<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Harry Agustiana</title>
    <description>The latest articles on DEV Community by Harry Agustiana (@harryagustiana).</description>
    <link>https://dev.to/harryagustiana</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F2761971%2Ffecc5812-e25b-43ff-81c4-398560a06443.jpeg</url>
      <title>DEV Community: Harry Agustiana</title>
      <link>https://dev.to/harryagustiana</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/harryagustiana"/>
    <language>en</language>
    <item>
      <title>Passive Website Health Checks with curl: Security Headers, TTFB, and Indexing Signals</title>
      <dc:creator>Harry Agustiana</dc:creator>
      <pubDate>Fri, 13 Nov 2026 03:03:38 +0000</pubDate>
      <link>https://dev.to/harryagustiana/passive-website-health-checks-with-curl-security-headers-ttfb-and-indexing-signals-795</link>
      <guid>https://dev.to/harryagustiana/passive-website-health-checks-with-curl-security-headers-ttfb-and-indexing-signals-795</guid>
      <description>&lt;p&gt;When someone asks you to sanity-check a live website before a campaign or after a DNS cutover, you rarely have SSH keys or CMS admin access right away. Even when you do, firing an active vulnerability scanner against a production server without formal change approval is a bad idea: aggressive crawlers trip WAF rate limits, pollute analytics, and submit junk to contact endpoints.&lt;/p&gt;

&lt;p&gt;A single passive HTTP request gives you a fast, non-intrusive baseline. From the public response headers and the first few kilobytes of HTML alone, you can verify transport locks, redirect latency, crawler directives, and social metadata in under a minute.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fe0a9eb1ng30twqylh5vm.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fe0a9eb1ng30twqylh5vm.png" alt="Four-layer table comparing signals, healthy baselines, and silent failures across security locks, server latency, crawl readiness, and search metadata" width="800" height="507"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  1. Check redirect hops and browser security headers with &lt;code&gt;curl&lt;/code&gt;
&lt;/h2&gt;

&lt;p&gt;Start by following redirects with &lt;code&gt;-sIL&lt;/code&gt; so you see every intermediate &lt;code&gt;301&lt;/code&gt; or &lt;code&gt;302&lt;/code&gt; hop along with the final response headers:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -sIL --max-redirs 5 https://example.com &lt;span class="o"&gt;|&lt;/span&gt; grep -iE &lt;span class="s2"&gt;"^(http/|location:|strict-transport-security:|content-security-policy:|x-frame-options:|x-content-type-options:|x-robots-tag:|x-powered-by:)"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Look for four specifics in that output (documented in detail in the &lt;a href="https://cheatsheetseries.owasp.org/cheatsheets/HTTP_Headers_Cheat_Sheet.html" rel="noopener noreferrer"&gt;OWASP HTTP Headers Cheat Sheet&lt;/a&gt;):&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Single-hop canonical redirect:&lt;/strong&gt; Entering &lt;code&gt;http://example.com&lt;/code&gt; should jump straight to the canonical &lt;code&gt;https://&lt;/code&gt; URL in one redirect. Chains like &lt;code&gt;http://example.com&lt;/code&gt; -&amp;gt; &lt;code&gt;https://example.com&lt;/code&gt; -&amp;gt; &lt;code&gt;https://www.example.com&lt;/code&gt; -&amp;gt; &lt;code&gt;https://www.example.com/&lt;/code&gt; waste round trips on high-latency mobile networks.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;Strict-Transport-Security&lt;/code&gt; (HSTS):&lt;/strong&gt; Confirms browsers pin HTTPS for repeat visits. Test subdomains carefully before adding &lt;code&gt;includeSubDomains&lt;/code&gt; or &lt;code&gt;preload&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Framing and injection locks:&lt;/strong&gt; Check for a &lt;a href="https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/CSP" rel="noopener noreferrer"&gt;&lt;code&gt;Content-Security-Policy&lt;/code&gt;&lt;/a&gt; header (at minimum &lt;code&gt;frame-ancestors 'self'&lt;/code&gt; to block clickjacking, paired with &lt;code&gt;X-Frame-Options: SAMEORIGIN&lt;/code&gt; for legacy clients) and &lt;code&gt;X-Content-Type-Options: nosniff&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;No leaked runtime versions:&lt;/strong&gt; Strip &lt;code&gt;X-Powered-By: PHP/x.y.z&lt;/code&gt; (&lt;code&gt;expose_php = Off&lt;/code&gt; in &lt;code&gt;php.ini&lt;/code&gt; or &lt;code&gt;fastcgi_hide_header X-Powered-By;&lt;/code&gt; in Nginx) so automated scanners cannot filter targets by minor version banner.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;In Nginx, remember the &lt;code&gt;always&lt;/code&gt; parameter on &lt;code&gt;add_header&lt;/code&gt;; without it, Nginx omits security headers on &lt;code&gt;404&lt;/code&gt; and &lt;code&gt;500&lt;/code&gt; responses, and remember that any child &lt;code&gt;location&lt;/code&gt; block with its own &lt;code&gt;add_header&lt;/code&gt; overrides parent-level headers unless repeated or moved to an &lt;code&gt;include&lt;/code&gt; snippet:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight nginx"&gt;&lt;code&gt;&lt;span class="n"&gt;add_header&lt;/span&gt; &lt;span class="nc"&gt;Strict&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Transport&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Security&lt;/span&gt; &lt;span class="s2"&gt;"max-age=31536000"&lt;/span&gt; &lt;span class="n"&gt;always&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="n"&gt;add_header&lt;/span&gt; &lt;span class="nc"&gt;Content&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Security&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Policy&lt;/span&gt; &lt;span class="s2"&gt;"frame-ancestors 'self';"&lt;/span&gt; &lt;span class="n"&gt;always&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="n"&gt;add_header&lt;/span&gt; &lt;span class="no"&gt;X&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Frame&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Options&lt;/span&gt; &lt;span class="s2"&gt;"SAMEORIGIN"&lt;/span&gt; &lt;span class="n"&gt;always&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="n"&gt;add_header&lt;/span&gt; &lt;span class="no"&gt;X&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Content&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Type&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Options&lt;/span&gt; &lt;span class="s2"&gt;"nosniff"&lt;/span&gt; &lt;span class="n"&gt;always&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  2. Measure DNS, TLS, and Time to First Byte (TTFB)
&lt;/h2&gt;

