<?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 Exploit The Edge on Medium]]></title>
        <description><![CDATA[Stories by Exploit The Edge on Medium]]></description>
        <link>https://medium.com/@ExploitTheEdge?source=rss-2724d5aaf00d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*oNYWxW2Ncw0JTiYlCvBoQg.png</url>
            <title>Stories by Exploit The Edge on Medium</title>
            <link>https://medium.com/@ExploitTheEdge?source=rss-2724d5aaf00d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 02:02:24 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@ExploitTheEdge/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[How a Lazy Admin Taught Me Everything About Web App Pentesting]]></title>
            <link>https://medium.com/@ExploitTheEdge/how-a-lazy-admin-taught-me-everything-about-web-app-pentesting-4c9a1e4748d1?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/4c9a1e4748d1</guid>
            <category><![CDATA[tryhackme]]></category>
            <category><![CDATA[privilege-escalation]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[pentesting]]></category>
            <category><![CDATA[defensive-security]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Mon, 16 Nov 2026 09:01:11 GMT</pubDate>
            <atom:updated>2026-11-16T09:01:11.764Z</atom:updated>
            <content:encoded><![CDATA[<p><em>One TryHackMe room, a forgotten backup file, and the sudo misconfiguration that handed me root — plus what defenders should actually watch for.</em></p><p><em>Disclosure: this story was drafted with AI assistance from our own guide and checked against the sources cited there.</em></p><p>The room is called <em>Lazy Admin</em>. The name tells you the ending before you start.</p><p>I deployed the TryHackMe instance, connected to the VPN, and ran the first nmap scan. Two ports: SSH on 22, Apache on 80. Nothing fancy. But the web root hid a /content/ directory running SweetRice CMS 1.5.1 — a project abandoned since 2018 — and a /backup/ directory serving a MySQL dump to anyone who asked.</p><p>That backup file changed everything.</p><h4>The Reconnaissance That Mattered</h4><p>Most write-ups lead with tool output. The real lesson is knowing <em>where</em> to look. A full TCP SYN scan with service detection (nmap -sS -sV -sC -p- -T4) gave me the baseline. But gobuster against the common wordlist revealed the attack surface: /content/ for the CMS, /content/attachment/ for uploads, and /backup/mysql_backup.sql sitting wide open.</p><p>The SQL dump contained an admin row:</p><pre>INSERT INTO `users` VALUES (1,&#39;manager&#39;,&#39;42f749ade7f9e195bf475f37a44cafcb&#39;,...);</pre><p>Thirty-two hex characters. Unsalted MD5. hashid confirmed it; hashcat -m 0 with rockyou.txt cracked it in seconds: Password123.</p><h4>From Admin Panel to Shell</h4><p>SweetRice’s admin panel lives at /content/as/. The credentials worked. The <strong>Ad Manager</strong> feature accepts image uploads for advertisements — but validation happens client-side. A PHP reverse shell renamed to shell.php.jpg bypasses the extension check and lands in /content/attachment/ with a randomized name.</p><p>One curl to trigger it. One nc listener catching the connection. A quick Python PTY upgrade later, I had a shell as itguy and the user flag.</p><h4>The Sudo Misconfiguration That Gave Root</h4><p>sudo -l is the command that separates CTF players from people who understand Linux internals.</p><pre>User itguy may run the following commands on THM-Chal:<br>    (root) NOPASSWD: /usr/bin/perl /home/itguy/backup.pl</pre><p>The script itself was trivial:</p><pre>#!/usr/bin/perl<br>system(&quot;/bin/sh&quot;, &quot;/etc/copy.sh&quot;);</pre><p>It executes /etc/copy.sh as root. That file was writable by itguy. I overwrote it with a reverse shell payload, made it executable, started a second listener, and ran the sudo command. Root shell. Root flag.</p><p>This pattern is documented on GTFOBins. It’s not a novel exploit — it’s a configuration failure that persists in production environments.</p><h4>What Blue Teams Should Actually Detect</h4><p>The offensive path is clean. The defensive gaps are where the learning sticks.</p><p>| Stage | Detection Opportunity | Mitigation | | — — — -| — — — — — — — — — — — | — — — — — — | | Backup disclosure | Web logs: GET /backup/mysql_backup.sql 200 | Deny public access to /backup/; store backups off-box | | Weak admin hash | Audit password hashes for 32-char MD5 | Enforce bcrypt/argon2; require MFA for CMS admin | | Authenticated RCE | WAF/ModSecurity: multipart upload with .php in filename | Disable PHP execution in upload dirs (php_admin_flag engine off); validate MIME + extension server-side | | Sudo perl abuse | Auditd watch on /etc/copy.sh; sudo -l anomalies | Remove NOPASSWD; restrict sudo to specific args; use Defaults!