<?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 HackerMe on Medium]]></title>
        <description><![CDATA[Stories by HackerMe on Medium]]></description>
        <link>https://medium.com/@hackerme?source=rss-d1d602509c70------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*E98lW4M61j4gIFeh6C1M4g.jpeg</url>
            <title>Stories by HackerMe on Medium</title>
            <link>https://medium.com/@hackerme?source=rss-d1d602509c70------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 08:27:07 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@hackerme/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[Port 1433 Exposed: How Weak SQL Server Passwords Led to a Miner and a RAT — HackerMe Security]]></title>
            <link>https://medium.com/@hackerme/port-1433-exposed-how-weak-sql-server-passwords-led-to-a-miner-and-a-rat-hackerme-security-0dc12294ee82?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/0dc12294ee82</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[database]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 17:44:25 GMT</pubDate>
            <atom:updated>2026-11-13T18:30:01.309Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GhFqNler1e2twvyS.png" /></figure><p>Port 1433 is the door many website databases still leave on the public internet. A campaign described by FortiGuard Labs this week is about something more basic than stopping injection in the application: a Microsoft SQL Server left reachable from outside, with a weak administrator password.</p><p>In research published on 12 November 2026, FortiGuard researcher Xiaopeng Zhang says the lab found the campaign in early October. The malware scans for SQL Servers open on the SQL Server default port, 1433, then tries to log in. After a successful login it uses built-in SQL Server features to run commands on the Windows server, and installs a Monero miner and a remote-access trojan. Infosecurity Magazine notes that the report does not say how many servers were compromised or who is behind the campaign.</p><p>If your site or app sits on SQL Server, the useful question is not whether you are in the unnamed victim list. It is whether that SQL Server port is reachable from the internet, and whether the sa or admin password would survive a long password list.</p><h3>What Fortinet reported</h3><p>FortiGuard says the initial payload is a Go-based program that embeds the open-source Naabu port scanner. It retrieves a target address range from an attacker-controlled server, then looks for hosts with TCP port 1433 open. A companion brute-force tool then tries usernames such as sa and admin against a list of more than 2,700 passwords. CSO Online, reporting on the same research, uses the same figures.</p><p>After a login, the malware runs Transact-SQL (T-SQL) commands on the server. Those commands enable xp_cmdshell, a SQL Server feature that runs operating-system commands, then download or unpack a loader. Fortinet and Infosecurity both describe the loader pulling further malware from the attackers&#39; infrastructure.</p><p>One payload, named to look like a Microsoft Edge update, is an XMRig-based Monero miner that uses the compromised server’s CPU. Another, named after a Windows system process, was identified as a Gh0st RAT variant. Fortinet says its features include keylogging and data theft, with more than 30 control commands, and plugins for screen capture and remote desktop. The C2 server had already shut down before Fortinet’s analysis, so that part of the work was static.</p><p>The malware also tries to stay. Fortinet says the first payload creates a scheduled task that runs every 30 minutes, with a name that imitates a Microsoft Edge update. The downloader sets another task; the Gh0st RAT uses a registry Run key, a Windows service and a further scheduled task. Infosecurity also reports a copy placed in the Startup folder so it ran at boot, and says each compromised server was registered with the attackers’ backend.</p><h3>Why an exposed port 1433 matters to site owners</h3><p>Many websites and web apps sit on SQL Server databases. Some of those databases end up on a host or a cloud virtual machine with port 1433 open to the world, often because of a leftover firewall rule, a development shortcut or a “temporary” remote-admin path that was never closed. A weak or reused sa password then turns that exposure into a full server compromise. That is a database security failure as much as a firewall one. CSO quotes Sectigo&#39;s Jason Soroko on the same point: remove direct internet access to SQL Server, restrict connections to approved systems, and enforce strong, unique credentials.</p><p>This is the same class of failure as an open login form that never rate-limits guesses, only the target is the database instead of the site. Our <a href="https://security.hackerme.ai/brute-force-login-protection-what-to-configure-what-to-test-and-what-to-ask-your-developer/">brute-force login protection guide</a> is about closing that pattern on a website. If you later find a miner or a remote-access tool on the box, <a href="https://security.hackerme.ai/is-brute-force-malware-telling-the-technique-from-the-infection-and-what-to-do-next/">telling a brute-force technique from an infection</a> helps you decide what to rebuild and what to rotate. And if the database lives with a host, be clear about <a href="https://security.hackerme.ai/website-security-what-your-host-handles-and-what-you-still-need-to-check/">what your host handles and what you still need to check</a>, including which ports are public.</p><h3>What to check this week</h3><ol><li><strong>Keep port 1433 off the public internet.</strong> SQL Server should accept connections from the application server or a private network, not from every address on the internet. Ask your host or cloud provider which public ports are open, and close 1433 if it is among them.</li><li><strong>Lock down or disable the </strong><strong>sa account.</strong> Do not leave a default or shared administrator password in place. If you do not need sa, disable it and use named accounts with only the rights they need.</li><li><strong>Use strong, unique passwords.</strong> The password list in this campaign had more than 2,700 entries. A reused or obvious admin password is enough.</li><li><strong>Leave </strong><strong>xp_cmdshell disabled.</strong> Fortinet says the attackers turned it on after they logged in so they could run operating-system commands. Unless you have a documented reason to use it, keep it off, and watch for attempts to enable it.</li><li><strong>Look for leftover persistence if the server was ever exposed.</strong> Unexpected scheduled tasks that imitate a Microsoft Edge update, unfamiliar programs in the Startup folder, or a new Windows service deserve a closer look, and usually a restore from a clean backup.</li></ol><p><a href="https://hackerme.ai/">Free Website Security Scanner | HackerMe</a></p><p>A public look at security for a website can show what your website exposes from the outside, on a domain you own or have permission to test. It is not a substitute for a database-port and credential review, and it should not be used as a brute-force test against anyone else’s server. We build HackerMe.</p><h3>FAQ</h3><h3>How do I know if SQL Server is reachable from the internet?</h3><p>Ask your host or cloud firewall which public ports are open. Port 1433 is the SQL Server default port. If that port is reachable from addresses you do not trust, close it and allow only the application servers that need it.</p><h3>Should I disable the sa account?</h3><p>If you do not need it, yes. If you do, give it a strong unique password, restrict where it can log in from, and do not use it as the application’s everyday account.</p><h3>What is xp_cmdshell, and why does it matter?</h3><p>It is a built-in SQL Server feature that runs operating-system commands. Fortinet says this campaign enabled it after a successful login, then used it to drop a miner and a remote-access trojan. Keep it disabled unless you have a clear, temporary reason to turn it on.</p><p><em>Originally published at </em><a href="https://security.hackerme.ai/port-1433-exposed-sql-server-passwords/"><em>https://security.hackerme.ai</em></a><em> on November 13, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0dc12294ee82" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Prompt Injection Attacks in GitHub Actions: What the Codex Issue Teaches About AI Agents]]></title>
            <link>https://medium.com/@hackerme/prompt-injection-attacks-in-github-actions-what-the-codex-issue-teaches-about-ai-agents-6e1b08292361?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e1b08292361</guid>
            <category><![CDATA[github-actions]]></category>
            <category><![CDATA[information-security]]></category>
            <category><![CDATA[security]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 17:42:25 GMT</pubDate>
            <atom:updated>2026-11-13T18:06:57.718Z</atom:updated>
            <content:encoded><![CDATA[<h3>Prompt Injection Attacks in GitHub Actions: What the Codex Issue Teaches About AI Agents — HackerMe Security</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3qNkU0eVsP5lHhq0.png" /></figure><p>Prompt injection attacks are no longer only a chatbot problem. GitHub Actions security now includes a simpler question: if a workflow hands an AI agent the text of a GitHub issue, a comment or a pull-request description, who actually wrote that text?</p><p>On 11 November 2026, security firm Depi disclosed a path that started with a crafted GitHub issue in OpenAI’s Codex repository and could have reached the credentials needed to publish the official @openai/codex package on npm. That is the shape of an npm supply chain attack, even though coverage of Depi&#39;s write-up does not describe any malicious release reaching users. SecurityWeek, in a 13 November roundup, says OpenAI has fixed the issue and awarded an initial $1,000 bounty.</p><p>You do not need to use Codex for this to matter. The same pattern now shows up in agency and product repositories: an agent that labels issues, summarises pull requests or runs a step in CI, reading text that anyone can type into the repo.</p><h3>What the researchers reported</h3><p>Depi’s write-up, by Pierre Martin (Worty) and Garance de la Brosse (Madder), says the entry point was an issue-labeling workflow in the openai/codex repository. The workflow used a Codex action to read newly opened issues and suggest labels. It was set up so that anyone who could open an issue could have that issue&#39;s title and body processed by the agent. Both fields were passed straight into the agent&#39;s prompt. SecurityWeek describes the same starting point as a prompt injection in a GitHub Actions workflow.