&lt;p&gt;Visual metrics like Largest Contentful Paint depend heavily on how fast the server delivers the first byte of HTML. Google's &lt;a href="https://web.dev/articles/ttfb" rel="noopener noreferrer"&gt;&lt;code&gt;web.dev&lt;/code&gt; TTFB documentation&lt;/a&gt; sets the "good" threshold at &lt;strong&gt;0.8 seconds (800 ms) or less&lt;/strong&gt;, and flags anything over &lt;strong&gt;1.8 seconds&lt;/strong&gt; as poor.&lt;/p&gt;

&lt;p&gt;Print the exact connection phases with &lt;code&gt;curl&lt;/code&gt;'s &lt;code&gt;-w&lt;/code&gt; format string:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -sL -o /dev/null -w &lt;span class="s2"&gt;"status=%{http_code} redirects=%{num_redirects} dns=%{time_namelookup}s tls=%{time_appconnect}s ttfb=%{time_starttransfer}s\n"&lt;/span&gt; https://example.com
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If &lt;code&gt;time_appconnect&lt;/code&gt; is fast (&lt;code&gt;&amp;lt; 0.15s&lt;/code&gt;) while &lt;code&gt;time_starttransfer&lt;/code&gt; exceeds &lt;code&gt;1.0s&lt;/code&gt;, the bottleneck sits behind TLS termination: usually uncached database queries on the homepage, cold serverless functions, or origin servers located far from your visitors without CDN edge caching.&lt;/p&gt;

&lt;h2&gt;
  
  
  3. Inspect the first 64 KB of HTML for indexing and mobile traps
&lt;/h2&gt;

&lt;p&gt;Security checklists often ignore the HTML body, while SEO crawlers often ignore response headers. Reading the opening &lt;code&gt;&amp;lt;head&amp;gt;&lt;/code&gt; markup catches deployment mistakes that break search visibility silently:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -sL --max-time 7 https://example.com &lt;span class="o"&gt;|&lt;/span&gt; head -c 65536 &lt;span class="o"&gt;|&lt;/span&gt; grep -iE &lt;span class="s2"&gt;"&amp;lt;title|name=[\"'](description|viewport|robots|generator)[\"']|rel=[\"']canonical[\"']|property=[\"']og:image[\"']|&amp;lt;h1"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Verify five items in that output:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;No accidental &lt;code&gt;noindex&lt;/code&gt;:&lt;/strong&gt; Check both &lt;code&gt;&amp;lt;meta name="robots"&amp;gt;&lt;/code&gt; and the HTTP &lt;code&gt;X-Robots-Tag&lt;/code&gt; header. Leaving staging &lt;code&gt;noindex&lt;/code&gt; rules active after launch drops the domain from Google Search and AI search tools; see &lt;a href="https://developers.google.com/search/docs/crawling-indexing/block-indexing" rel="noopener noreferrer"&gt;Google Search Central's &lt;code&gt;noindex&lt;/code&gt; documentation&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Mobile viewport present:&lt;/strong&gt; Confirm &lt;code&gt;&amp;lt;meta name="viewport" content="width=device-width, initial-scale=1"&amp;gt;&lt;/code&gt; exists so mobile browsers do not shrink a 980px desktop canvas.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Self-referencing &lt;code&gt;rel="canonical"&lt;/code&gt;:&lt;/strong&gt; Prevents ad campaign query strings (&lt;code&gt;?utm_source=...&lt;/code&gt;, &lt;code&gt;?gclid=...&lt;/code&gt;) from splitting indexing signals across duplicate URLs.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Complete title, H1, and &lt;code&gt;og:image&lt;/code&gt;:&lt;/strong&gt; Broken or missing &lt;code&gt;og:image&lt;/code&gt; URLs render empty link cards whenever customers share the site on Slack, WhatsApp, or LinkedIn.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Clean &lt;code&gt;&amp;lt;title&amp;gt;&lt;/code&gt; and &lt;code&gt;&amp;lt;meta name="description"&amp;gt;&lt;/code&gt;:&lt;/strong&gt; Compromised CMS plugins frequently serve normal visuals to logged-in admins while injecting spam keywords into the public &lt;code&gt;&amp;lt;title&amp;gt;&lt;/code&gt; for anonymous requests and search bots.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  4. Safety rules if you automate passive checks
&lt;/h2&gt;