/usr/bin/perl !requiretty |</p><p>The hardening commands are straightforward:</p><pre># 1. Remove backup exposure<br>chmod 700 /var/www/html/backup<br><br># 2. Disable PHP in uploads<br>echo &#39;php_admin_flag engine off&#39; &gt;&gt; /etc/apache2/sites-available/000-default.conf<br>systemctl reload apache2<br><br># 3. Fix sudoers<br>visudo  # replace NOPASSWD line with specific command + args</pre><h4>What This Room Still Teaches in 2024</h4><p>SweetRice hasn’t been patched since 2018. CVE-2018–19564 remains unpatched. The tooling has evolved — gobuster v3 uses a dir subcommand, hashcat runs on OpenCL 3.0, TryHackMe&#39;s AttackBox runs in the browser — but the root causes haven&#39;t changed:</p><ul><li>Backup files exposed to the web root</li><li>Unsalted MD5 hashes in production databases</li><li>Client-side validation on file uploads</li><li>NOPASSWD sudo rules calling scripts that trust writable files</li></ul><p>Every one of these appears in real incident reports. The room is easy. The lessons are not.</p><p>**Full technical guide with every command and output: <a href="https://edge.3dynamix.uk/cybersecurity/tryhackme-lazy-admin-complete-walkthrough-defensive-lessons/">https://edge.3dynamix.uk/cybersecurity/tryhackme-lazy-admin-complete-walkthrough-defensive-lessons/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4c9a1e4748d1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Avalanche Isn’t One Blockchain — It’s a Network of Them]]></title>
            <link>https://medium.com/@ExploitTheEdge/avalanche-isnt-one-blockchain-it-s-a-network-of-them-671d06996126?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/671d06996126</guid>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[web3]]></category>
            <category><![CDATA[defi]]></category>
            <category><![CDATA[avalanche]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Wed, 14 Oct 2026 08:00:31 GMT</pubDate>
            <atom:updated>2026-10-14T08:00:31.736Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A 2026 look at subnets, Teleporter messaging, and the upgrades that quietly changed how Avalanche works since 2022</em></p><p><em>Disclosure: this story was drafted with AI assistance from our own guide and checked against the sources cited there.</em></p><p>Most people describe Avalanche as “another EVM chain.” That’s not wrong, exactly — but it misses the point. Avalanche was never designed to be a single chain competing for the same block space as everyone else. It was designed to be a <em>network of chains</em>, each with its own validator set, its own virtual machine, and its own rules — all settling back to a shared, permissionless Proof-of-Stake base layer.</p><p>If you last looked at Avalanche around 2022, a lot has changed quietly underneath. Here’s what the architecture actually looks like now, and what’s different.</p><h4>Three chains, one validator set</h4><p>At the center is the <strong>Primary Network</strong> — the default validator set that every validator joins. It secures three built-in chains, each with a distinct job:</p><ul><li><strong>P-Chain (Platform Chain)</strong> — handles validator registration, subnet creation, and staking. Runs the PlatformVM.</li><li><strong>C-Chain (Contract Chain)</strong> — the EVM-compatible chain most people actually interact with. Runs Coreth, a modified go-ethereum client.</li><li><strong>X-Chain (Exchange Chain)</strong> — a UTXO-based chain for asset creation and trading. Runs the Avalanche Virtual Machine (AVM). It’s not EVM-compatible, and by 2026 most DeFi activity has moved off it to C-Chain or subnets — X-Chain is now mostly AVAX transfers and native asset trades.</li></ul><p>Validators validate all three Primary Network chains by default. That’s the base layer. <strong>Subnets</strong> are where it gets more interesting.</p><h4>Subnets: shared security, custom rules</h4><p>A subnet is a dynamic set of validators that agree on the state of one or more blockchains. A validator can belong to multiple subnets, but each blockchain is validated by exactly one subnet. In practice, this means teams can spin up a chain with its own gas rules, its own validator requirements (hardware, geography, even KYC), and its own virtual machine — while still drawing validators from the same Proof-of-Stake pool that secures the Primary Network.</p><p>To be a subnet validator, you still need to be a Primary Network validator first (≥2,000 AVAX staked). Subnet-specific rewards come from the subnet’s own emission or fees, not from Primary Network inflation.</p><h4>Two consensus protocols, not one</h4><p>Avalanche runs two leaderless consensus protocols, both based on repeated sub-sampling of the validator set:</p><ul><li><strong>Avalanche Consensus</strong> — DAG-based, used by X-Chain, finalizing transactions in roughly 1–2 seconds. Good fit for linear asset transfers.</li><li><strong>Snowman Consensus</strong> — a linearized version used by P-Chain and C-Chain, producing a total order of blocks. Smart contract execution needs that strict ordering, which DAG structures don’t naturally provide.</li></ul><p>Both give probabilistic finality with no fixed block time, though in practice C-Chain blocks land roughly every 1–2 seconds. Sampling parameters are configurable per subnet, so throughput and finality characteristics aren’t one-size-fits-all across the network.</p><h4>What’s actually new since 2022</h4><p>The biggest shift isn’t a single flashy feature — it’s that subnets stopped being isolated islands.