</p><p>From there, Depi describes a chain that could have given an attacker a path to publish a malicious version of @openai/codex. Writeble, reporting on the same disclosure on 12 November, says the chain was demonstrated against a test environment, and that the write-up does not describe any malicious release reaching users. Depi says it reported the chain to OpenAI through Bugcrowd with a full proof of concept, and that OpenAI &quot;deployed a fix right away.&quot; Depi and SecurityWeek both report an initial bounty of $1,000; Depi says the report is still under further review.</p><p>Depi also says @openai/codex is downloaded millions of times a month. That figure is Depi&#39;s, not an independent count, and this article does not add a download number of its own.</p><p>We are not walking through the exploit chain here. The useful fact for anyone who maintains a repository is the starting point: a workflow trusted text that strangers could write.</p><h3>Why prompt injection attacks matter to site owners</h3><p>Many site builders, agencies and product teams now let AI agents triage issues or run steps in GitHub Actions. That is convenient. It is also an AI agent security problem if the agent is treated as a trusted reviewer of untrusted input.</p><p>If an agent reads issues, comments or pull-request descriptions, the people who wrote those texts can try to give it instructions. Our <a href="https://security.hackerme.ai/ai-agent-security/">guide to AI agent security</a> is about that broader risk: an agent is only as trustworthy as the tools, credentials and inputs you give it.</p><p>The Codex case is also a reminder that software supply chain security can fail one step before a release job. A workflow that only “labels issues” still runs in an environment that may share tokens, caches or later jobs with the pipeline that publishes a package. Token hygiene after an incident, and the habit of rotating anything a workflow could have exposed, is the same discipline we described after the <a href="https://security.hackerme.ai/apify-breach-api-token-rotation/">Apify API token rotation</a>. For the APIs your site or app already exposes, the same idea applies: <a href="https://security.hackerme.ai/api-security-vulnerabilities/">API security vulnerabilities</a> often start with a trusted path that accepted more input than it should have.</p><h3>What to check this week</h3><ol><li><strong>List the workflows that read outside text.</strong> Search for jobs that run when an issue is opened, a comment is posted, or a pull request arrives, and that pass that text to an agent, a script or a prompt. If anyone can open an issue in the repo, treat that text as untrusted.</li><li><strong>Give those jobs the least privilege they need.</strong> Run issue-triage and labeler workflows with read-only tokens. Do not attach package-publish, release or write permissions to a job whose only task is to read someone else’s text.</li><li><strong>Keep release workflows away from shared caches.</strong> Privileged jobs that build or publish should not trust a cache that a less-trusted workflow can write. If a labeler job and a release job can share a cache, they are closer than they look.</li><li><strong>Pin and review the actions you call.</strong> Know which action versions run in those workflows, and who can change them.</li><li><strong>Decide what an agent is allowed to do.</strong> If it can run commands, open a network path or write to the repository, untrusted text is not “just data.”</li></ol><p>None of these steps is a substitute for OpenAI’s fix in Codex itself. They are the checks that still apply in your own repositories.</p><p>While you are reviewing what your own site exposes from the outside, a <a href="https://hackerme.ai/">HackerMe</a> can show outside-in findings on a domain you own or have permission to test. It is not a GitHub Actions auditor, and it will not scan your workflows. We build HackerMe.</p><h3>FAQ</h3><h3>Was a malicious Codex package published to npm?</h3><p>No source reviewed for this article describes a malicious @openai/codex release reaching users. Depi says OpenAI fixed the core bug. Writeble notes that the chain was demonstrated against a test environment.</p><h3>Do I need to patch Codex on my machine?</h3><p>Depi says OpenAI fixed the core bug that enabled the chain, so there is nothing for end users to patch on their machines. The takeaway is about how your own workflows treat untrusted text, not a client-side patch.</p><h3>What should I change in GitHub Actions if we use an AI agent?</h3><p>Do not pass issue titles, issue bodies, comments or pull-request descriptions to an agent in a job that has write, release or publish permissions. Use read-only tokens for those jobs, and keep privileged release workflows from trusting shared caches.</p><p><em>Originally published at </em><a href="https://security.hackerme.ai/prompt-injection-attacks-github-issue-codex/"><em>https://security.hackerme.ai</em></a><em> on November 13, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e1b08292361" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Web Hosting Ransomware: What Three Host Attacks Teach Site Owners About Backups]]></title>
            <link>https://medium.com/@hackerme/web-hosting-ransomware-what-three-host-attacks-teach-site-owners-about-backups-6f926ff237f4?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/6f926ff237f4</guid>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 15:33:08 GMT</pubDate>
            <atom:updated>2026-11-13T15:49:51.045Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*emX-PfMGcDQzN82p.png" /></figure><p>Web hosting ransomware used to be a story about one customer’s server. In the first half of November 2026 it became a story about the host itself. Three hosting providers, Beget and ESTT in Russia and ActiveCloud in Belarus, were hit within about ten days. Thousands of websites, mailboxes and virtual servers went dark, and at one of them, backups were lost too.</p><p>You do not need to host in that region for this to matter. The same question applies to every site owner: if your host is the one that gets hit, what do you still control?</p><h3>What happened at each host</h3><h3>Beget: a wiper from the management layer</h3><p>Beget’s own incident timeline says the attack began at about 02:00 Moscow time on 2 November, when attackers reached its management servers. Half an hour later they launched malware designed to delete data. Beget detected it at about 02:40 and disconnected all servers from the network at 02:55.</p><p>BleepingComputer, reporting on Beget’s statements, says all data in Beget’s cloud (VPS, S3 storage and Kubernetes) was lost, along with CDN settings and IP configurations. VPS binary backups were completely lost; VPS file backups partly survived and, by Beget’s 6 November assessment, would be available for only about half of VPS users. About 1,000 servers had to be checked by hand for hidden scripts before going back online. Beget said it found no sign the attacker tried to steal data; the attack looked aimed at destruction.</p><p>Shared hosting fared better: Beget’s timeline says more than 55,000 shared hosting websites had been restored from backups, but recovery took days.</p><h3>ESTT: ransomware in the virtualization clusters</h3><p>ESTT reported a serious incident on 8 November. According to its statement, quoted by CNews and Habr, preliminary findings pointed to ransomware that hit its virtualization clusters and cloud management systems. Customers’ dedicated physical servers were not affected. ESTT said it finished restoring its cloud infrastructure on 13 November.</p><h3>ActiveCloud: customer sites and email down for days</h3><p>ActiveCloud said its infrastructure came under attack early on 10 November, and later confirmed ransomware had encrypted part of its cloud. The Record reports that customer websites, corporate email and virtual servers were disrupted, along with the company’s own support channels. Local outlets reported that customers could not restore from backups themselves while recovery was under way, and that some virtual machines later failed to boot. ActiveCloud says it found no evidence customer data was stolen.</p><p>No one has publicly linked the three attacks or named who is behind them.</p><h3>Who is affected</h3><p>Directly, customers of these three providers. Indirectly, anyone whose website, store or app lives on a single provider and relies only on that provider’s backups. The pattern in all three cases is the same: the attackers went after the systems that manage the whole platform, not one customer’s site. When that layer falls, every site on it falls with it, and backups that the same layer can reach are at risk too.</p><h3>What to check this week</h3><ol><li><strong>Where your backups live.</strong> Ask your host whether backups are stored on separate infrastructure, and whether the same control panel or management system can delete them. Our <a href="https://security.hackerme.ai/website-backup-testing-restore-checklist/">website backup testing checklist</a> walks through the questions.</li><li><strong>An off-host copy you control.</strong> Keep at least one recent copy of your files and database outside your hosting provider, in an account with its own login and MFA.</li><li><strong>A real restore test.</strong> A backup you have never restored is a guess. Restore to a staging server once, and time it.</li><li><strong>DNS and domain access.</strong> If you can move your domain’s DNS quickly, you can point visitors to a backup copy or a status page while the host recovers.</li><li><strong>The split of responsibility.</strong> Know <a href="https://security.hackerme.ai/website-security-what-your-host-handles-and-what-you-still-need-to-check/">what your host handles and what you still need to check</a> yourself, and read our <a href="https://security.hackerme.ai/ransomware-preparedness-for-small-business-websites-defenses-isolation-and-verification/">ransomware preparedness guide for small business websites</a>.</li></ol><p>None of these steps stops an attack on your provider. They decide how long you are offline when one happens. While you are at it, a <a href="https://hackerme.ai/">check your website security with HackerMe</a> can show what your site exposes from the outside.</p><h3>The dark web angle</h3><p>Attacks on providers often start with access that was bought, not built. Initial-access brokers sell logins to admin panels, VPNs and management systems on underground forums, and one set of provider credentials can reach thousands of customer sites. None of the three companies has said how the attackers got in, so we are not claiming that happened here. It is the reason provider-level admin accounts deserve MFA and tight access, and the reason your backup should not depend on them.</p><h3>FAQ</h3><h3>Does my host’s backup protect me from ransomware?</h3><p>Only partly. If the attackers reach the host’s management layer, backups that layer can access can be encrypted or deleted too, as Beget’s lost VPS binary backups show. Keep your own copy outside the host.</p><h3>Was customer data stolen in these attacks?</h3><p>All three providers said they found no evidence of customer data theft. The damage reported was encryption, deletion and long outages.</p><h3>How often should I test a website restore?</h3><p>At least every few months, and after any major change to the site. Restore to a staging server and confirm the site and database actually work.</p><p><em>Originally published at </em><a href="https://security.hackerme.ai/web-hosting-ransomware-backups/"><em>https://security.hackerme.ai</em></a><em> on November 13, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6f926ff237f4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Apify Breach: How to Handle API Token Rotation After a Breach at a Tool Your Website Uses]]></title>
