Works with the AI Agents you already use
PHP2Shell Patch and Exposure Check (CVE-2026-109618)
Respond to the Oct 2026 PHP RCE flaws (CVE-2026-109618/109619): find every PHP runtime, check fixed versions, patch or contain, restart workers, hunt from Oct 17. CLEAR / CONTAINED / AT RISK.
$19
PHP2Shell Patch and Exposure Check (CVE-2026-109618)
Example session with this skill installed
We're a small agency hosting 40 client WordPress sites on 3 Ubuntu servers (PHP-FPM, one old box still on PHP 7.4) plus 2 Laravel apps on Forge. Help us respond to PHP2Shell today.
- Read your context and instructions
- Compiled the php2shell patch
PHP2Shell response: agency estate (2026-10-23 10:40)
Overall verdict: AT RISK. One internet-facing server runs PHP 7.4, for which there is no official fix.
Advisory checked: php.net 2026-10-21 release and the CERT-EU alert. No differences from the skill's snapshot.
Estate status
| System | Exposure | PHP (running) | SAPI | Status | Action / owner | Due |
|---|---|---|---|---|---|---|
| web-01 (18 WP sites) | Internet | 8.3.36 | FPM | PATCHED | Verified php-fpm8.3 -v after restart | Done 09:55 |
| web-02 (16 WP sites) | Internet | 8.2.31 | FPM | EXPOSED | apt upgrade to 8.2.35, restart FPM | 11:00, Sam |
| legacy-01 (6 WP sites) | Internet | 7.4.33 | mod_php | EXPOSED | Contain now: serve cached pages behind Cloudflare, block wp-admin except office IPs; migrate to 8.3 | Contain 11:30; migrate by Fri |
| forge-app-1 | Internet | 8.4.24 | FPM | EXPOSED | Forge: Update PHP 8.4, then restart Horizon and queue workers | 11:15, Priya |
| forge-app-2 | Internet | 8.4.27 | FPM | MITIGATED -> PATCHED? | Queue workers started before the patch; restart, then re-verify | 11:15 |
Inventory notes (php_exposure_check.sh)
- web-02: the distro package shows 8.2.31; the Ubuntu advisory lists no backport yet, so it stays EXPOSED until upgraded.
- forge-app-2: php artisan queue:work has been running since Oct 19, so it is still on the old runtime.
- Found an old staging site on legacy-01 (staging.client7.com) that wasn't on our list. Added it.
Interim mitigation on legacy-01 (temporary, expires Fri)
Advisory config settings applied with a container memory cap (they reduce exposure, they are not a fix). Cloudflare WAF rules on. Uploads disabled.
Compromise hunt (window from Oct 17)
- web-01 and web-02: wp core verify-checksums and plugin checksums clean; no PHP files in uploads; no new admin users. Nothing found
so far.
- legacy-01: 1 unknown PHP file in /wp-content/uploads/2026/10/ created Oct 18.
Escalated: host isolated, evidence copied, DB and API keys rotated, rebuild planned from clean image on PHP 8.3.
Notices drafted
Client email (40 sites), a specific notice for client 7 with incident wording to review with counsel, and an internal update every 2 h.
Still UNKNOWN
Two clients host their own DNS and CDN, so the WAF status there is unconfirmed.
Connects securely to your tools. The creator never sees your data.
About this skill
On 21 October 2026 the PHP project shipped fixes for two critical, unauthenticated remote-code-execution flaws, CVE-2026-109618 and CVE-2026-109619, nicknamed "PHP2Shell". Governments and major sites took systems offline. Exploitation attempts were reported days before disclosure. There is no official fixed release for PHP 8.1 and older. If you run WordPress, Drupal, Joomla, Moodle, Magento, Laravel or anything else on PHP, you need to know today which systems are exposed, patch them properly, and check whether anyone got in.
This skill gives your AI agent a calm, defensive incident-response procedure, a read-only host inventory script, a version checker, a compromise-hunt checklist and ready-to-send notices. It:
- Re-checks the live php.net advisory and your national CERT alert first, then compares them with a dated snapshot of the fixed versions (8.5.12, 8.4.27, 8.3.36, 8.2.35) and reports anything that changed
- Triages in minutes, putting internet-facing, mod_php, PHP 8.1-or-older and personal-data systems first
- Inventories every PHP runtime with a read-only script: CLI and FPM binaries, SAPIs, mod_php, distro packages (with backport warnings), interim php.ini settings, long-running workers that still run old code, and PHP container images. It changes nothing on the host
- Covers what a script can't see: registries and CI images, managed hosts and control panels (Forge, cPanel, Plesk), vendor appliances, forgotten staging and marketing sites
- Plans patching branch by branch, including the hard case: 8.1 and older have no official fix, so it lays out the options and keeps those systems marked EXPOSED until resolved
- Makes sure fixes are actually live: restarts FPM, Apache, queue workers, Horizon, Octane and FrankenPHP, then verifies the running version
- Explains the advisory's interim mitigations accurately, with their limits: config settings reduce exposure but are not a fix, and WAF rules only cover routed traffic
- Hunts for compromise from 17 October: new or changed PHP files, web shells in upload folders, new admin users, cron jobs, SSH keys, out-of-memory crash clusters and unusual outbound traffic. If anything turns up, it covers isolating, preserving evidence, rotating secrets and rebuilding
- Drafts an internal status update, a client notice for agencies and hosts, and a maintenance banner, stating only what was verified
- Gives each system a status (PATCHED, MITIGATED, EXPOSED or UNKNOWN) and an overall verdict: CLEAR, CONTAINED or AT RISK
For developers, sysadmins, agencies, freelancers and hosting providers responsible for PHP sites.
Defensive and honest: no exploit code, ever. A system is never marked patched without an observed running version, it never claims "no compromise", and every fact taken from the snapshot rather than a live source is flagged.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Security.
Incident facts are evolving; always re-checks the live advisory. Inventory limited to systems you can access. Defensive only: no exploit code or exploitability testing. Not legal advice on breach notification.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
- Software
- Bash
Python 3Web browsing
SSHorconsoleorhosting control panelGit
Bash runs the read-only inventory script on your Linux hosts. Python runs the version checker. Browsing checks the live php.net and CERT advisories.