</p><p><strong>Avalanche Warp Messaging (AWM)</strong>, introduced in 2023, gave subnets a native way to send verifiable cross-subnet messages using BLS multi-signature aggregation from the sending subnet’s validators — no trusted bridge required. <strong>Teleporter</strong>, built on top of AWM in 2024, generalized this into a standard interface (TeleporterMessenger) that any subnet can use to send and receive messages, replacing the custom bridge contracts that used to be the norm.</p><p><strong>HyperSDK</strong>, also from 2024, opened the door to custom, non-EVM virtual machines written in Go — used by projects like Dexalot’s order-book DEX and various gaming subnets. Some of these projects cite throughput claims north of 10,000 TPS; that’s project-specific and worth verifying per subnet rather than taking as a network-wide figure.</p><p>On the C-Chain side, the <strong>Cortina upgrade</strong> (April 2024) brought EIP-4844 blob transactions, EIP-6780’s SELFDESTRUCT deprecation, and a modified fee market — aligning C-Chain more closely with Ethereum’s own roadmap. The <strong>Etna upgrade</strong> (late 2024) adjusted the emission curve, targeting a long-term terminal supply of 720 million AVAX.</p><p>Another addition that didn’t exist in the earlier design: <strong>Evergreen Subnets</strong>, permissioned subnets launched in 2023 with their own validator sets and KYC enforcement, aimed squarely at regulated institutional use cases.</p><h4>Tokenomics, briefly</h4><p>AVAX has a hard cap of 720,000,000 tokens. Validators need 2,000 AVAX staked to join the Primary Network; delegators need 25 AVAX minimum, with a minimum 2% delegation fee set by the validator. Base fees on both C-Chain and P-Chain are burned entirely, EIP-1559-style, which creates deflationary pressure as usage rises. Circulating supply and live staking APY move daily, so treat any snapshot figure as a point-in-time check rather than a constant.</p><h4>Where the ecosystem activity actually sits</h4><p>According to DefiLlama’s recent trend data, Trader Joe and Benqi lead C-Chain DEX and lending volume respectively, with Pangolin v3 present but holding a smaller share. Subnet-native DeFi — Dexalot’s order book being the clearest example — represents a different category entirely, built on HyperSDK rather than the EVM.</p><h4>The practical takeaway</h4><p>If you’re building or operating in this ecosystem in 2026, the checklist looks different than it did a few years ago: read the Subnet-EVM and HyperSDK docs before picking a VM framework, test Teleporter messaging on Fuji testnet before touching mainnet, keep validator nodes on avalanchego v1.11+ for Etna compatibility, and if you’re eyeing institutional use cases, look specifically at Evergreen Subnet requirements rather than assuming standard subnet rules apply.</p><p>The architecture rewards understanding — Avalanche in 2026 isn’t the same network it was when subnets were a whitepaper concept and bridges were the only way across.</p><p>For the full technical breakdown — RPC commands, validator staking calls, subnet deployment steps, and source citations — read the complete guide: <a href="https://edge.3dynamix.uk/crypto/avalanche-in-2026-architecture-subnets-and-what-changed-since/">https://edge.3dynamix.uk/crypto/avalanche-in-2026-architecture-subnets-and-what-changed-since/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=671d06996126" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[IPFS in 2026: The Decentralized Web Finally Works]]></title>
            <link>https://medium.com/@ExploitTheEdge/ipfs-in-2026-the-decentralized-web-finally-works-1c6eb641fef8?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/1c6eb641fef8</guid>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[ipfs]]></category>
            <category><![CDATA[web3]]></category>
            <category><![CDATA[crypto]]></category>
            <category><![CDATA[decentralization]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Mon, 12 Oct 2026 08:00:33 GMT</pubDate>
            <atom:updated>2026-10-12T08:00:33.744Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Content-addressed storage stopped being a crypto experiment and became something you can actually ship on</em></p><p><em>Disclosure: this story was drafted with AI assistance from our own guide and checked against the sources cited there.</em></p><h4>The internet still has a single point of failure</h4><p>When AWS us-east-1 has a bad day, huge chunks of the web disappear with it. Governments still block domains at the DNS or IP level. The reason is structural: the web addresses content by <em>location</em> — you ask a DNS name for a specific server, and if that server is down or blocked, the content is gone, even if a thousand copies exist elsewhere.</p><p>IPFS flips that model. Instead of asking “where is this file,” you ask “give me the file with this hash.” That hash is called a CID (content identifier), and any peer holding the matching data can serve it. No single server, no single point of censorship.