            <link>https://medium.com/@hackerme/apify-breach-how-to-handle-api-token-rotation-after-a-breach-at-a-tool-your-website-uses-b683cc075e03?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/b683cc075e03</guid>
            <category><![CDATA[apify]]></category>
            <category><![CDATA[security]]></category>
            <category><![CDATA[rest-api]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 13:39:45 GMT</pubDate>
            <atom:updated>2026-11-13T13:39:45.701Z</atom:updated>
            <content:encoded><![CDATA[<p>Many websites and online stores depend on tools that their owners never log into. A scraper collects competitor prices every night. An automation pulls leads into the CRM. A plugin fetches reviews or search results. Each of those tools holds an API token, and when the company behind the tool is breached, that token becomes your problem.</p><p>That is what happened this week with Apify.</p><h3>What happened at Apify</h3><p>Apify is a Czech web scraping and browser automation platform. According to BleepingComputer, it reports about 80,000 customers.</p><p>On its status page, Apify says that on 9 and 10 November 2026 an unauthorized third party gained access to parts of its internal systems. It detected the activity on 11 November at 12:13 UTC and cut off access. In a later update it said the attacker used a compromised credential to obtain a number of account credentials and accessed related Apify accounts, and that it has seen no further activity since 11 November.</p><p>As precautions, Apify:</p><ul><li>rotated various critical internal secrets;</li><li>revoked some OAuth connections to external integrations, including GitHub;</li><li>rotated the API tokens and Apify Proxy passwords of affected customers and reset active sessions;</li><li>told some customers to reset their two-factor authentication, because their 2FA codes may have been exposed, while saying passwords were safe.</li></ul><p>The investigation is still ongoing, and Apify says a post-mortem will follow. It has not said how many accounts were affected. Apify has since said it has notified customers whose personal information or stored data was accessed in the incident, and that anyone who has not received an email has not been identified as affected so far.</p><h3>Who is affected</h3><p>You may be affected even if you have never heard of Apify, because the person who set it up may have been a developer or an agency. Typical places an Apify token lives in a website business:</p><ul><li>environment variables on your hosting or in a serverless function;</li><li>automation tools such as n8n, Zapier or Make;</li><li>a WordPress or WooCommerce plugin that imports products, prices or reviews;</li><li>scheduled jobs that build SEO reports or lead lists;</li><li>GitHub repositories that build Apify Actors.</li></ul><p>If Apify rotated your token, the old one stopped working. Jobs that depend on it may now be failing quietly, and nobody notices until a report is empty or a price list is stale.</p><h3>What to check this week</h3><h3>1. Find every place the token is stored</h3><p>Ask your developer or agency for a list. Tokens are easy to copy into five places and forget four of them. A breach is the moment to make the list, not to guess.</p><h3>2. Replace it, then rebuild</h3><p>Create a new token in the Apify Console and update every place on the list. One developer who ran 33 Actors on Apify wrote on DEV Community that a secret environment variable only reaches a run through a new build, so Actors had to be rebuilt after the value changed. Test each job with a small run.</p><h3>3. Reconnect integrations deliberately</h3><p>Apify revoked GitHub and some other OAuth connections. Reconnect only the ones you still use, and give each the narrowest access it needs.</p><h3>4. Reset 2FA if you were asked</h3><p>If Apify’s email told you to reset two-factor authentication, do it in the Apify Console. Do not click links in look-alike emails; go to the console yourself. Breach news is a favourite hook for phishing.</p><h3>5. Check the bill</h3><p>Look for runs, usage or proxy traffic you did not start between 9 November and the day you rotated. Stolen tokens are often used to run up charges rather than to steal data.</p><h3>The dark-web angle: why tokens get stolen</h3><p>API tokens and session cookies are valuable because they skip the password and often skip 2FA as well. Access to developer and cloud accounts is commonly traded on criminal forums and Telegram channels, alongside logs from infostealer malware. A token that works is worth more than a password that needs a code. That is why the fastest safe response to any vendor breach is to treat every token from that vendor as burned and replace it, even before the vendor confirms your account was touched.</p><p>This is the same pattern we covered in our guides on <a href="https://security.hackerme.ai/api-security-vulnerabilities/">API security vulnerabilities</a>, on <a href="https://security.hackerme.ai/multifactor-authentication-for-website-owners-picking-methods-and-rolling-them-out-safely/">multifactor authentication for website owners</a>, and on <a href="https://security.hackerme.ai/ai-agent-security/">AI agent security</a>.</p><h3>What to ask your developer</h3><ul><li>Where is our Apify token stored, and has it been replaced everywhere?</li><li>Did any scheduled job fail after 11 November?</li><li>Do we use one token per service, so we can rotate one without breaking everything?</li></ul><p>If you want an outside view of how your own site looks to an attacker, you can run a <a href="https://hackerme.ai/">free website security scan</a> with HackerMe.</p><h3>FAQ</h3><h3>Was my data stolen in the Apify breach?</h3><p>Apify says it is still investigating and will contact affected users directly. An early customer notice found no indication that stored account data was affected, but Apify later said it has notified customers whose personal information or stored data was accessed. If you did not get an email, still check your integrations, because a rotated token can break jobs either way.</p><h3>Why did my Apify scraper or automation stop working?</h3><p>If Apify rotated your API token, the old one stopped working immediately. Replace it everywhere it is stored and rebuild any Actor that reads it from a secret environment variable.</p><h3>Should I rotate API tokens after any vendor breach?</h3><p>Yes. Replacing a token costs minutes; leaving a stolen one active can cost money and data. Keep one token per service so rotation is quick.</p><p>We build HackerMe.</p><p>Sources: Apify status page (https://status.apify.com/), BleepingComputer, DEV Community account by an Apify customer.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b683cc075e03" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[WordPress Mixed Content Fix: How to Find the HTTP File That’s Breaking HTTPS]]></title>
            <link>https://medium.com/@hackerme/wordpress-mixed-content-fix-how-to-find-the-http-file-thats-breaking-https-88b34b5b6f46?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/88b34b5b6f46</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[web-security]]></category>
            <category><![CDATA[https]]></category>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[website-development]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Thu, 12 Nov 2026 07:44:49 GMT</pubDate>
            <atom:updated>2026-11-12T07:44:49.606Z</atom:updated>
            <content:encoded><![CDATA[<p>If your WordPress site has a certificate and loads over <a href="https://,/">https://,</a> but the padlock is missing or a slider, form, or checkout step stopped working, mixed content is the usual suspect. This guide shows a site or store owner how to do a WordPress mixed content fix properly: find the exact http:// file, work out where WordPress is storing it, and change it at the source.</p><p>Mixed content means the page itself arrives over HTTPS, but something on it, such as an image, script, stylesheet, font, or iframe, is still requested over plain HTTP. Chrome’s DevTools documentation on <a href="https://developer.chrome.com/docs/devtools/security">the Security panel</a> describes these pages as only partially protected, because the HTTP parts can be read or changed in transit.</p><h3>Why it matters on a store, not just a blog</h3><p>MDN’s page on <a href="https://developer.mozilla.org/en-US/docs/Web/Security/Defenses/Mixed_content">mixed content</a> explains that browsers try to upgrade some requests, such as images, audio, and video, to HTTPS, and block the rest. For an owner, that means two different symptoms:</p><ul><li><strong>The page looks fine, but the padlock is downgraded.</strong> Usually an old image or media file.</li><li><strong>Something quietly breaks.</strong> A blocked script or stylesheet can take a menu, a chat widget, a payment button, or a page-builder layout down with it, and visitors just see a broken page.</li></ul><p>The second kind is the one that costs sales, and it often shows up only on one product template or one old landing page.</p><h3>Step 1: Find the exact http:// file</h3><p>Do not start with a plugin. Start by finding the URL.</p><ol><li>Open the page in a private window, so your logged-in admin view and cached files do not hide anything.