&lt;p&gt;If you wrap these checks into a lightweight web endpoint or internal monitoring script, treat every user-supplied URL as untrusted input to prevent Server-Side Request Forgery (SSRF):&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Allow only &lt;code&gt;http:&lt;/code&gt; and &lt;code&gt;https:&lt;/code&gt; schemes on standard ports (&lt;code&gt;80&lt;/code&gt; and &lt;code&gt;443&lt;/code&gt;), and reject URLs containing embedded &lt;code&gt;username:password&lt;/code&gt; credentials.&lt;/li&gt;
&lt;li&gt;Block raw IPv4/IPv6 literals as well as &lt;code&gt;localhost&lt;/code&gt;, &lt;code&gt;.local&lt;/code&gt;, &lt;code&gt;.internal&lt;/code&gt;, and &lt;code&gt;.arpa&lt;/code&gt; hostnames, and validate resolved DNS records if your runtime permits private-network routing.&lt;/li&gt;
&lt;li&gt;Cap redirect hops (for example, max 4 redirects), enforce a strict fetch timeout (around 5 to 7 seconds), and stream at most the first 128 KB of HTML so a multi-gigabyte response cannot exhaust worker memory.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For a ready-made browser tool that runs these exact passive checks on Cloudflare Workers and prints a plain-language summary in English or Indonesian, you can test any public URL on &lt;a href="https://wanohdigital.biz.id/cek-website" rel="noopener noreferrer"&gt;Wanoh Digital Website Check&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Quick pre-launch verification routine
&lt;/h2&gt;

&lt;p&gt;Keep those three &lt;code&gt;curl&lt;/code&gt; commands in a shell alias or post-deploy script and run them right after every DNS switch, certificate renewal, or CMS upgrade. One passive request takes less than two seconds and catches the exact misconfigurations that otherwise sit unnoticed until search traffic or customer conversions drop.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure note: AI tools assisted with structuring and editing this guide; all commands, configurations, and technical references were verified by the author, Harry Agustiana.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>webdev</category>
      <category>security</category>
      <category>seo</category>
      <category>beginners</category>
    </item>
    <item>
      <title>Post-Patch Triage for PHP2Shell: A Practical Checklist for Laravel Servers and Docker</title>
      <dc:creator>Harry Agustiana</dc:creator>
      <pubDate>Thu, 22 Oct 2026 05:10:56 +0000</pubDate>
      <link>https://dev.to/harryagustiana/post-patch-triage-for-php2shell-a-practical-checklist-for-laravel-servers-and-docker-1ffd</link>
      <guid>https://dev.to/harryagustiana/post-patch-triage-for-php2shell-a-practical-checklist-for-laravel-servers-and-docker-1ffd</guid>
      <description>&lt;p&gt;PHP2Shell (CVE-2026-109618 and CVE-2026-109619) is fixed in PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35. This post covers what comes after the update: a triage you can run on a Laravel server, with Docker notes. Commands assume Linux, and paths vary by stack, so adjust them. Everything below is read-only inspection, and none of it replaces a proper forensic process.&lt;/p&gt;

&lt;p&gt;Use 2026-10-20 as your cutoff date. That is when the CVE identifiers started circulating, before the patch existed.&lt;/p&gt;

&lt;h3&gt;
  
  
  1. Confirm the version that serves web traffic
&lt;/h3&gt;

&lt;p&gt;The CLI version is not necessarily the version behind your web server.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;php -v
php-fpm -v            &lt;span class="c"&gt;# or php-fpm8.2 -v on Debian/Ubuntu&lt;/span&gt;
apachectl -M &lt;span class="o"&gt;|&lt;/span&gt; grep -i php
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If you installed from a distribution package, the version string can look old while the fix is present. Check the changelog for the CVE instead:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="c"&gt;# Debian/Ubuntu&lt;/span&gt;
apt changelog php8.2-fpm &lt;span class="o"&gt;|&lt;/span&gt; grep -i &lt;span class="s1"&gt;'CVE-2026-1096'&lt;/span&gt;
&lt;span class="c"&gt;# RHEL family&lt;/span&gt;
rpm -q --changelog php-fpm &lt;span class="o"&gt;|&lt;/span&gt; grep -i &lt;span class="s1"&gt;'CVE-2026-1096'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Updating Laravel does not update PHP. Restart PHP-FPM, Apache and any long-running workers (queue workers, Octane) after the update.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. Docker
&lt;/h3&gt;

&lt;p&gt;Each container carries its own PHP. The host's PHP does not matter here.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;docker &lt;span class="nb"&gt;exec&lt;/span&gt; &lt;span class="o"&gt;&amp;lt;&lt;/span&gt;container&lt;span class="o"&gt;&amp;gt;&lt;/span&gt; php -v
docker build --pull --no-cache -t myapp:2026-10-22 .
docker compose pull &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; docker compose up -d --build
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Watch for tags pinned to an old patch version (for example &lt;code&gt;php:8.2.30-fpm&lt;/code&gt;). &lt;code&gt;--pull&lt;/code&gt; will not move those. Edit the tag. Rebuild queue worker and scheduler containers too, not only the web one.&lt;/p&gt;

&lt;h3&gt;
  
  
  3. Files that should not be there
&lt;/h3&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="c"&gt;# PHP files modified since the cutoff, excluding dependency folders&lt;/span&gt;
find /var/www -type f -name &lt;span class="s1"&gt;'*.php'&lt;/span&gt; -newermt &lt;span class="s1"&gt;'2026-10-20'&lt;/span&gt; \
  -not -path &lt;span class="s1"&gt;'*/vendor/*'&lt;/span&gt; -not -path &lt;span class="s1"&gt;'*/node_modules/*'&lt;/span&gt;

&lt;span class="c"&gt;# PHP-like files where none should live&lt;/span&gt;
find /var/www/app/storage /var/www/app/public/uploads -type f \
  \&lt;span class="o"&gt;(&lt;/span&gt; -name &lt;span class="s1"&gt;'*.php'&lt;/span&gt; -o -name &lt;span class="s1"&gt;'*.phtml'&lt;/span&gt; -o -name &lt;span class="s1"&gt;'*.phar'&lt;/span&gt; \&lt;span class="o"&gt;)&lt;/span&gt;

&lt;span class="c"&gt;# If the deploy directory is a git checkout&lt;/span&gt;
git -C /var/www/app status --porcelain
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Legitimate deploys also change files, so compare against your deploy history before panicking.&lt;/p&gt;

