall between untrusted hosts and trusted DHCP servers.”</p><p><strong>Dynamic ARP Inspection : </strong><a href="https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst4500/12-2/25ew/configuration/guide/conf/dynarp.html">Cisco</a> defines <strong>ARP inspection</strong> as “a security feature that validates <strong>A</strong>ddress <strong>R</strong>esolution <strong>P</strong>rotocol (<strong>ARP</strong>) packets in a network.”</p><p>ARP inspection will <strong>validate</strong> and <strong>rate-limit</strong> ARP packets as necessary; if an ARP packet’s MAC and IP address do not match, the protocol will intercept, log, and discard the packet. ARP inspection uses the DHCP binding database filled from DHCP snooping as its list of binding IP addresses.</p><h3>Sniffing &amp; spoofing : Wireshark pratique</h3><ul><li><strong>Sniffing</strong> : Capturer le trafic réseau pour analyser ce qui passe entre des machines.</li><li><strong>Spoofing</strong> : Faire croire qu’on est une autre machine (ex : ARP spoofing).</li></ul><h3>L2 MAC Flooding &amp; ARP Spoofing(TryHackMe)</h3><h4>Network Discovery</h4><p>To know what is the ip address and the network’s CIDR prefix ip address show eth1 or the shorthand version: ip a s eth1</p><p>and we use nmap to know how many other live hosts are there</p><h4>Passive Network Sniffing</h4><p>for a more verbose output that prints each packet (minus its link level header) in ASCII format:</p><p>tcpdump -A -i eth1</p><p>Now, let’s take a closer look at the captured packets! We can redirect them into a <strong>pcap</strong> file providing a destination file via the <strong>-w</strong> argument:</p><p>tcpdump -A -i eth1 -w /tmp/tcpdump.pcap</p><p>Example to transfer the packet capture using <strong>scp</strong> and open it in Wireshark:</p><p>scp admin@MACHINE_IP:/tmp/tcpdump.pcap .<br>wireshark tcpdump.pcap</p><h4>Sniffing while MAC Flooding</h4><p>MAC flooding could trigger an alarm in a SOC. No, seriously, suspicious layer 2 traffic can easily be detected and reported by state-of-the-art and properly configured network devices.</p><p>This way, you can leave the <strong>tcpdump</strong> process running in the foreground on the first SSH session:</p><p>tcpdump -A -i eth1 -w /tmp/tcpdump2.pcap</p><p>Now, on the second SSH session, buckle up and let <a href="http://manpages.ubuntu.com/manpages/bionic/man8/macof.8.html"><strong>macof</strong></a> run against the interface to start flooding the switch:</p><p>macof -i eth1</p><p>After around 30 seconds, stop both <strong>macof</strong> and <strong>tcpdump</strong> (Ctrl+C).</p><p>As in the previous task, transfer the <strong>pcap</strong> to your machine (<strong>kali/AttackBox)</strong> and take a look:</p><p>scp admin@MACHINE_IP:/tmp/tcpdump2.pcap .<br>wireshark tcpdump2.pcap</p><h4>Man-in-the-Middle: Intro to ARP Spoofing</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/466/1*501r28DIqyrC1utTJOtPSQ.png" /></figure><p>As you may have noticed, MAC Flooding can be considered a real “noisy” technique. In order to reduce the risk of detection and DoS we will leave <strong>macof</strong> aside for now. Instead, we are going to perform so-called <strong>ARP cache poisoning</strong> attacks ‘against Alice and Bob’, in an attempt to become a fully-fledged Man-in-the-Middle (MITM).</p><p>using <strong>ettercap</strong> to launch an ARP Spoofing attack:</p><p>ettercap -T -i eth1 -M arp</p><h4>MITM &amp; ARP spoofing</h4><ul><li><strong>MITM (Man-In-The-Middle)</strong> : L’attaquant s’intercale entre deux machines et <strong>intercepte ou modifie</strong> leurs communications.</li><li><strong>ARP Spoofing/Poisoning</strong> : Technique pour usurper l’adresse MAC d’une machine et ainsi rediriger le trafic vers soi.</li></ul><p>Théorie : ARP n’est pas sécurisé parce qu’il n’a pas de mécanisme d’authentification.</p><h3>Détection &amp; contournement : Techniques d’évasion</h3><ul><li>Une fois qu’on connaît les attaques, il faut <strong>savoir comment elles sont détectées (IDS/IPS)</strong> et quelles techniques peuvent les rendre <strong>moins visibles ou plus difficiles à détecter</strong>.</li><li>Cela inclut l’<strong>évasion de signature IDS</strong>, fragmentation de paquets, etc.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2a1a3d69f0b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>