</li><li>Open the browser’s developer tools and look at the Console. Google’s web.dev guide to <a href="https://web.dev/articles/fixing-mixed-content">fixing mixed content</a> notes that Chrome reports mixed content there as warnings and errors, naming the http:// URL.</li><li>In Chrome you can also open the Security panel. Its Non-secure origins section links to the matching requests in the Network panel.</li><li>Write each URL down next to the page you found it on.</li></ol><p>The same web.dev guide points out the catch: the console only shows problems for the page you are viewing, and clears when you navigate. So check the pages that matter one by one: home, a product page, cart, checkout, the contact form, and any promo landing page you still link to.</p><h3>Step 2: Work out where WordPress keeps that URL</h3><p>The http:// address tells you a lot about where it lives:</p><ul><li><strong>It is your own domain, under /wp-content/uploads/.</strong> The link is almost always inside post or page content, a page-builder block, or a widget, saved from before the site moved to HTTPS.</li><li><strong>It is your own domain, under /wp-content/themes/ or /wp-content/plugins/.</strong> The theme or plugin is building the URL itself, often from an old setting or a hard-coded path. Check the theme options first, then the plugin’s settings.</li><li><strong>It is another company’s domain.</strong> A tracking pixel, font, map, video, or review widget pasted in years ago. Usually it sits in a header or footer script box, a custom HTML widget, or a plugin that inserts code.</li></ul><h3>Step 3: Check the two address settings first</h3><p>Go to Settings, then General. The WordPress Address (URL) and Site Address (URL) should both begin with <a href="https://./">https://.</a> The WordPress handbook page on <a href="https://developer.wordpress.org/advanced-administration/security/https/">HTTPS for WordPress</a> covers this setup, including sites behind a proxy or load balancer.</p><p>Since WordPress 5.7, Site Health can also help. The core team’s note on <a href="https://make.wordpress.org/core/2021/02/22/improved-https-detection-and-migration-in-wordpress-5-7/">HTTPS detection and migration in WordPress 5.7</a> explains that Site Health can detect whether your hosting already supports HTTPS and offer a one-click switch. After a switch on a site with existing content, WordPress rewrites old http:// links to your own home URL on the fly as pages are shown.</p><p>That is useful, but know its limits. It covers links to your own site address, not third-party files. And it fixes the output, not the stored content, so the old links are still in the database.</p><h3>Step 4: The lasting mixed content fix in WordPress, with a backup first</h3><p>For your own domain, the lasting fix is to update the stored URLs from http:// to <a href="https://./">https://.</a> You have three common routes:</p><ul><li><strong>Edit by hand.</strong> Fine for a handful of posts or one widget.</li><li><strong>A search-and-replace plugin.</strong> Practical for many posts. Use a well-maintained one and read its instructions.</li><li><strong>WP-CLI, if your host offers it.</strong> The <a href="https://developer.wordpress.org/cli/commands/search-replace/">wp search-replace command</a> handles PHP serialized data, which matters because many themes and page builders store settings that way. It has a --dry-run option that reports what would change without saving, and its documentation shows how to skip a column such as guid.</li></ul><p>Whatever route you use, take a full backup first and make sure you can restore it. A database-wide replace is hard to undo, and a mistyped string can break a lot at once. The WordPress handbook page on <a href="https://developer.wordpress.org/advanced-administration/security/backup/">backups</a> covers what to include, and our <a href="https://security.hackerme.ai/website-backup-testing-restore-checklist/">website backup and restore checklist</a> walks through testing that a restore really works.</p><p>Replace the exact old address, such as http://www.yourshop.com, with the new one. Do not run a blanket swap of every &quot;http:&quot; in the database, which would also rewrite links to other sites that may not support HTTPS at all.</p><h3>Step 5: Fix the theme, plugin, and third-party URLs</h3><p>For files that come from your theme or a plugin:</p><ul><li>Look for a URL field in the theme options, customizer, or plugin settings that still says <a href="http://,/">http://,</a> and update it.</li><li>If a child theme or custom code hard-codes the address, ask your developer to fix it there.</li><li>Update the theme or plugin. If it still loads an http:// file in its current version, report it to the developer.</li></ul><p>For third-party files, the web.dev guide suggests checking whether the same file loads at its https:// address. If it does, update the snippet. If it does not, host the file yourself, find another provider, or remove the widget.</p><h3>Step 6: Clear caches and check again</h3><p>Once the links are fixed, clear every cache layer you have: a caching plugin, your host’s cache, and a CDN if you use one. Then repeat Step 1 in a fresh private window on the same list of pages. A fix that only shows up in the admin view has not reached your visitors yet.</p><h3>What about “just add a header”?</h3><p>You will see advice to add the Content-Security-Policy directive <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/upgrade-insecure-requests">upgrade-insecure-requests</a>. MDN explains that it tells the browser to request HTTP resources over HTTPS instead. It can be a reasonable safety net, but it only works if each file is actually available over HTTPS, and it does not fix the stored links. If you want to go further with CSP, our <a href="https://security.hackerme.ai/content-security-policy/">Content Security Policy guide</a> covers rolling it out without breaking the site.</p><p>Some plugins that “force SSL” work in a similar way: they rewrite links as pages are served. They can hide the symptom while the old links stay in your content, so treat them as a stopgap.</p><h3>What this check does and does not prove</h3><p>Clearing every mixed content warning on the pages you checked shows that those pages, at that moment, loaded all their files over HTTPS. It does not prove the site is secure. A clean padlock means the connection is encrypted. It says nothing about outdated plugins, weak admin passwords, or a malicious script that is itself served over HTTPS. Pages you did not open may still have problems.</p><p>Re-check after a theme change, a new plugin, a page-builder update, a CDN change, or a site migration. Those are the moments mixed content usually comes back. If your browser still says “Not Secure” after all this, our earlier guide on <a href="https://hackerme.substack.com/p/website-says-not-secure-how-to-fix">why a website says “Not Secure” and how to fix it</a> covers the HTTP redirect and certificate causes.</p><h3>A free outside-in scan, if you want a second look</h3><p>We build HackerMe, a free outside-in website security scanner owned by Agura. If you want a second look at HTTPS redirects, TLS settings, and security headers, you can <a href="https://brandingme.ai/app/r/_xKsnffGTOLEPOEJN7GAiw">run a free website security scan</a> of a domain you own or have explicit permission to test. The scan reports what it observed from outside on that run. It cannot see inside your WordPress admin or database, and it does not replace checking your key pages in a browser.</p><p>Find the file. Fix it where it is stored. Then check again on the pages your customers actually use.</p><p><em>Originally published on the </em><a href="https://hackerme.substack.com/p/wordpress-mixed-content-fix-how-to"><em>HackerMe Substack</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=88b34b5b6f46" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Website Says “Not Secure”? How to Fix It, and What to Check First]]></title>
            <link>https://medium.com/@hackerme/website-says-not-secure-how-to-fix-it-and-what-to-check-first-d04ca73f6e5e?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/d04ca73f6e5e</guid>
            <category><![CDATA[https]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[website-development]]></category>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[web-security]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Wed, 11 Nov 2026 11:47:37 GMT</pubDate>
            <atom:updated>2026-11-12T06:20:58.760Z</atom:updated>
            <content:encoded><![CDATA[<p>Most of the time the cause is a plain HTTP page, one leftover HTTP file, or a certificate problem. Here is how to tell which one you have.</p><p>If your website says “Not Secure” in the browser bar, the fix depends on which of three things is happening: the page is loading over plain HTTP, an HTTPS page is pulling in a file over HTTP, or the browser does not trust the certificate. This guide walks a site or store owner through how to fix each one, in the order that usually finds the cause fastest.</p><p>You do not need a security team for this. You need a private browser window, the browser’s developer tools, and access to your hosting or platform settings.</p><h3>First, read what the browser is actually telling you</h3><p>Click the icon to the left of the address. Chrome’s help page on <a href="https://support.google.com/chrome/answer/95617">checking whether a site’s connection is secure</a> separates a plain “Not secure” label from a red “Not secure” or “Dangerous” warning. The first usually means the connection is not private. The second is more serious, and Google advises visitors not to enter personal information on that page.</p><p>Then look at the address itself:</p><ul><li><strong>It starts with </strong><a href="http://./"><strong>http://.</strong></a> The page is not using HTTPS at all. Chrome has marked every HTTP page as “Not secure” since Chrome 68 in July 2018, as the Chromium team announced in <a href="https://blog.chromium.org/2018/02/a-secure-web-is-here-to-stay.html">A secure web is here to stay</a>.</li><li><strong>It starts with </strong><a href="https://,/"><strong>https://,</strong></a><strong> but the label still appears.</strong> Usually something on the page is loaded over HTTP (mixed content), or a form posts to an HTTP address.</li><li><strong>You get a full-page warning instead of the site.</strong> That is a certificate problem, not a label. Expired, issued for the wrong name, or missing part of its chain.</li></ul><p>Write down which one you see and on which URL. A homepage that looks fine says nothing about the product page, the cart, or a landing page from last year’s campaign.