&lt;h3&gt;
  
  
  4. Persistence
&lt;/h3&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;sudo ls -la /etc/cron.d /var/spool/cron/crontabs 2&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;/dev/null
systemctl list-timers --all
awk -F: &lt;span class="s1"&gt;'$3 &amp;gt;= 1000 {print $1}'&lt;/span&gt; /etc/passwd
sudo find /home /root -name authorized_keys -exec ls -l &lt;span class="o"&gt;{&lt;/span&gt;&lt;span class="o"&gt;}&lt;/span&gt; +
last -n 20
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;You are looking for anything you cannot explain: a job, a timer, a user or a key nobody on your team added.&lt;/p&gt;

&lt;h3&gt;
  
  
  5. Logs
&lt;/h3&gt;

&lt;p&gt;The researchers suggest watching for unusual multipart requests followed by memory limit errors, repeated HTTP 500 responses, oversized PHPSESSID response cookies and unexplained worker restarts.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;grep -i &lt;span class="s1"&gt;'allowed memory size'&lt;/span&gt; /var/log/php*-fpm.log &lt;span class="o"&gt;|&lt;/span&gt; tail -50
awk &lt;span class="s1"&gt;'$9 == 500 {print $1}'&lt;/span&gt; /var/log/nginx/access.log &lt;span class="o"&gt;|&lt;/span&gt; sort &lt;span class="o"&gt;|&lt;/span&gt; uniq -c &lt;span class="o"&gt;|&lt;/span&gt; sort -rn &lt;span class="o"&gt;|&lt;/span&gt; head
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The status field position assumes the common log format. Adjust if yours differs.&lt;/p&gt;

&lt;h3&gt;
  
  
  6. Rotate what a worker could have held
&lt;/h3&gt;

&lt;p&gt;The flaws can expose worker memory, so assume anything the app handles could have been read. At minimum:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;database passwords&lt;/li&gt;
&lt;li&gt;third-party API keys and mail credentials&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;APP_KEY&lt;/code&gt;, with care: rotating it invalidates sessions and anything encrypted with it. Recent Laravel versions support &lt;code&gt;APP_PREVIOUS_KEYS&lt;/code&gt; for graceful rotation, so check the docs for your version.&lt;/li&gt;
&lt;li&gt;active sessions (clear the session table or files)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Then run &lt;code&gt;php artisan config:clear&lt;/code&gt; and redeploy.&lt;/p&gt;

&lt;h3&gt;
  
  
  7. Where a scanner fits
&lt;/h3&gt;

&lt;p&gt;A filesystem scanner is one useful input here. &lt;a href="https://github.com/hryagstn/laravel-scalpel" rel="noopener noreferrer"&gt;Laravel Scalpel&lt;/a&gt; (open source, my project) covers a defined slice: PHP and PHP-like extensions in non-PHP zones such as &lt;code&gt;public&lt;/code&gt; and &lt;code&gt;storage&lt;/code&gt;, common obfuscation patterns, dangerous &lt;code&gt;.htaccess&lt;/code&gt; and &lt;code&gt;.user.ini&lt;/code&gt; directives, &lt;code&gt;.env&lt;/code&gt; integrity, and a baseline diff. I ran it against three Laravel applications and it found no backdoor files. Three practical notes for using any in-process scanner after an RCE:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Trigger it from outside the app's user.&lt;/strong&gt; An in-app run uses the same OS user as PHP, so an attacker with code execution could tamper with it. I run it both ways: from inside the application, and through an external runner (the &lt;code&gt;sentinel.sh&lt;/code&gt; pattern from n8n-bastion) on a schedule that posts findings to Telegram. I weigh the external result more. Send output somewhere the web user cannot write, and use &lt;code&gt;scalpel:verify&lt;/code&gt; if you enable signing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Make the baseline refresh the last step, not the first.&lt;/strong&gt; &lt;code&gt;scalpel:baseline --force&lt;/code&gt; bakes in whatever is on disk at that moment. I refreshed mine after the update, which is safe only because scheduled external diffs had been running before it, so any unexpected change would already have raised an alert. Without that, scan and diff first, and archive the previous snapshot (by default &lt;code&gt;storage/app/private/scalpel/baseline.json&lt;/code&gt;) somewhere the web user cannot write. Enable output signing before creating the first baseline, so a regenerated baseline gets flagged.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Check that silence means something.&lt;/strong&gt; No alert is evidence only if the scheduled runs executed during the window and the alert level covers the changes you care about. Confirm both in your runner's execution history. Alerts are filtered by severity (&lt;code&gt;--fail-on&lt;/code&gt;, plus any CRITICAL or HIGH filter in the runner), so a low-severity change may not page you.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Treat vendor and cache paths deliberately.&lt;/strong&gt; By default &lt;code&gt;vendor/&lt;/code&gt; and &lt;code&gt;bootstrap/cache&lt;/code&gt; are skipped by the content scanners (the baseline diff still hashes them). After an incident, &lt;code&gt;php artisan scalpel:scan --include-vendor&lt;/code&gt; is worth the extra time.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;None of this covers cron, keys, logs or memory, which is exactly why steps 3 to 5 exist.&lt;/p&gt;

&lt;h3&gt;
  
  
  8. After the update, and if you cannot update yet
&lt;/h3&gt;

&lt;p&gt;The fix adds a new hard memory limit, and hitting it ends the process without running shutdown handlers. Review &lt;code&gt;memory_limit&lt;/code&gt; and &lt;code&gt;post_max_size&lt;/code&gt; against what your app really needs.&lt;/p&gt;