</p><p>IPFS has been around for years, but it spent a long time as a protocol you had to fight with to use in production. That changed. The implementation stabilized, the networking got faster, and — critically — the ecosystem solved the one problem that used to kill every IPFS pitch: persistence.</p><h4>The CID, decoded</h4><p>Every piece of content on IPFS gets a self-describing hash like:</p><pre>bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi</pre><p>Since Kubo (the reference Go implementation, formerly go-ipfs) version 0.17, <strong>CIDv1 is the default</strong>. That string encodes the CID version, the content codec, the hash algorithm, and the hash itself. You can inspect it directly: ipfs cid info &lt;CID&gt;.</p><p>Large files are split into 256 KiB blocks, each with its own CID, linked together through a Merkle DAG. Directories with huge numbers of files use HAMT (Hash Array Mapped Trie) sharding — a feature that’s been in go-ipfs since version 0.7.0 back in 2020 — so you don’t end up with one giant unwieldy metadata block.</p><p>Mutability comes from IPNS, which publishes a pointer that can be updated to point at new CIDs over time. Since Kubo 0.18, IPNS over PubSub can resolve in seconds rather than minutes, though resolution speed still varies with network conditions.</p><h4>The gap that actually mattered: pinning</h4><p>Here’s the thing earlier writeups on IPFS tended to skip past: <strong>IPFS does not replicate content automatically.</strong> If you add a file to your node and then take your node offline, that file disappears from the network unless someone else explicitly pinned it. Adding something to IPFS does not mean it’s safe.</p><p>This is the part that’s genuinely different now. There’s a real pinning ecosystem — Pinata, Web3.Storage, Filebase, Crust, and others — that lets you push a CID to a service with an SLA and multi-region replication, instead of running a node 24/7 yourself. The workflow is simple: configure the remote pinning service once (ipfs pin remote service add), then pin any CID to it (ipfs pin remote add). You can verify the pin succeeded by checking the service or just curling a gateway URL for that CID.</p><p>For longer-term, cryptoeconomically enforced storage, there’s Filecoin — a separate blockchain that uses IPFS’s content addressing but adds storage proofs and a retrieval market. You don’t need Filecoin to use IPFS day to day, but several pinning services (Web3.Storage, Filebase, Lighthouse) will handle Filecoin deals for you behind the scenes, so you don’t have to touch the deal-making tooling directly.</p><h4>How people actually fetch this stuff</h4><p>Most users aren’t running a node — they’re hitting an HTTP gateway. There are a few patterns in use: path gateways (ipfs.io/ipfs/&lt;CID&gt;), subdomain gateways (&lt;CID&gt;.ipfs.dweb.link) that give better origin isolation, and trustless gateways where a browser-based client like Helia actually verifies the returned blocks against the CID instead of just trusting the gateway operator.</p><p>For anyone building something, the practical options to know are ipfs.io, dweb.link, gateway.pinata.cloud, cloudflare-ipfs.com, and ipfs.filebase.io. You can also run your own gateway with ipfs gateway --port 8080 --subdomains if you want to avoid rate limits or keep lookups private.</p><h4>What this means if you’re shipping something</h4><p>The realistic use cases right now are static sites, NFT metadata, backups, and datasets — not yet dynamic applications. A typical static site workflow looks like: build your site, add it to IPFS to get a root CID, pin that CID to a service, publish it under an IPNS key, and let a subdomain gateway resolve that IPNS name for visitors.</p><p>For sharing something like a dataset where recipients need to verify integrity, you can export a CAR file (ipfs dag export) as a portable archive, and the recipient re-imports and checks it against the root CID before pulling the data.</p><p>And since IPFS is public by default, anything sensitive needs to be encrypted <em>before</em> it goes in — the network itself doesn’t make your data private, only resistant to disappearing.</p><p>None of this requires blind trust in a single company’s infrastructure anymore. It requires a node, a pinning service, and a gateway — pieces that now actually fit together.</p><p>If you want the full technical walkthrough — commands, daemon flags, low-resource node profiles, and a step-by-step checklist to pin and publish something this week — the complete guide is here: <a href="https://edge.3dynamix.uk/crypto/ipfs-in-2026-content-addressed-storage-that-actually-works/">https://edge.3dynamix.uk/crypto/ipfs-in-2026-content-addressed-storage-that-actually-works/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1c6eb641fef8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I Built a SOC Home Lab to Catch Attackers Hiding in Plain Sight]]></title>
            <link>https://medium.com/@ExploitTheEdge/i-built-a-soc-home-lab-to-catch-attackers-hiding-in-plain-sight-aba4923adbd3?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/aba4923adbd3</guid>
            <category><![CDATA[blue-team]]></category>