</p><h3>How to fix “Not Secure” when the page loads over HTTP</h3><p>If the address bar shows <a href="http://,/">http://,</a> the site either has no certificate or is not sending visitors to the HTTPS version.</p><ol><li><strong>Check that HTTPS works at all.</strong> Type the same address with https:// in a private window. If it loads cleanly, the certificate exists and the problem is the redirect. If it fails, turn on a certificate in your host or platform first. Most hosts and hosted store platforms offer one in their dashboard. <a href="https://letsencrypt.org/docs/faq/">Let’s Encrypt’s FAQ</a> explains how its free certificates work if your host uses them.</li><li><strong>Redirect HTTP to HTTPS on the server.</strong> Google’s guidance on <a href="https://developers.google.com/search/docs/crawling-indexing/site-move-with-url-changes">moving a site with URL changes</a>, which covers HTTP to HTTPS moves, recommends server-side permanent redirects such as 301 or 308, and redirecting straight to the final URL instead of chaining hops.</li><li><strong>Update the site’s own idea of its address.</strong> On WordPress, the WordPress Address and Site Address in the general settings should both begin with <a href="https://./">https://.</a> The WordPress developer handbook page on <a href="https://developer.wordpress.org/advanced-administration/security/https/">HTTPS for WordPress</a> covers the setup in more detail. If you use a CDN or proxy in front of the site, check that its SSL mode matches how the origin server is configured, or you can end up in a redirect loop.</li></ol><p>Once every page redirects to HTTPS, you can look at HSTS, which tells browsers to use HTTPS for your domain even if someone types <a href="http://./">http://.</a> Roll it out carefully. Our guide to <a href="https://security.hackerme.ai/hsts-for-website-owners-how-to-roll-out-strict-transport-security-without-locking-yourself-out/">HSTS for website owners</a> explains how to do that without locking yourself out of a subdomain that still needs HTTP.</p><h3>When HTTPS is on and the page still says “Not Secure”</h3><p>This is the case owners find most confusing, because the certificate looks fine. The usual cause is mixed content: the page itself arrives over HTTPS, but an image, script, stylesheet, font, iframe, or form target still uses an http:// address.</p><p>MDN’s page on <a href="https://developer.mozilla.org/en-US/docs/Web/Security/Defenses/Mixed_content">mixed content</a> explains that browsers try to upgrade some of these requests, such as images, audio, and video, to HTTPS and block the rest. In practice that means two different symptoms. Sometimes the page looks normal but the padlock is downgraded. Sometimes a script or embed is blocked, and a menu, slider, chat widget, or checkout step quietly stops working.</p><p>To find the file:</p><ol><li>Open the page in a private window.</li><li>Open the browser’s developer tools and look at the Console. Mixed content warnings name the exact http:// URL.</li><li>Reload with the Network tab open and filter for “http:” to catch anything the console missed.</li><li>Repeat on the pages that matter most to you: product pages, cart, checkout, contact forms, and any promo landing page.</li></ol><p>To fix it, change each reference at the source. Google’s web.dev guide to <a href="https://web.dev/articles/fixing-mixed-content">fixing mixed content</a> suggests checking whether the resource is available at the https:// version of the same URL, and if it is, updating the link in your code or content. If it is not available over HTTPS, host the file yourself or find another provider.</p><p>Common places for old http:// links: images pasted into old posts, theme settings, page-builder blocks, hard-coded footer scripts, and third-party embeds. On WordPress, a search-and-replace tool can update old URLs in the database. Take and test a backup first, because a database-wide replace is hard to undo.</p><p>The Content-Security-Policy directive <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/upgrade-insecure-requests">upgrade-insecure-requests</a>, documented on MDN, tells the browser to request HTTP resources over HTTPS. It can be a useful safety net, but it only works if each resource is actually available over HTTPS. It does not fix the source, so keep fixing the links.</p><h3>When the browser blocks the page with a certificate warning</h3><p>A full-page warning means the browser does not trust the certificate. Owner-side causes you can check:</p><ul><li><strong>Expired certificate.</strong> Look at the expiry date in the certificate viewer. If auto-renewal failed, find out why before you renew by hand, or it will happen again.</li><li><strong>Wrong name.</strong> A certificate for example.com does not automatically cover <a href="http://www.example.com/">www.example.com</a> or shop.example.com. Check every hostname you actually use.</li><li><strong>Incomplete chain.</strong> The server should send the intermediate certificates too. <a href="https://letsencrypt.org/docs/certificate-compatibility/">Let’s Encrypt’s certificate compatibility notes</a> are a reminder that older devices may trust certificate chains differently.</li></ul><p>The <a href="https://cheatsheetseries.owasp.org/cheatsheets/Transport_Layer_Security_Cheat_Sheet.html">OWASP Transport Layer Security Cheat Sheet</a> covers the server-side settings in depth, for you or your developer.</p><p>Do not tell customers to click through the warning. If a visitor’s device clock is wrong, or their network inspects traffic, they may see warnings your other visitors do not. That is worth knowing, but rule out the causes on your side first.</p><h3>What this check does and does not prove</h3><p>Walking these steps shows that the pages you opened, at that moment, loaded over HTTPS with a trusted certificate and without mixed content. It does not prove the site is secure. A padlock means the connection is encrypted. It says nothing about outdated plugins, weak admin passwords, or a malicious script that is itself served over HTTPS. If you want the bigger picture of what HTTPS does and does not cover, our post on <a href="https://security.hackerme.ai/www-or-https-which-is-safer/">whether www or HTTPS is safer</a> explains it.</p><p>Re-check after any theme change, new app or plugin, CDN change, or certificate renewal. Those are the moments a “Not Secure” label usually comes back.</p><h3>A free outside-in scan, if you want a second look</h3><p>We build HackerMe, a free outside-in website security scanner owned by Agura. If you want a second look at TLS settings, HTTPS redirects, and security headers, you can <a href="https://brandingme.ai/app/r/fG7rJ9HWypVTNvv6Q_mUPg">run a free website security scan</a> of a domain you own or have explicit permission to test. The scan reports what it observed from outside on that run. It cannot see inside your server or admin panel, and it does not replace checking your key pages in a browser.</p><p>Read the browser first. Fix the cause it points to. Then check again on the pages your customers actually use.</p><p><em>Originally published on the </em><a href="https://hackerme.substack.com/p/website-says-not-secure-how-to-fix"><em>HackerMe Substack</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d04ca73f6e5e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Will Your TLS Certificate Renew—and Reach Your Visitors?]]></title>
            <link>https://medium.com/@hackerme/will-your-tls-certificate-renew-and-reach-your-visitors-cc891ca72606?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/cc891ca72606</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[sysadmin]]></category>
            <category><![CDATA[ssl]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[web-development]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Mon, 09 Nov 2026 15:10:00 GMT</pubDate>
            <atom:updated>2026-11-09T19:53:30.740Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Subtitle: Six checks — expiry and renewal margin, name and chain, every hostname you serve, protocol versions, renewal that reaches the running server, and an outside monitor.</em></p><p><em>A browser security warning can stop customers from reaching your site. The checks below are for anyone who keeps a site online — whether you run the servers or own the business.</em></p><p>Published by the HackerMe team (Agura). The checks are vendor-neutral. Each one is labeled &quot;safe on any public site&quot; (it makes an ordinary HTTPS connection, the same as any visitor) or &quot;maintainer check&quot; (it needs access to the server or the certificate tooling).</p><p>You’ve configured automatic certificate renewal. How do you know it worked?</p><p>It matters more each year. Let’s Encrypt <a href="https://letsencrypt.org/2025/01/22/ending-expiration-emails/">stopped sending expiration emails on June 4, 2025</a>, so its users no longer get a reminder when renewal silently fails. And the CA/Browser Forum’s <a href="https://cabforum.org/2025/04/11/ballot-sc081v3-introduce-schedule-of-reducing-validity-and-data-reuse-periods/">Ballot SC-081</a> reduces the maximum lifetime of public TLS certificates from 398 days to 47 days, in steps from March 2026 to March 2029. Renewal has to be checked, not assumed.</p><p>Replace example.com with a domain you own or have explicit permission to test. Commands shown use OpenSSL 3.x syntax.</p><h3>1. Expiry date and renewal margin (safe on any public site)</h3><pre>openssl s_client -connect example.com:443 -servername example.com &lt;/dev/null 2&gt;/dev/null \<br>  | openssl x509 -noout -subject -issuer -enddate</pre><p>-servername sends SNI, so a shared server returns the right certificate.</p><p>To check remaining lifetime without relying on date (GNU and BSD differ), use -checkend. The threshold below is an <strong>example</strong>—set THRESHOLD_DAYS to match your renewal window:</p><pre>THRESHOLD_DAYS=21   # example; adjust to your renewal window<br>SECONDS_LEFT=$((THRESHOLD_DAYS * 86400))<br>if openssl s_client -connect example.com:443 -servername example.com &lt;/dev/null 2&gt;/dev/null \<br>  | openssl x509 -noout -checkend &quot;$SECONDS_LEFT&quot;; then<br>  echo &quot;More than ${THRESHOLD_DAYS} days left&quot;<br>else<br>  echo &quot;Expires within ${THRESHOLD_DAYS} days (or no certificate)&quot;<br>fi</pre><p>-checkend exits 0 when the certificate will <strong>not</strong> expire within the given number of seconds, and non-zero otherwise.</p><ul><li><strong>Expected:</strong> a notAfter date, and a comfortable margin for your chosen threshold.</li><li><strong>Finding:</strong> an expiry inside your renewal window.