&lt;p&gt;Only updating is a complete fix. If you are stuck, these reduce exposure to specific paths and do not make the server safe:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;CVE-2026-109618: set &lt;code&gt;memory_limit=-1&lt;/code&gt; so applications cannot override it, keep &lt;code&gt;post_max_size&lt;/code&gt; low, and enforce request size limits at the web server or proxy. Do not use RLIMIT_AS or RLIMIT_DATA as a substitute, since they can trigger the same vulnerable path.&lt;/li&gt;
&lt;li&gt;CVE-2026-109619: set &lt;code&gt;session.upload_progress.enabled=0&lt;/code&gt; and restart every worker. On PHP 8.5, also set &lt;code&gt;fatal_error_backtraces=0&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;A WAF rule is a stopgap. Cloudflare itself said attackers adapted payloads around its rules.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you find real evidence of intrusion, rebuild from a clean image and a known good backup. Hand cleaning rarely catches everything.&lt;/p&gt;

</description>
      <category>php</category>
      <category>security</category>
      <category>laravel</category>
      <category>devops</category>
    </item>
    <item>
      <title>What a Filesystem Scan Can and Cannot Tell You After a Laravel Intrusion</title>
      <dc:creator>Harry Agustiana</dc:creator>
      <pubDate>Thu, 08 Oct 2026 03:49:28 +0000</pubDate>
      <link>https://dev.to/harryagustiana/what-a-filesystem-scan-can-and-cannot-tell-you-after-a-laravel-intrusion-1hn5</link>
      <guid>https://dev.to/harryagustiana/what-a-filesystem-scan-can-and-cannot-tell-you-after-a-laravel-intrusion-1hn5</guid>
      <description>&lt;p&gt;Most Laravel security tools look at your source code, config, and dependencies before you deploy. That is useful, but it answers one question: &lt;em&gt;is my code safe?&lt;/em&gt; After a compromise, you need a different question answered: &lt;em&gt;has anything on this server been added or changed that should not be there?&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;That is the gap &lt;a href="https://github.com/hryagstn/laravel-scalpel" rel="noopener noreferrer"&gt;Laravel Scalpel&lt;/a&gt; was built for. It is an open-source intrusion evidence scanner that runs inside a Laravel app and inspects the deployed filesystem.&lt;/p&gt;

&lt;p&gt;A quick update first. Scalpel is now at v1.11.0 with 66 GitHub stars, and it was recently covered by &lt;a href="https://laravel-news.com/laravel-scalpel" rel="noopener noreferrer"&gt;Laravel News&lt;/a&gt;. More important to me, issues and pull requests are now coming from other developers, and the repo has contributors besides me. This post covers how the scanner works and one lesson from the Laravel News test that is worth knowing if you plan to run it in CI.&lt;/p&gt;

&lt;h2&gt;
  
  
  What it checks
&lt;/h2&gt;

&lt;p&gt;&lt;code&gt;php artisan scalpel:scan&lt;/code&gt; runs six scanners by default. Five look at the current files, and one compares them with a saved snapshot.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Structural.&lt;/strong&gt; Executable PHP files in places PHP should never live, by default &lt;code&gt;public/&lt;/code&gt; and &lt;code&gt;storage/&lt;/code&gt;. This covers &lt;code&gt;.php&lt;/code&gt; and less common extensions like &lt;code&gt;.phtml&lt;/code&gt; and &lt;code&gt;.phar&lt;/code&gt;, plus double extensions such as &lt;code&gt;shell.php.jpg&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Obfuscated code.&lt;/strong&gt; Common backdoor patterns such as &lt;code&gt;eval(base64_decode(...))&lt;/code&gt;, compressed payload execution, dynamic function calls, direct evaluation of request input, and long encoded strings.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;code&gt;.htaccess&lt;/code&gt; and &lt;code&gt;.user.ini&lt;/code&gt;.&lt;/strong&gt; Handler mappings that let the web server run scripts, external redirects, and PHP directives like &lt;code&gt;auto_prepend_file&lt;/code&gt;. That last one is a classic persistence trick, because it runs a hidden file on every request.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Environment.&lt;/strong&gt; A missing or unreadable &lt;code&gt;.env&lt;/code&gt;, a &lt;code&gt;.env&lt;/code&gt; under &lt;code&gt;public/&lt;/code&gt;, an empty &lt;code&gt;APP_KEY&lt;/code&gt;, keys that drift from &lt;code&gt;.env.example&lt;/code&gt;, and &lt;code&gt;APP_DEBUG=true&lt;/code&gt; in production.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Baseline diff.&lt;/strong&gt; Record a known-good state, then report added, modified, and deleted files later.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;php artisan scalpel:baseline   &lt;span class="c"&gt;# record SHA-256, size, and mtime&lt;/span&gt;
php artisan scalpel:diff       &lt;span class="c"&gt;# compare the current files against it&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Create the baseline only when you trust the application state, and turn on HMAC signing before you create the first one. A signed baseline lets &lt;code&gt;scalpel:diff&lt;/code&gt; detect a baseline that was regenerated without the signing key.&lt;/p&gt;

&lt;h2&gt;
  
  
  The limit you should understand first
&lt;/h2&gt;

&lt;p&gt;Scalpel runs in the same process and with the same permissions as your application. If an attacker can change your code, they can also change the scanner or its config. A filesystem scan detects evidence of an intrusion. It is not a firewall and it does not contain anything.&lt;/p&gt;

&lt;p&gt;For that reason, run it from an external trigger where possible, keep code directories read-only, and ship results to storage outside the server being scanned.&lt;/p&gt;

&lt;h2&gt;
  
  
  What an independent test found
&lt;/h2&gt;