            <category><![CDATA[threat-detection]]></category>
            <category><![CDATA[splunk]]></category>
            <category><![CDATA[soc]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Mon, 05 Oct 2026 08:00:33 GMT</pubDate>
            <atom:updated>2026-10-05T08:00:33.374Z</atom:updated>
            <content:encoded><![CDATA[<p><em>How to emulate LOLBins and write Splunk SPL rules that actually cut through the noise</em></p><p><em>Disclosure: this story was drafted with AI assistance from our own guide and checked against the sources cited there.</em></p><p>Here’s a problem most blue teamers learn the hard way: attackers don’t need malware. They just need certutil.exe.</p><p>Living-off-the-Land Binaries — signed, trusted Microsoft tools repurposed for payload delivery and evasion — are documented extensively by the LOLBAS project, which tracks over 130 of them. If you’re building detection skills in a home SOC lab, emulating this behavior and then writing the SPL rules to catch it is one of the fastest ways to understand how real detection engineering works.</p><p>Here’s the workflow, end to end.</p><h4>Step 1: Get Windows telemetry into its own index</h4><p>By default, Splunk dumps everything into the main index. For a lab focused on Windows endpoint data — System, Security, PowerShell, and Sysmon logs — that&#39;s inefficient for retention and query performance. On the Universal Forwarder, point those inputs at a dedicated windows index inside inputs.conf, then restart the forwarder service.</p><p>There’s a gotcha here that trips up a lot of lab builders: assigning index = windows in inputs.conf doesn&#39;t create that index on the indexer. If it doesn&#39;t already exist, Splunk silently drops the events — no error, no fallback to main. You have to create it manually under <strong>Settings &gt; Indexes &gt; New Index</strong>, set a max size and frozen time period appropriate for a lab, and verify with:</p><pre>index=windows earliest=-24h | stats count by sourcetype</pre><p>If that comes back empty, go check the index actually exists before you assume your forwarder is broken.</p><h4>Step 2: Generate real telemetry with a LOLBin</h4><p>Certutil.exe is mapped to MITRE ATT&amp;CK techniques T1105 (Ingress Tool Transfer) and T1140 (Deobfuscate/Decode Files), and it’s a go-to example because the flags matter so much for detection. From a writable directory on the target host:</p><pre>certutil.exe -urlcache -split -f &quot;http://example.com&quot; test.txt</pre><p>-urlcache fetches the resource, -split chunks the download to dodge simple size-based inspection, and -f forces an overwrite. This activity should show up in both Sysmon Event ID 1 (Process Create) and PowerShell Script Block Logging (Event ID 4104) — assuming Sysmon is actually configured to log command lines, which the default config does not do.</p><p>Once you’ve got this working, the same pattern extends to other common download vectors: mshta.exe (T1218.005), regsvr32.exe (T1218.010), rundll32.exe (T1218.011), and powershell.exe (T1059.001). Each one gives you a different command-line signature to detect against.</p><h4>Step 3: Fix your Sysmon config before you fix your SPL</h4><p>Default Sysmon logging misses the fields you actually need. Deploy a community configuration — SwiftOnSecurity/sysmon-config or olafhartong/sysmon-modular are the standard choices — and make sure Event IDs 1 (Process Creation), 3 (Network Connection), 7 (Image Load), 11 (File Create), and 22 (DNS Query) are enabled at minimum. Install with sysmon.exe -accepteula -i sysmonconfig.xml and confirm events are flowing with Get-WinEvent.</p><h4>Step 4: Write a rule that isn’t garbage</h4><p>A broad search for certutil.exe will bury you in legitimate noise — AD CS, WSUS, and SCCM all use it for routine certificate operations. The fix is isolating the specific flag combinations associated with payload staging:</p><pre>index=windows<br>( sourcetype=XmlWinEventLog OR sourcetype=WinEventLog:Security )<br>certutil.exe<br>(&quot;-urlcache&quot; OR &quot;-split&quot; OR &quot;-f&quot; OR &quot;-decode&quot; OR &quot;-verifyctl&quot;)<br>| eval suspicious=case(<br> like(CommandLine, &quot;% -urlcache % -split % -f %&quot;), &quot;download_split_force&quot;,<br> like(CommandLine, &quot;% -decode %&quot;), &quot;decode_operation&quot;,<br> 1=1, &quot;other_flag&quot;<br>)<br>| stats count min(_time) as earliest max(_time) as latest by Computer User ProcessName CommandLine ParentImage suspicious<br>| convert ctime(earliest) ctime(latest)<br>| sort - latest</pre><p>This filters to the relevant sourcetypes, matches the flag combos most associated with staging activity, classifies the variant, and aggregates by host and user so you’re not drowning in duplicate alerts.</p><p>Even this will generate false positives from legitimate admin activity like certutil -verifyctl. Suppress them with an allowlist lookup built from a baseline week of clean activity:</p><pre>| lookup certutil_allowlist CommandLineHash OUTPUT approved<br>| where isnull(approved)</pre><h4>Step 5: Turn it into an alert, then prove it fires</h4><p>Save the refined query as a scheduled alert — not real-time, which burns search-head resources unnecessarily. Run it every 5 minutes against a -5m@m to @m window, trigger per-result, and throttle for an hour per Computer, CommandLine so a single actor doesn&#39;t flood your queue. Add email or webhook actions to a SOAR or Slack endpoint if you want enrichment.