</li><li><strong>Doesn’t prove:</strong> that renewal is configured. A valid date today says nothing about tomorrow.</li></ul><h3>2. Name match and chain (safe on any public site)</h3><pre>openssl s_client -connect example.com:443 -servername example.com \<br>  -verify_hostname example.com &lt;/dev/null 2&gt;&amp;1 | grep &#39;Verify return code&#39;</pre><ul><li><strong>Expected:</strong> Verify return code: 0 (ok).</li><li><strong>Finding:</strong> 62 (hostname mismatch), 10 (certificate has expired), or 21 (unable to verify the first certificate). Code 21 usually means the server does not send its intermediate certificate. Some browsers fill the gap, so the site can look fine in a browser and still fail for API clients.</li><li><strong>Doesn’t prove:</strong> trust everywhere. It reflects your machine’s trust store.</li></ul><h3>3. Every name you serve (safe on any public site, plus maintainer check)</h3><p>Certificate Transparency logs record publicly trusted certificates as they are issued (<a href="https://developer.mozilla.org/en-US/docs/Web/Security/Defenses/Certificate_Transparency">MDN: Certificate Transparency</a>). Searching them finds forgotten hostnames:</p><pre>curl -fsS &#39;https://crt.sh/?q=%25.example.com&amp;output=json&#39; | jq -r &#39;.[].name_value&#39; | sort -u</pre><p>Run checks 1 and 2 against each name that still resolves.</p><ul><li><strong>Finding:</strong> a live hostname nobody owns, or one served from a separate load balancer with a manually uploaded certificate.</li><li><strong>Doesn’t prove:</strong> a complete inventory. Private certificates are not logged, and crt.sh is a community service that can be slow or return unrelated names.</li></ul><h3>4. Protocol versions (safe on any public site)</h3><p><a href="https://www.rfc-editor.org/rfc/rfc8996">RFC 8996</a> deprecates TLS 1.0 and 1.1, and the <a href="https://cheatsheetseries.owasp.org/cheatsheets/Transport_Layer_Security_Cheat_Sheet.html">OWASP TLS Cheat Sheet</a> says to prefer TLS 1.3, allow TLS 1.2 for compatibility, and disable the older versions.</p><pre>for v in tls1 tls1_1 tls1_2 tls1_3; do<br>  printf &#39;%s: &#39; &quot;$v&quot;<br>  openssl s_client -connect example.com:443 -servername example.com -&quot;$v&quot; \<br>    -cipher &#39;DEFAULT:@SECLEVEL=0&#39; &lt;/dev/null 2&gt;/dev/null | grep -m1 &#39;^New,&#39;<br>done</pre><ul><li><strong>Expected:</strong> a cipher for tls1_2 and tls1_3. For tls1 and tls1_1, (NONE) is consistent with the server rejecting those versions—but (NONE) alone does <strong>not</strong> prove the server blocked them; your local OpenSSL build may refuse the old protocol even when the server would accept it.</li><li><strong>Careful:</strong> recent OpenSSL builds refuse old protocols themselves. @SECLEVEL=0 relaxes that, but confirm a surprising result with a second tool.</li><li><strong>Doesn’t prove:</strong> a good cipher configuration. For that, compare your server against a maintained template such as the <a href="https://ssl-config.mozilla.org/">Mozilla SSL Configuration Generator</a>.</li></ul><h3>5. Renewal reaches the running server (maintainer check)</h3><p>A successful renewal on disk is not the same as visitors receiving the new certificate. Separate three steps. With Certbot (<a href="https://eff-certbot.readthedocs.io/en/stable/using.html#testing">testing flags</a>):</p><p><strong>(a) Test renewal</strong> against the staging server (no certificates saved to disk):</p><pre>certbot renew --dry-run</pre><p>By default, --dry-run runs pre/post hooks but <strong>does not</strong> run deploy hooks.</p><p><strong>(b) Test that the deploy hook runs.</strong> Add --run-deploy-hooks so a dry run also executes deploy hooks (<a href="https://eff-certbot.readthedocs.io/en/stable/using.html#testing">Certbot </a><a href="https://eff-certbot.readthedocs.io/en/stable/using.html#testing">--run-deploy-hooks</a>). The hooks use the <strong>current active</strong> certificate, not the temporary test certificate from the dry run—so this confirms the hook fires (for example, reloads the web server), not that a brand-new cert was loaded:</p><pre>certbot renew --dry-run --run-deploy-hooks</pre><p><strong>© Verify the certificate actually served.</strong> After a real renewal (or after you expect the new file to be live), compare the on-disk certificate with what the TLS terminator presents:</p><pre>openssl x509 -in /etc/letsencrypt/live/example.com/cert.pem -noout -serial -enddate<br>openssl s_client -connect example.com:443 -servername example.com &lt;/dev/null 2&gt;/dev/null \<br>  | openssl x509 -noout -serial -enddate</pre><ul><li><strong>Expected:</strong> the same serial in both <strong>when</strong> you are checking the component that terminates TLS and reads that file (for example, the origin web server).</li><li><strong>Caveat:</strong> behind a CDN or load balancer, the on-disk origin certificate and the public certificate can differ <strong>intentionally</strong>. Compare only against the device that should be serving the file you are reading; do not treat a mismatch through a CDN edge as proof of a failed reload.</li><li><strong>Finding (direct terminator):</strong> a newer certificate on disk than on the wire — renewal worked, delivery did not.</li></ul><h3>6. Monitor from outside (maintainer check to set up)</h3><p>Put your hostnames in hosts.txt, one per line, and run this daily from a machine that is not the web server. -verify_return_error makes OpenSSL abort with a non-zero exit when verification fails (<a href="https://docs.openssl.org/3.5/man1/openssl-s_client/">OpenSSL s_client</a>). The 21-day window is an <strong>example</strong>—adjust THRESHOLD_DAYS to your renewal policy:</p><pre>#!/bin/sh<br>THRESHOLD_DAYS=21   # example; adjust to your renewal window<br>while read -r h; do<br>  [ -n &quot;$h&quot; ] || continue<br>  out=$(timeout 15 openssl s_client -connect &quot;$h:443&quot; -servername &quot;$h&quot; \<br>        -verify_hostname &quot;$h&quot; -verify_return_error &lt;/dev/null 2&gt;/dev/null)<br>  rc=$?<br>  if [ &quot;$rc&quot; -ne 0 ]; then<br>    echo &quot;ALERT $h: no connection or verification failed (exit $rc)&quot;<br>  elif ! printf &#39;%s\n&#39; &quot;$out&quot; | openssl x509 -noout -checkend $((THRESHOLD_DAYS * 86400)) &gt;/dev/null 2&gt;&amp;1; then<br>    echo &quot;ALERT $h: no certificate, or it expires within ${THRESHOLD_DAYS} days&quot;<br>  fi<br>done &lt; hosts.txt</pre><p>-checkend fails if the certificate expires within the given number of seconds. Send every ALERT line to a person, not just a log.</p><h3>Illustration: the certificate that renewed but never arrived</h3><p><em>This is a constructed example to show the process, not a check run on any customer site.</em></p><p>A team runs Certbot on a schedule, and the renewal log shows success. The outside monitor (check 6) still reports that www.example.com expires within their chosen window (21 days in this example):</p><ol><li>Check 1 confirms the served certificate has nine days left.</li><li>Check 5 shows a newer serial on disk than on the wire (compared at the TLS terminator). Renewal worked, delivery did not.</li><li>The cause: the web server reads the certificate only at start, and nothing reloaded it after renewal. certbot renew --dry-run alone would not have caught that—deploy hooks do not run on a dry run unless you add --run-deploy-hooks.</li><li>The fix: a deploy hook that reloads the server, then check 5 again until both serials match at the terminator.</li><li>Check 3 also finds api.example.com on a separate load balancer with a manually uploaded certificate. It gets an owner and a date.</li></ol><p>Without the outside check, the first warning would have been a browser error.</p><h3>Make it routine</h3><p>Run checks 1 and 2 after every deployment and certificate change, check 3 monthly, and check 6 daily. A clean result means the certificates you tested were valid when you tested them. It does not mean the site is secure.</p><p>Disclosure: we are the team behind HackerMe, a free automated website security scanner: <a href="https://brandingme.ai/app/r/6xMIXpfJLCYHdS8JhXYxfg">https://brandingme.ai/app/r/6xMIXpfJLCYHdS8JhXYxfg</a></p><p>The checks above work without it. What do you monitor that we missed? We would like to hear it in the comments.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cc891ca72606" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can Another Site Put Your Page in a Frame?]]></title>
            <link>https://medium.com/@hackerme/can-another-site-put-your-page-in-a-frame-2d8beb911a87?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/2d8beb911a87</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[web-security]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Sat, 07 Nov 2026 20:23:00 GMT</pubDate>
            <atom:updated>2026-11-12T06:14:17.498Z</atom:updated>
            <content:encoded><![CDATA[<p>How to review X-Frame-Options and CSP frame-ancestors, starting from a settings page that a homepage check would miss.</p><p>Here is a concrete example, not a report from a customer site. A visitor thinks they are dismissing a cookie banner. The click actually lands on your real settings page, loaded in an invisible iframe on another site, and turns off two-factor authentication. That is clickjacking: the visitor intended one thing and confirmed another.</p><p>A homepage check can hide exactly that gap:</p><pre>$ curl -s -o /dev/null -D - https://example.com/ | grep -i -E &#39;^(HTTP/|x-frame-options|content-security-policy)&#39;<br>HTTP/2 200<br>x-frame-options: SAMEORIGIN</pre><pre>$ curl -s -o /dev/null -D - <a href="https://example.com/settings">https://example.com/settings</a> | grep -i -E &#39;^(HTTP/|location|x-frame-options|content-security-policy)&#39;<br>HTTP/2 200</pre><p>Both responses are 200 pages, not redirects. The homepage restricts framing; /settings sends no framing policy. Browsers will let another site embed the settings document. Framing defenses do not replace authentication or CSRF controls; they only limit who may embed the page.</p><p>The checks below read headers every visitor already gets, except the local iframe test, which you should run only on a system you own or have explicit permission to test. Background: the <a href="https://cheatsheetseries.owasp.org/cheatsheets/Clickjacking_Defense_Cheat_Sheet.html">OWASP Clickjacking Defense Cheat Sheet</a>, <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options">MDN on X-Frame-Options</a>, <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/frame-ancestors">MDN on CSP frame-ancestors</a> and the <a href="https://wstg.owasp.org/v4.2/4-Web_Application_Security_Testing/11-Client-side_Testing/09-Testing-for-Clickjacking/">OWASP testing guide for clickjacking</a>.