&lt;p&gt;The Laravel News review ran Scalpel 1.9.0 on a fresh Laravel app. With a clean state and a baseline, the default scan reported nothing. After adding four harmless fixtures (a fake &lt;code&gt;eval(base64_decode(...))&lt;/code&gt; inside &lt;code&gt;if (false)&lt;/code&gt;, an &lt;code&gt;avatar.php.jpg&lt;/code&gt;, an &lt;code&gt;.htaccess&lt;/code&gt; handler mapping, and a &lt;code&gt;.user.ini&lt;/code&gt; with &lt;code&gt;auto_prepend_file&lt;/code&gt;), the content scanners caught them.&lt;/p&gt;

&lt;p&gt;The more interesting result was a false positive pattern. After &lt;code&gt;php artisan optimize&lt;/code&gt; compiled the framework views, the next scan reported about 100 &lt;code&gt;MEDIUM&lt;/code&gt; and two &lt;code&gt;HIGH&lt;/code&gt; findings inside &lt;code&gt;storage/framework/views&lt;/code&gt;. The structural scanner allows that directory, but the obfuscated-code scanner still reads the compiled files.&lt;/p&gt;

&lt;p&gt;If your app caches views in production, test this before you use Scalpel as a deployment gate. You can add &lt;code&gt;storage/framework/views&lt;/code&gt; to &lt;code&gt;content_scan_excluded_paths&lt;/code&gt;, but that means skipping content checks for every compiled view. Alternatively, run &lt;code&gt;php artisan optimize:clear&lt;/code&gt; and recreate the baseline before scanning.&lt;/p&gt;

&lt;h2&gt;
  
  
  Using it in CI
&lt;/h2&gt;

&lt;p&gt;Both &lt;code&gt;scalpel:scan&lt;/code&gt; and &lt;code&gt;scalpel:diff&lt;/code&gt; support table, JSON, GitHub Actions annotation, and SARIF output. &lt;code&gt;--fail-on&lt;/code&gt; sets the lowest severity that fails a job.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;php artisan scalpel:scan --format=sarif --fail-on=MEDIUM
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Exit code &lt;code&gt;0&lt;/code&gt; means a complete scan with no findings, &lt;code&gt;1&lt;/code&gt; means findings at or above your threshold, and &lt;code&gt;2&lt;/code&gt; means findings below it or an incomplete scan, so an unreadable directory never looks like a clean result. A &lt;code&gt;ScanFinished&lt;/code&gt; event is also dispatched, so you can send your own Slack, mail, or webhook alert without parsing command output.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;composer require hryagstn/laravel-scalpel
php artisan vendor:publish --tag=scalpel-config
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;It needs PHP 8.2+ and supports Laravel 10 to 13. If you hit a false positive or a case it misses, please open an issue. Real-world reports like the one above are what make the scanner better.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Source and docs: &lt;a href="https://github.com/hryagstn/laravel-scalpel" rel="noopener noreferrer"&gt;https://github.com/hryagstn/laravel-scalpel&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;Command simulator: &lt;a href="https://hryagstn.github.io/laravel-scalpel/" rel="noopener noreferrer"&gt;https://hryagstn.github.io/laravel-scalpel/&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>laravel</category>
      <category>php</category>
      <category>security</category>
      <category>opensource</category>
    </item>
    <item>
      <title>I Almost Shipped a Broken Database on Tencent EdgeOne Makers. Here's What Fixed It</title>
      <dc:creator>Harry Agustiana</dc:creator>
      <pubDate>Sat, 29 Aug 2026 06:25:27 +0000</pubDate>
      <link>https://dev.to/harryagustiana/i-almost-shipped-a-broken-database-on-tencent-edgeone-makers-heres-what-fixed-it-19h2</link>
      <guid>https://dev.to/harryagustiana/i-almost-shipped-a-broken-database-on-tencent-edgeone-makers-heres-what-fixed-it-19h2</guid>
      <description>&lt;p&gt;Here's a mistake I bet a lot of developers coming from traditional hosting will make on their first serverless edge project: assuming a local database file will just... work.&lt;/p&gt;

&lt;p&gt;I made that mistake building &lt;strong&gt;OptiQ&lt;/strong&gt;, a sales-tracking app for small optical shops, on &lt;strong&gt;Tencent EdgeOne Makers&lt;/strong&gt;. This post is less "here's a polished tutorial" and more "here's the debugging path I took," because I think the reasoning matters more than the final answer.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Project, Briefly
&lt;/h2&gt;

&lt;p&gt;Small optical shops have a messy sales pattern that most POS templates don't handle well. A customer might buy:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Just a frame&lt;/li&gt;
&lt;li&gt;Just a lens&lt;/li&gt;
&lt;li&gt;A complete pair (frame + lens)&lt;/li&gt;
&lt;li&gt;Softlens&lt;/li&gt;
&lt;li&gt;Small accessories (cleaning solution, frame screws) — no prescription needed
On top of that, the shop needs to track refraction data (OD/OS, cylinder, axis) per customer, and &lt;em&gt;sometimes&lt;/em&gt; link it to a sale — but not always. OptiQ is my answer to that: a small, relational, low-cost app built specifically around that flexibility.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Low-cost was non-negotiable here. I'm the only IT person at my company, so I don't get to justify infrastructure spend for a side project — it has to run on free tiers, or it doesn't ship.&lt;/p&gt;

&lt;h2&gt;
  
  
  Mistake #1: Treating an Edge Function Like a Regular Server
&lt;/h2&gt;

&lt;p&gt;For context, EdgeOne Makers is the rebrand of EdgeOne Pages (as of June 2026) — it grew from "frontend hosting" into a full-stack platform for Web and AI Agent apps, with support for React, Vue, Next.js, Astro, and a few agent frameworks. That expanded scope is what made it worth using for something beyond a static site.&lt;/p&gt;