</p><p>To validate the pipeline, re-run a modified version of the certutil command on the target host and check <strong>Activity &gt; Triggered Alerts</strong> in Splunk Web within the schedule window. Drill into the result to confirm the raw event, extracted fields, and suspicious classification all line up.</p><p>From there, a timechart panel by suspicious category and a top-offending-hosts panel give you ongoing visibility without re-running searches manually.</p><h4>Why this matters beyond one binary</h4><p>The real value isn’t the certutil rule itself — it’s the repeatable loop: confirm the index exists, confirm telemetry arrives, emulate the technique, write a narrow rule, suppress the false positives, alert, validate, document. Map each rule to its MITRE ATT&amp;CK technique ID and you start building a coverage matrix instead of a pile of disconnected searches.</p><p>If you want the full walkthrough — including the Sysmon field reference, the complete LOLBin variant table, the dashboard panels, and the full detection engineering checklist — the complete guide is here: <a href="https://edge.3dynamix.uk/cybersecurity/soc-home-lab-how-to-emulate-lolbins-write-splunk-spl-rules/">https://edge.3dynamix.uk/cybersecurity/soc-home-lab-how-to-emulate-lolbins-write-splunk-spl-rules/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aba4923adbd3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Guessing Subnet Masks: The Only CIDR Reference You’ll Need in 2026]]></title>
            <link>https://medium.com/@ExploitTheEdge/stop-guessing-subnet-masks-the-only-cidr-reference-youll-need-in-2026-b0c846af9eed?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/b0c846af9eed</guid>
            <category><![CDATA[cidr]]></category>
            <category><![CDATA[ipv6]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[subnetting]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Fri, 02 Oct 2026 08:02:09 GMT</pubDate>
            <atom:updated>2026-10-02T08:02:09.103Z</atom:updated>
            <content:encoded><![CDATA[<p><em>From /24 LANs to /127 IPv6 links — practical calculations, Linux commands, and the RFC-backed rules that prevent silent network failures</em></p><p><em>Disclosure: this story was drafted with AI assistance from our own guide and checked against the sources cited there.</em></p><p>You’re staring at a terminal, trying to carve a /24 into department VLANs. Engineering needs 60 hosts, Sales needs 25, Management needs 10, plus two point-to-point links. You could guess. You could use an online calculator. Or you could apply the one rule that makes it deterministic: allocate largest subnets first, verify no overlaps, and move on.</p><p>That’s Variable Length Subnet Masking (VLSM) — the only addressing model that’s mattered since CIDR replaced classful addressing in 1993 (RFC 1519). Yet every few months, someone still asks whether a /26 mask is 255.255.255.192 or .224, or why their /31 point-to-point link works on Linux but not on that one vendor’s gear.</p><p>This is the reference I wish existed when I stopped memorizing tables and started understanding the bitwise logic underneath.</p><h4>What a Subnet Mask Actually Does</h4><p>A subnet mask is a bitmask. Contiguous 1 bits mark the network portion; 0 bits mark the host portion. Always contiguous — no gaps — because routing decisions are made by simple bitwise AND.</p><pre>IP address:   192.168.10.45   -&gt; 11000000.10101000.00001010.00101101<br>Mask /24:     255.255.255.0   -&gt; 11111111.11111111.11111111.00000000<br>Network:      192.168.10.0    -&gt; 11000000.10101000.00001010.00000000</pre><p>The network address is IP AND mask. The broadcast address is network OR (NOT mask). Usable IPv4 hosts = 2^(host bits) - 2 (all-zeros and all-ones reserved). This comes straight from RFC 791, RFC 950, and RFC 4632.</p><h4>CIDR Notation Replaced Classful Addressing Decades Ago</h4><p>Modern networks use VLSM — any prefix length from /0 to /32 (IPv4) or /0 to /128 (IPv6). The common ones:</p><p>| CIDR | Dotted Mask | Usable Hosts | Typical Use | | — — — | — — — — — — -| — — — — — — — | — — — — — — -| | /24 | 255.255.255.0 | 254 | Standard LAN | | /26 | 255.255.255.192 | 62 | Department / VLAN | | /30 | 255.255.255.252 | 2 | Point-to-point link | | /31 | 255.255.255.254 | 2* | P2P links (RFC 3021) | | /32 | 255.255.255.255 | 1 | Loopback / host route |</p><ul><li>/31 allows 2 usable addresses on point-to-point links per RFC 3021.</li></ul><h4>Quick Calculations Without Leaving the Terminal</h4><p><strong>ipcalc</strong> (Debian/Ubuntu/Fedora/Arch):</p><pre>ipcalc 192.168.10.45/26</pre><p>Output includes network, broadcast, host range, and wildcard mask.