</p><h3>Read the framing headers on a real GET</h3><pre>curl -s -D headers.txt -o body.html https://example.com/account<br>grep -i -E &#39;^(HTTP/|location|x-frame-options|content-security-policy):&#39; headers.txt</pre><p>Use a normal GET. curl -I sends HEAD, and some servers answer HEAD with different headers than a browser&#39;s GET (<a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Methods/HEAD">MDN: HEAD</a>). Confirm what you got: the real page, a login redirect, or an error. An unauthenticated request to /account often returns the login redirect, and those headers describe the redirect, not the account page.</p><p>You want either X-Frame-Options set to DENY or SAMEORIGIN, or a Content-Security-Policy that includes frame-ancestors (for example &#39;none&#39; or &#39;self&#39;), or both. If neither header restricts framing, add one. Browsers allow other sites to embed the document when both are missing (<a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options">MDN: X-Frame-Options</a>).</p><p>A missing header is the gap to fix. Whether someone can then make a user perform an action still depends on the page having a state-changing control, the user being signed in inside a frame, and any confirmation step.</p><h3>The policy has to be an HTTP header</h3><p>X-Frame-Options and CSP frame-ancestors are enforced only as HTTP response headers. A &lt;meta http-equiv=&quot;X-Frame-Options&quot; ...&gt; tag has no effect, and frame-ancestors is not supported in a &lt;meta&gt; CSP (<a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options">MDN: X-Frame-Options</a>, <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/frame-ancestors">MDN: frame-ancestors</a>). If you find the restriction only in the HTML body and not in headers.txt, move it onto the response headers. A CDN or reverse proxy may also strip or rewrite them (<a href="https://cheatsheetseries.owasp.org/cheatsheets/Clickjacking_Defense_Cheat_Sheet.html">OWASP Clickjacking Defense Cheat Sheet</a>).</p><h3>Read the value carefully</h3><p>What you seeWhat it usually meansX-Frame-Options: DENYNo framing, including same-origin.X-Frame-Options: SAMEORIGINFraming only when every ancestor shares the page&#39;s origin (<a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options">MDN</a>).X-Frame-Options: ALLOW-FROM ...Obsolete. Modern browsers ignore the header entirely when they see this directive (<a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options">MDN</a>). Prefer CSP frame-ancestors.frame-ancestors &#39;none&#39;Similar intent to DENY.frame-ancestors &#39;self&#39;Similar intent to same-origin framing for this directive.frame-ancestors &#39;self&#39; https://partner.exampleAn explicit allow list. Review every entry.</p><p>CSP frame-ancestors does not fall back to default-src. A policy that sets default-src &#39;none&#39; still allows embedding by anyone unless frame-ancestors is set (<a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/frame-ancestors">MDN: frame-ancestors</a>). Flag ALLOW-FROM, an empty restriction, or a frame-ancestors list with origins you do not recognize, and replace them.</p><h3>Check the pages that change state, not only the homepage</h3><p>Framing risk is highest on pages where a single click changes something: approve a transfer, change an email address, disable MFA, confirm a delete. Repeat the GET above on /login, /account, /settings, /checkout and /admin. If a sensitive path answers with a redirect to login, check the page itself with a signed-in test account. Still verify error pages and mobile routes; the <a href="https://wstg.owasp.org/v4.2/4-Web_Application_Security_Testing/11-Client-side_Testing/09-Testing-for-Clickjacking/">OWASP testing guide</a> notes that alternate or mobile variants are sometimes left without the same controls.</p><h3>Load your own page in a local test frame</h3><p>Save a local HTML file that contains only this iframe, pointed at a page you own or are authorized to test:</p><pre>&lt;iframe src=&quot;https://example.com/account&quot; width=&quot;800&quot; height=&quot;600&quot;&gt;&lt;/iframe&gt;</pre><p>If your real page renders inside the frame, add a framing header on that response. A blank frame is not enough: the URL may have failed to load, redirected, or errored. Open DevTools and confirm the browser refused the frame because of X-Frame-Options or frame-ancestors.</p><h3>Make the header survive deployment</h3><p>With nginx’s default inheritance, add_header directives from a parent block are inherited by a location only when that location defines no add_header of its own. One extra add_header inside a location replaces the inherited set (<a href="https://nginx.org/en/docs/http/ngx_http_headers_module.html">nginx headers module</a>). Inheritance can be changed with add_header_inherit starting in nginx 1.29.3. On older builds, put shared security headers in an included snippet and include that snippet in every block that sets any add_header. Prefer always when you need the header on non-success status codes.</p><p>Cookies marked SameSite=Lax or Strict are withheld from many cross-site iframe requests, which can reduce authenticated clickjacking. SameSite alone does not block framing. Simple if (top !== self) top.location = ... frame-buster scripts have well-documented bypasses in the same OWASP materials. Prefer HTTP framing headers first.</p><p>A framing header on the pages you checked stops those documents from being embedded. It does not cover UI-redress that uses a separate top-level window, which needs action confirmation and server-side authorization instead (<a href="https://cheatsheetseries.owasp.org/cheatsheets/Clickjacking_Defense_Cheat_Sheet.html">OWASP Clickjacking Defense Cheat Sheet</a>).</p><p>This piece was first published on Dev.to and is republished here with that original as the canonical URL: <a href="https://dev.to/hackerme/can-another-site-put-your-page-in-a-frame-6-clickjacking-checks-and-what-a-pass-doesnt-prove-5nn">https://dev.to/hackerme/can-another-site-put-your-page-in-a-frame-6-clickjacking-checks-and-what-a-pass-doesnt-prove-5nn</a></p><p>The HackerMe team wrote this. We build a free scanner you can point at response headers on a domain you own or have explicit permission to test: <a href="https://brandingme.ai/app/r/Chw4NoNBy6COfEq1iQGDow">HackerMe</a>.</p><h3>Sources</h3><ul><li>OWASP Clickjacking Defense Cheat Sheet: <a href="https://cheatsheetseries.owasp.org/cheatsheets/Clickjacking_Defense_Cheat_Sheet.html">https://cheatsheetseries.owasp.org/cheatsheets/Clickjacking_Defense_Cheat_Sheet.html</a></li><li>MDN, X-Frame-Options: <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options">https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/X-Frame-Options</a></li><li>MDN, CSP frame-ancestors: <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/frame-ancestors">https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/frame-ancestors</a></li><li>OWASP WSTG, Testing for Clickjacking: <a href="https://wstg.owasp.org/v4.2/4-Web_Application_Security_Testing/11-Client-side_Testing/09-Testing-for-Clickjacking/">https://wstg.owasp.org/v4.2/4-Web_Application_Security_Testing/11-Client-side_Testing/09-Testing-for-Clickjacking/</a></li><li>MDN, HEAD: <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Methods/HEAD">https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Methods/HEAD</a></li><li>nginx headers module: <a href="https://nginx.org/en/docs/http/ngx_http_headers_module.html">https://nginx.org/en/docs/http/ngx_http_headers_module.html</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2d8beb911a87" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Missing Content-Security-Policy: What It Means and What to Check Next]]></title>
            <link>https://medium.com/@hackerme/missing-content-security-policy-what-it-means-and-what-to-check-next-7ec4aa273630?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/7ec4aa273630</guid>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[web-security]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 13:29:09 GMT</pubDate>
            <atom:updated>2026-11-12T06:17:41.609Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0pGyt4qdTqQooftQlwdl2Q.png" /></figure><p>A plain-language check for the CSP header, Report-Only vs enforce, and a short story of how CSP can stop an injected script.</p><p>If you run a small site or a marketing site in front of a product, you may have never typed “Content-Security-Policy” into a config file. Automated reports still call out a missing CSP header. That can sound like the site is broken. The finding alone does not mean the site was hacked. It means the page is missing a browser rule that limits where scripts and other resources may load from.</p><p>Think of CSP as instructions to the visitor’s browser: which script hosts are allowed, whether inline script may run, whether the page can be framed. <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/CSP">MDN’s CSP guide</a> puts XSS mitigation first. <a href="https://cheatsheetseries.owasp.org/cheatsheets/Content_Security_Policy_Cheat_Sheet.html">OWASP</a> stresses that you still need to write the application carefully; CSP is a second layer.</p><h3>A short scenario</h3><p>A comment field reflects unsanitized input into the page. An attacker leaves markup that loads https://evil.example/x.js (or an inline script). Without CSP—or with a policy that still allows almost any script—the browser may run that code for every visitor. A suitable enforced CSP that blocks inline scripts and limits script sources to trusted origins (or uses nonces/hashes) can stop the browser from executing that injection even when the markup still reaches the page. CSP does not fix the injection bug; it narrows what the browser will run.</p><h3>A five-minute outside check</h3><p>From a terminal (follow redirects; test the final URL):</p><pre>curl -sL -D headers.txt -o page.html -w &#39;final_url=%{url_effective}\n&#39; https://your-domain.com/<br>head -1 headers.txt<br>grep -i &#39;^content-security-policy:&#39; headers.txt<br>grep -i &#39;^content-security-policy-report-only:&#39; headers.txt</pre><p>What you want eventually is a Content-Security-Policy: line on the HTTPS page people actually use.</p><ul><li>No Content-Security-Policy: line: no enforced policy was found in the response headers; also check the HTML for a meta CSP tag.</li><li>Only Content-Security-Policy-Report-Only: the browser is not blocking yet — that mode is for learning. Review what would be blocked; Report-Only itself does not break the site.</li><li>Prefer the HTTP header over a meta tag when you need framing controls or reporting.