&lt;p&gt;My plan was simple: static frontend on Pages, API logic on Pages Functions, SQLite file for storage. Deploy was smooth — genuinely, push-to-live in minutes, no CDN or SSL setup needed.&lt;/p&gt;

&lt;p&gt;Then I hit the wall: &lt;strong&gt;edge functions are stateless.&lt;/strong&gt; There's no persistent filesystem, and a given request might get routed to a different edge instance than the last one. A &lt;code&gt;.db&lt;/code&gt; file written on one request could be gone — or inconsistent — by the next. This isn't an EdgeOne-specific limitation; it's true of serverless edge compute generally. But it's an easy thing to &lt;em&gt;forget&lt;/em&gt; if your mental model still comes from VPS-style deployments.&lt;/p&gt;

&lt;p&gt;Lesson: figure out your state strategy before you write a single API route, not after your first weird bug report.&lt;/p&gt;

&lt;h2&gt;
  
  
  Mistake #2 (Almost): Picking a Database That Sleeps When You Need It Awake
&lt;/h2&gt;

&lt;p&gt;Once I accepted I needed a managed database, I evaluated three options:&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Option 1 — EdgeOne's native KV store.&lt;/strong&gt; Fast, simple, and built in. But it's key-value only — no joins, no relational queries. For OptiQ, that would mean hand-rolling indexes for every access pattern (&lt;code&gt;transactions by customer&lt;/code&gt;, &lt;code&gt;transactions by date range&lt;/code&gt;) and computing monthly totals by pulling records into app code and summing manually. It works, but it's the kind of thing that quietly turns into technical debt.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Option 2 — Supabase.&lt;/strong&gt; Full Postgres, generous free tier, great DX. Then I read the fine print: free projects &lt;strong&gt;pause after ~7 days of inactivity&lt;/strong&gt; and need a manual restore to wake back up. A shop that doesn't sell something every single day could open the app to a database that's asleep. That's a bad failure mode for a business tool, even a small one.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Option 3 — Turso.&lt;/strong&gt; SQLite (libSQL), but run as a managed service and queried over HTTP — which means it works fine from a stateless edge function, unlike a local file. Real SQL (joins, constraints, aggregates), no idle-pause behavior. This is what I went with, and the schema I'd already sketched needed almost zero changes to work with it.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Part of the Schema That Actually Matters
&lt;/h2&gt;

&lt;p&gt;Most of OptiQ's schema is unremarkable — customers, products, transactions, transaction line items. The one decision worth calling out:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight sql"&gt;&lt;code&gt;&lt;span class="k"&gt;CREATE&lt;/span&gt; &lt;span class="k"&gt;TABLE&lt;/span&gt; &lt;span class="n"&gt;transactions&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;
  &lt;span class="n"&gt;id&lt;/span&gt; &lt;span class="kt"&gt;INTEGER&lt;/span&gt; &lt;span class="k"&gt;PRIMARY&lt;/span&gt; &lt;span class="k"&gt;KEY&lt;/span&gt; &lt;span class="no"&gt;AUTOINCREMENT&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;customer_id&lt;/span&gt; &lt;span class="kt"&gt;INTEGER&lt;/span&gt; &lt;span class="k"&gt;REFERENCES&lt;/span&gt; &lt;span class="nf"&gt;customers&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;id&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="n"&gt;refraction_id&lt;/span&gt; &lt;span class="kt"&gt;INTEGER&lt;/span&gt; &lt;span class="k"&gt;REFERENCES&lt;/span&gt; &lt;span class="nf"&gt;refractions&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;id&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="c1"&gt;-- nullable, on purpose&lt;/span&gt;
  &lt;span class="n"&gt;total&lt;/span&gt; &lt;span class="kt"&gt;REAL&lt;/span&gt; &lt;span class="k"&gt;NOT&lt;/span&gt; &lt;span class="kc"&gt;NULL&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;refraction_id&lt;/code&gt; is nullable. That's the whole trick. Without it, I'd have needed separate transaction tables (or awkward sentinel values) for "sale with a prescription attached" vs "sale without one" — like buying a bottle of lens cleaner shouldn't require a fake prescription record just to satisfy a NOT NULL constraint. One nullable foreign key instead of a schema fork.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where EdgeOne Makers' Full-Stack Positioning Actually Paid Off
&lt;/h2&gt;

&lt;p&gt;The thing I didn't expect going in: not having to stitch together three separate vendors. Static hosting, edge API functions, and (once I wired it up) the database connection all lived in one deploy pipeline. If you've ever debugged a CORS or env-var mismatch between a frontend host, a serverless function provider, and a database vendor that don't share tooling, you'll get why that's worth mentioning.&lt;/p&gt;

&lt;h2&gt;
  
  
  If You're About to Build Something Similar
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Decide your state/database approach &lt;em&gt;before&lt;/em&gt; your first API route — not as a fix after deploying.&lt;/li&gt;
&lt;li&gt;Check what happens to your database on &lt;em&gt;inactivity&lt;/em&gt;, not just under load. "Free tier" often hides a failure mode in the fine print.&lt;/li&gt;
&lt;li&gt;Use nullable foreign keys deliberately when your domain has "sometimes this relationship exists, sometimes it doesn't" — it's cheaper than modeling every variant as a separate table.
OptiQ is live at &lt;a href="https://optiq.logota.biz.id/" rel="noopener noreferrer"&gt;optiq.logota.biz.id&lt;/a&gt; — built on Tencent EdgeOne Makers as part of Codepolitan's DevHandal 2026 Batch 2.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;#TencentEdgeOne #EdgeOneMakers #CODEPOLITAN #EdgeOne&lt;/p&gt;