</p><p><strong>iproute2</strong> (standard on Linux):</p><pre>ip -brief address show<br>sudo ip address add 10.0.5.10/24 dev eth0</pre><p><strong>nmcli</strong> (NetworkManager):</p><pre>nmcli con mod &quot;Wired connection 1&quot; ipv4.addresses 192.168.20.5/24<br>nmcli con mod &quot;Wired connection 1&quot; ipv4.gateway 192.168.20.1<br>nmcli con mod &quot;Wired connection 1&quot; ipv4.dns &quot;1.1.1.1 9.9.9.9&quot;<br>nmcli con mod &quot;Wired connection 1&quot; ipv4.method manual<br>nmcli con up &quot;Wired connection 1&quot;</pre><p><strong>Python one-liner</strong> (no extra deps):</p><pre>python3 -c &quot;import ipaddress; n=ipaddress.ip_network(&#39;192.168.10.0/26&#39;); print(f&#39;Network: {n.network_address}\nBroadcast: {n.broadcast_address}\nUsable: {n.num_addresses-2}\nFirst: {list(n.hosts())[0]}\nLast: {list(n.hosts())[-1]}&#39;)&quot;</pre><h4>VLSM Example: Carving a /24 Into Departments</h4><p>You have 10.10.0.0/24 (256 addresses). Requirements:</p><ul><li>Engineering: 60 hosts</li><li>Sales: 25 hosts</li><li>Management: 10 hosts</li><li>Two P2P links</li></ul><p>Allocate largest first:</p><p>| Subnet | CIDR | Range | Use | | — — — — | — — — | — — — -| — — -| | 10.10.0.0/26 | /26 | .1–.62 | Engineering (62 usable) | | 10.10.0.64/27 | /27 | .65–.94 | Sales (30 usable) | | 10.10.0.96/28 | /28 | .97–.110 | Management (14 usable) | | 10.10.0.112/30 | /30 | .113–.114 | P2P link 1 | | 10.10.0.116/30 | /30 | .117–.118 | P2P link 2 | | 10.10.0.120/29 | /29 | .121–.126 | Future growth |</p><p>Verify no overlaps:</p><pre>python3 -c &quot;<br>import ipaddress<br>nets = [ipaddress.ip_network(n) for n in [<br>    &#39;10.10.0.0/26&#39;,&#39;10.10.0.64/27&#39;,&#39;10.10.0.96/28&#39;,<br>    &#39;10.10.0.112/30&#39;,&#39;10.10.0.116/30&#39;,&#39;10.10.0.120/29&#39;<br>]]<br>for i, a in enumerate(nets):<br>    for b in nets[i+1:]:<br>        if a.overlaps(b):<br>            print(f&#39;OVERLAP: {a} &lt;-&gt; {b}&#39;)<br>            exit(1)<br>print(&#39;No overlaps&#39;)<br>&quot;</pre><h4>IPv6 Subnetting: Different Rules, Same Idea</h4><p>IPv6 uses prefix length only — no dotted masks. Standard LAN prefix is <strong>/64</strong> (RFC 4291). The lower 64 bits are the Interface Identifier (often EUI-64 or random per RFC 7217).</p><p>| Prefix | Use Case | | — — — — | — — — — — | | /32–/48 | ISP allocation to site | | /48 | Typical site assignment (65k /64 subnets) | | /56 | Small business / home (256 /64 subnets) | | /64 | Single LAN / VLAN (required for SLAAC) | | /127 | Point-to-point links (RFC 6164) | | /128 | Loopback / host route |</p><p><strong>Key difference from IPv4:</strong> IPv6 reserves the all-zeros interface identifier for the subnet-router anycast address (RFC 4291 §2.6.1). A /64 therefore has 2⁶⁴ − 1 usable unicast addresses. No broadcast exists — IPv6 uses multicast (ff02::1).</p><p>IPv6 on Linux:</p><pre>ip -6 address show<br>sudo ip -6 address add 2001:db8:1000:1::10/64 dev eth0<br>python3 -c &quot;import ipaddress; n=ipaddress.ip_network(&#39;2001:db8:1000:1::10/64&#39;); print(f&#39;Network: {n.network_address}\nPrefix: {n.prefixlen}&#39;)&quot;</pre><h4>Common Pitfalls That Cause Silent Failures</h4><ol><li><strong>Non-contiguous masks</strong> — invalid. 255.255.0.255 breaks routing. The kernel rejects non-contiguous prefixes when adding addresses via iproute2.</li><li><strong>Using /32 or /128 on LAN interfaces</strong> — valid for loopback/host routes and anycast, but not for regular LAN interfaces where you need a subnet.</li><li><strong>Forgetting the -2</strong> for IPv4 usable hosts (except /31 and /32).</li><li><strong>Mixing prefix lengths in the same L2 segment</strong> — causes silent reachability failures.</li><li><strong>IPv6 /64 requirement for SLAAC</strong> — DHCPv6 or static can use other lengths, but SLAAC <em>requires</em> /64 (RFC 4291).</li></ol><h4>Cheat Sheet: Convert Between Notations</h4><pre># CIDR -&gt; dotted mask (IPv4)<br>python3 -c &quot;import ipaddress; print(ipaddress.IPv4Network(&#39;0.0.0.0/26&#39;).netmask)&quot;<br># 255.255.255.192<br><br># Dotted mask -&gt; CIDR<br>python3 -c &quot;import ipaddress; print(ipaddress.IPv4Network(&#39;192.168.1.0/255.255.255.192&#39;).prefixlen)&quot;<br># 26<br><br># Wildcard mask (for ACLs) = NOT subnet mask<br>python3 -c &quot;import ipaddress; m=ipaddress.IPv4Network(&#39;0.0.0.0/26&#39;).netmask; print(ipaddress.IPv4Address(int(m) ^ 0xffffffff))&quot;<br># 0.0.0.63</pre><h4>Before You Commit a Subnet Plan</h4><ul><li>[ ] All subnets use contiguous masks (CIDR)</li><li>[ ] No overlapping ranges (verify with ipaddress module)</li><li>[ ] IPv4 usable hosts = 2^(32-prefix) - 2 (except /31, /32)</li><li>[ ] IPv6 LANs are /64 for SLAAC compatibility</li><li>[ ] P2P links use /31 (IPv4) or /127 (IPv6)</li><li>[ ] Gateway address reserved in each subnet (typically .1 or ::1)</li><li>[ ] Documentation updated (NetBox, Nautobot, spreadsheet, or diagram)</li><li>[ ] Reverse DNS zones planned for each prefix</li></ul><p>The full reference with RFC citations, additional examples, and a printable cheat sheet lives at <a href="https://edge.3dynamix.uk/networking/subnet-masks-and-cidr-a-practical-2026-reference/">https://edge.3dynamix.uk/networking/subnet-masks-and-cidr-a-practical-2026-reference/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b0c846af9eed" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[N8n’s “quiet” RCE problem]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">why your automation server suddenly deserves production-grade paranoia</p><p class="medium-feed-link"><a href="https://medium.com/@ExploitTheEdge/n8ns-quiet-rce-problem-72ee2f0f34ce?source=rss-2724d5aaf00d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@ExploitTheEdge/n8ns-quiet-rce-problem-72ee2f0f34ce?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/72ee2f0f34ce</guid>