</li></ul><p>Repeat the check on /login or another page that holds a session. Homepages and app pages often disagree.</p><h3>“We have a CSP” can still be a soft CSP</h3><p>When a header exists, open it. If scripts are allowed from almost anywhere, or inline script is still allowed without a nonce or hash, much of the XSS benefit is gone. A missing object-src line alone is not proof of weakness — it can fall back to the default rule.</p><p>You do not need a perfect policy on day one. You do need to know whether you have none, report-only, or a policy that still trusts almost everything.</p><h3>How teams usually roll it out</h3><ol><li>Add Report-Only and review what would be blocked in the browser console.</li><li>Fix legitimate scripts and styles (or allow them deliberately).</li><li>Switch to an enforced Content-Security-Policy header.</li><li>Prefer the HTTP header over a meta tag when you need framing controls or reporting.</li></ol><p>A basic same-origin starting point from OWASP looks like:</p><pre>Content-Security-Policy: default-src &#39;self&#39;; frame-ancestors &#39;self&#39;; form-action &#39;self&#39;;</pre><p>This is only a starting example. It is not a ready setting for every site, it may block existing pieces of the page, and it is not a Strict CSP. Test it first as Report-Only on staging.</p><h3>What this never proves</h3><p>A CSP on the pages you tested does not mean the site has no bugs, no malware, and no bad plugins. It means the browser got a policy for those responses. Outside scanners — including ours — only see what the site returns.</p><h3>About us</h3><p>We build HackerMe, an automated website security scanner by Agura / OMC Group. Security headers run on every website scan; a sample finding we show publicly is “Missing Content-Security-Policy header.” Deeper CSP analysis can check policy quality on Standard and higher scan types. HackerMe offers a free scan on a domain you own or have explicit permission to test: <a href="https://brandingme.ai/app/r/Xf3JFiq6lABO98_rFGscxQ">the free scanner</a></p><h3>Sources</h3><ul><li>MDN, Content Security Policy (CSP): <a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/CSP">https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/CSP</a></li><li>OWASP Content Security Policy Cheat Sheet: <a href="https://cheatsheetseries.owasp.org/cheatsheets/Content_Security_Policy_Cheat_Sheet.html">https://cheatsheetseries.owasp.org/cheatsheets/Content_Security_Policy_Cheat_Sheet.html</a></li><li>OWASP WSTG-CONF-12: <a href="https://wstg.owasp.org/latest/4-Web_Application_Security_Testing/02-Configuration_and_Deployment_Management/12-Content_Security_Policy/">https://wstg.owasp.org/latest/4-Web_Application_Security_Testing/02-Configuration_and_Deployment_Management/12-Content_Security_Policy/</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7ec4aa273630" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your Site Looks Fine to You. Google May Be Seeing Something Else.]]></title>
            <link>https://medium.com/@hackerme/your-site-looks-fine-to-you-google-may-be-seeing-something-else-585580f9624c?source=rss-d1d602509c70------2</link>
            <guid isPermaLink="false">https://medium.com/p/585580f9624c</guid>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[web-security]]></category>
            <category><![CDATA[small-business]]></category>
            <category><![CDATA[website-maintenance]]></category>
            <category><![CDATA[wordpress]]></category>
            <dc:creator><![CDATA[HackerMe]]></dc:creator>
            <pubDate>Wed, 04 Nov 2026 17:24:08 GMT</pubDate>
            <atom:updated>2026-11-04T17:24:08.673Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Written by the HackerMe team (Agura); we build HackerMe, an automated website security scanner. Everything below works without it.</em></p><p>A common story from small site owners goes like this: the homepage loads, the admin panel works, nothing looks broken. Then search traffic drops, or a customer says a Google result for the business shows pages about pills, casinos or replica watches. The site was compromised weeks ago, and nobody inside the business saw it.</p><p>That is not bad luck with timing. Some compromises are built so the owner does not notice. This article covers why that happens, what you can check from the outside, and where those checks stop.</p><h3>Why the owner is often the last to know</h3><p>Google’s spam policies describe two behaviors that fit this pattern.</p><p><strong>Hacked content</strong> is “any content placed on a site without permission, due to vulnerabilities in a site’s security.” Spam pages can be added under paths you never visit, so your normal browsing never touches them.</p><p><strong>Sneaky redirects</strong> send a visitor to a different URL than the one they requested, in a way meant to show “users and search engines different content.” Google’s own examples include showing search engines one type of content while redirecting users elsewhere, and showing desktop users a normal page while redirecting mobile users to a spam domain.</p><p>Put those together and you get a site that looks normal on the owner’s laptop, logged in, typing the URL directly, while a phone user arriving from a search result lands somewhere else.</p><h3>Five checks you can run from the outside</h3><p>None of these needs access to the server. Each one has a clear limit.</p><h3>1. Read the Security issues report in Search Console</h3><p>If you have verified the site in Google Search Console, open the Security issues report. Google says it shows its findings when an evaluation determines a site was hacked or shows harmful behavior, and that affected pages can carry a warning in search results or the browser.</p><p>An empty report is not a clean bill of health. It reflects what Google detected, not everything on the server.</p><h3>2. Search for your own domain with a site: query</h3><p>Search site:yourdomain.com and scan the titles. Add terms that have nothing to do with your business, such as site:yourdomain.com casino or site:yourdomain.com viagra.</p><p>Pages you did not create, or titles in a language your site does not use, are the red flag. A clean result covers only what is indexed; new spam may not be there yet.</p><h3>3. Request your pages the way a visitor from search would</h3><p>Redirects of this kind often depend on the referrer or the device. From a terminal, compare a plain request with one that claims to come from a search result on a phone:</p><pre>curl -s -D plain-headers.txt -o plain.html https://yourdomain.com/<br>curl -s -D search-headers.txt -o search.html \<br>     -A &quot;Mozilla/5.0 (iPhone; CPU iPhone OS 17_0 like Mac OS X)&quot; \<br>     -e &quot;https://www.google.com/&quot; https://yourdomain.com/<br>head -1 plain-headers.txt search-headers.txt<br>grep -i &#39;^location&#39; plain-headers.txt search-headers.txt</pre><p>These are normal GET requests that save both the headers and the page. We avoid curl -I here: it sends a HEAD request, and some servers and injected scripts answer HEAD differently from the GET a real visitor sends. Check what you actually received in each case: the real page, a redirect, or an error.</p><p>A redirect to an unfamiliar domain that appears only in the second request is a strong sign of trouble. Two limits matter. curl does not run JavaScript, so a redirect performed in the browser by an injected script will not show up as a Location header; look inside search.html for unfamiliar scripts (check 4), or repeat the test in a real phone browser opened from a search result. And malicious code can also key on IP ranges, cookies or time of day, or fire only once per visitor, so matching results are one data point, not a clean bill.</p><h3>4. Compare what the page sends with what you expect</h3><p>Save the HTML of a few key pages (curl -s https://yourdomain.com/ &gt; home.html) and look for script tags or links to domains you do not recognize. Do the same for a page deep in the site, not only the homepage.</p><p>Watch for an unknown external script, hidden links, or text that does not appear when you view the page normally. Injected code can also be loaded later by a legitimate-looking script, or served only to crawlers.</p><h3>5. Look at the sitemap and robots.txt</h3><p>Open /sitemap.xml and /robots.txt. Compromised sites sometimes get extra sitemaps listing thousands of spam URLs.</p><p>Entries you did not create are a finding. Their absence is not reassurance: spam pages do not need a sitemap to be indexed.</p><h3>What no outside check can tell you</h3><p>Everything above sees only what the site returns to a request. It cannot see:</p><ul><li>a backdoor file that is never linked and only answers a secret parameter,</li><li>a malicious admin account,</li><li>modified core or plugin files that behave normally until triggered,</li><li>stolen credentials or data that left the server.</li></ul><p>That is why finding something from the outside is a reason to start a proper cleanup, and finding nothing is not a reason to stop. Google’s hacked-site guidance on web.dev walks through the recovery steps: quarantine, identify the vulnerability, clean, and request a review. If the site handles customer data, bring in someone who can review the server itself.</p><h3>A routine that catches this earlier</h3><ul><li>Verify the site in Search Console and make sure alerts reach an address someone reads.</li><li>Run the site: query once a month.</li><li>Repeat check 3 after every plugin, theme or hosting change.</li><li>Keep the CMS, plugins and themes updated, and remove what you do not use.</li></ul><h3>About us</h3><p>HackerMe runs automated checks from the outside on a domain you own or have explicit permission to test. Like the checks above, it sees only what your site returns, so it can point you to things worth investigating, not confirm that a site is clean. <a href="https://brandingme.ai/app/r/MPEtMeuU9k8sh0qL0a0-OA">https://brandingme.ai/app/r/MPEtMeuU9k8sh0qL0a0-OA</a></p><h3>Sources</h3><ul><li>Google Search Central, Spam policies for Google web search (hacked content, sneaky redirects): <a href="https://developers.google.com/search/docs/essentials/spam-policies">https://developers.google.com/search/docs/essentials/spam-policies</a></li><li>Google Search Console Help, Security issues report: <a href="https://support.google.com/webmasters/answer/9044101">https://support.google.com/webmasters/answer/9044101</a></li><li>web.dev, Help for hacked sites: <a href="https://web.dev/articles/hacked">https://web.dev/articles/hacked</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=585580f9624c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>