</description>
      <category>tencentedgeone</category>
      <category>edgeonemakers</category>
      <category>codepolitan</category>
      <category>edgeone</category>
    </item>
    <item>
      <title>How a Broken Laravel App Turned Into Two Open Source Security Tools</title>
      <dc:creator>Harry Agustiana</dc:creator>
      <pubDate>Mon, 13 Jul 2026 05:11:32 +0000</pubDate>
      <link>https://dev.to/harryagustiana/how-a-broken-laravel-app-turned-into-two-open-source-security-tools-287c</link>
      <guid>https://dev.to/harryagustiana/how-a-broken-laravel-app-turned-into-two-open-source-security-tools-287c</guid>
      <description>&lt;p&gt;I now run a small self-hosted watchtower that pings me on Telegram the moment something looks wrong on any Laravel app I manage. It didn't start as a plan. It started with a friend's text message and a server that had already been compromised for who knows how long.&lt;/p&gt;

&lt;h2&gt;
  
  
  The app nobody thought was a target
&lt;/h2&gt;

&lt;p&gt;The app was PaketLebaranku.id — a small Laravel + MySQL + Livewire platform running on a modest VPS through Biznet Gio, built for a friend's small business. Its only job: track installment savings for people paying into a Lebaran package program, small recurring deposits that add up to something meaningful once a year.&lt;/p&gt;

&lt;p&gt;It's exactly the kind of app that never shows up in anyone's threat model. No enterprise budget, no dedicated ops team, no security monitoring — just a business owner and a friend who built the thing on the side. Which, as it turned out, is precisely why it got hit.&lt;/p&gt;

&lt;h2&gt;
  
  
  "The app is throwing a 500 error, can you check?"
&lt;/h2&gt;

&lt;p&gt;That was the whole message. I expected a stack trace, maybe a bad migration or an expired dependency. Instead, digging into the error turned up something far worse.&lt;/p&gt;

&lt;p&gt;The &lt;code&gt;.env&lt;/code&gt; file wasn't misconfigured — it was gone. Any Laravel developer knows what that means: database credentials, app keys, third-party secrets, either already exposed or already pulled by whoever got in.&lt;/p&gt;

&lt;p&gt;Then a second find: a handful of backdoor scripts sitting inside &lt;code&gt;public/icons&lt;/code&gt;, dressed up to look like they belonged there. Then a third: &lt;code&gt;.htaccess&lt;/code&gt; had been rewritten to let Python scripts execute inside a directory that should never be able to run anything.&lt;/p&gt;

&lt;p&gt;Three separate footholds, sitting quietly in an app tied to real people's savings, for who knows how long before a user happened to hit an error page.&lt;/p&gt;

&lt;h2&gt;
  
  
  The gap that actually mattered
&lt;/h2&gt;

&lt;p&gt;Once we'd contained it, the obvious question was "how do I catch this faster next time." But sitting with it longer, I realized the real failure wasn't a missing check — it was that nothing was watching at all. A scan run once, manually, after the fact doesn't help. The incident wasn't caught by a security tool; it was caught by luck, in the form of an annoyed user.&lt;/p&gt;

&lt;p&gt;That reframed what I actually needed to build: not just something that &lt;em&gt;could&lt;/em&gt; find this class of problem, but something that would look for it &lt;em&gt;without anyone having to remember to ask&lt;/em&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Laravel Scalpel: the checks I wish had already existed
&lt;/h2&gt;

&lt;p&gt;The first piece is &lt;strong&gt;Laravel Scalpel&lt;/strong&gt;, a narrow, fast forensic scanner for Laravel codebases. It doesn't try to be a full SAST suite — it answers the exact three questions that would have shortened this incident from days to minutes: are there files sitting where they shouldn't be, does the environment configuration look right, are there permission or execution paths that shouldn't exist.&lt;/p&gt;

&lt;p&gt;It's picked up some traction beyond my own use case — featured by the Laravel Artisan Community, 24+ GitHub stars, and a pull request currently open to get it listed in &lt;code&gt;awesome-laravel&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  N8N Bastion: making sure someone's actually watching
&lt;/h2&gt;

&lt;p&gt;The second piece, &lt;strong&gt;N8N Bastion&lt;/strong&gt;, is what closes the real gap. It's a self-monitoring stack built on n8n that turns infrastructure and application problems into a Telegram message — instead of a support ticket that arrives after the damage is done. Laravel Scalpel now runs inside it as a scheduled check, not a tool anyone has to remember to invoke.&lt;/p&gt;

&lt;p&gt;Together they cover both halves of the problem: Scalpel knows what to look for on a Laravel app, and Bastion makes sure it's actually looking, on a schedule, without a human in the loop.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why this is worth open-sourcing
&lt;/h2&gt;

&lt;p&gt;PaketLebaranku.id will never make a breach headline. It's too small, too obscure, too ordinary — and that's exactly the point. There are a lot of small businesses running real, sensitive workloads on a single self-managed VPS with no monitoring at all, not because the owners don't care, but because every security tool out there quietly assumes a team and a budget they don't have.&lt;/p&gt;

&lt;p&gt;Laravel Scalpel and N8N Bastion exist because of one real incident on one real app. I'm putting both out there because I'd bet a lot of other small Laravel apps are sitting exactly where PaketLebaranku.id was, right before that 500 error showed up.&lt;/p&gt;

&lt;p&gt;If you run a Laravel app on a VPS you manage yourself, I'd genuinely like to know what breaks or what's missing if you try these:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Laravel Scalpel: &lt;a href="https://github.com/hryagstn/laravel-scalpel" rel="noopener noreferrer"&gt;https://github.com/hryagstn/laravel-scalpel&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;N8N Bastion: &lt;a href="https://github.com/hryagstn/n8n-bastion" rel="noopener noreferrer"&gt;https://github.com/hryagstn/n8n-bastion&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

</description>
    </item>
  </channel>
</rss>