            <category><![CDATA[homelab]]></category>
            <category><![CDATA[rce]]></category>
            <category><![CDATA[n8n]]></category>
            <category><![CDATA[vulnerability]]></category>
            <category><![CDATA[servers]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Wed, 11 Feb 2026 09:08:29 GMT</pubDate>
            <atom:updated>2026-02-11T09:08:29.358Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Agentic AI is a new sysadmin headache]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">memory poisoning, tool poisoning, and prompt injection are getting real</p><p class="medium-feed-link"><a href="https://medium.com/@ExploitTheEdge/agentic-ai-is-a-new-sysadmin-headache-1a3822396cb3?source=rss-2724d5aaf00d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@ExploitTheEdge/agentic-ai-is-a-new-sysadmin-headache-1a3822396cb3?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/1a3822396cb3</guid>
            <category><![CDATA[data-poisoning]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[prompt-injection]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Wed, 11 Feb 2026 08:58:32 GMT</pubDate>
            <atom:updated>2026-02-11T08:58:32.012Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[12 Skills to Learn in 2025]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@ExploitTheEdge/12-skills-to-learn-in-2025-21086fa127c1?source=rss-2724d5aaf00d------2"><img src="https://cdn-images-1.medium.com/max/2600/0*-sLp-h4GOMQDJsyX" width="5184"></a></p><p class="medium-feed-snippet">To Improve Your Life</p><p class="medium-feed-link"><a href="https://medium.com/@ExploitTheEdge/12-skills-to-learn-in-2025-21086fa127c1?source=rss-2724d5aaf00d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@ExploitTheEdge/12-skills-to-learn-in-2025-21086fa127c1?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/21086fa127c1</guid>
            <category><![CDATA[growth]]></category>
            <category><![CDATA[learn]]></category>
            <category><![CDATA[skills]]></category>
            <category><![CDATA[development]]></category>
            <category><![CDATA[learning]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Tue, 15 Oct 2024 09:54:03 GMT</pubDate>
            <atom:updated>2024-10-15T09:54:03.055Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Understanding Optical Fibers]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@ExploitTheEdge/understanding-optical-fibers-e45cfc07b5ab?source=rss-2724d5aaf00d------2"><img src="https://cdn-images-1.medium.com/max/2600/0*vJLod00lwU1OZvQB" width="3417"></a></p><p class="medium-feed-snippet">Technology, Types, and Connectors</p><p class="medium-feed-link"><a href="https://medium.com/@ExploitTheEdge/understanding-optical-fibers-e45cfc07b5ab?source=rss-2724d5aaf00d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@ExploitTheEdge/understanding-optical-fibers-e45cfc07b5ab?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/e45cfc07b5ab</guid>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[light]]></category>
            <category><![CDATA[how-it-works]]></category>
            <category><![CDATA[optical-fiber]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Fri, 11 Oct 2024 10:15:16 GMT</pubDate>
            <atom:updated>2024-10-11T10:15:16.662Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[How To Automate Your Daily Tasks]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@ExploitTheEdge/how-to-automate-your-daily-tasks-3d08a7f57c3e?source=rss-2724d5aaf00d------2"><img src="https://cdn-images-1.medium.com/max/2600/0*3MjY3bP7JodABZeW" width="6720"></a></p><p class="medium-feed-snippet">With Windows Task Scheduler</p><p class="medium-feed-link"><a href="https://medium.com/@ExploitTheEdge/how-to-automate-your-daily-tasks-3d08a7f57c3e?source=rss-2724d5aaf00d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@ExploitTheEdge/how-to-automate-your-daily-tasks-3d08a7f57c3e?source=rss-2724d5aaf00d------2</link>
            <guid isPermaLink="false">https://medium.com/p/3d08a7f57c3e</guid>
            <category><![CDATA[windows]]></category>
            <category><![CDATA[computer-science]]></category>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[task-management]]></category>
            <category><![CDATA[information-technology]]></category>
            <dc:creator><![CDATA[Exploit The Edge]]></dc:creator>
            <pubDate>Wed, 09 Oct 2024 19:37:00 GMT</pubDate>
            <atom:updated>2024-10-09T19:37:00.994Z</atom:updated>
        </item>
    </channel>
</rss>