The Next php2shell: A 12-Point Checklist to Have in Place Before the Next Pre-Auth RCE
The next php2shell will not be in PHP. It will be in a web runtime, a reverse proxy, a file-transfer appliance, or a logging library, and it will arrive the same way: a leak, an emergency release, a four-day head start for attackers, and a long tail of hosts nobody knew they had. The runtime changes each time. The response does not.
This checklist is built from the php2shell week. Each item is something a team either had ready on October 20 or spent the week wishing it had. It is split into what to have in place before the disclosure, what to do in the first 24 hours, and what to do in the week after.
Before the disclosure
1. An external inventory that starts from what you own, not from what you remember
The organizations that patched from a list on October 21 had one thing in common: an inventory built from the organizational structure outward, covering subsidiaries, acquired brands, and vendor-hosted assets under company domains. Organizations that started from a list of known domains spent the week discovering the PHP servers behind subsidiary brands acquired years ago, agency-built campaign sites, and admin consoles nobody had touched since 2016.
Check: can you produce, today, every internet-facing asset attributed to every entity in your corporate structure, including the ones acquired in the last three years?
2. Technology fingerprints on every asset, kept current
IONIX had every customer’s PHP assets mapped on the morning of October 21, before the fix shipped, because the fingerprinting already existed as part of continuous discovery. The question at the next disclosure will be a different runtime, so the fingerprint has to cover the whole stack: web servers, application runtimes, frameworks, admin consoles, CDN and WAF vendors.
Check: for any technology name, can you filter your inventory to the assets running it, including assets that strip their version header?
3. A list of end-of-life software on internet-facing hosts
72.5% of internet-facing hosts that advertise a PHP version run a branch older than 8.2. Those are the hosts that got no regular fix and will be unpatched at the next disclosure too. Every organization has an equivalent list for its own runtimes.
Check: which internet-facing hosts run runtimes past their vendor’s end-of-life date, and who owns each one?
4. A WAF coverage audit, including block mode and origin lockdown
IONIX’s State of the WAF report found 47% of live internet-facing web applications have no blocking firewall, and 29% of WAF-protected applications with a testable origin answer requests that bypass the WAF. A WAF rule only buys time for assets that meet all three conditions: behind the WAF, rules in block mode, origin reachable only through the WAF.
Check: which internet-facing assets meet all three, and which are exposed directly?
5. A pre-agreed decision on what “potentially affected” means
For php2shell, no safe exploitability test was possible, because confirming the bug crashes the server. IONIX said so in its public advisory and told customers to patch every potentially affected asset in priority order. Teams that had already decided “if we cannot confirm, we patch everything on the list in priority order” moved. Teams that waited for confirmation waited for something that could not arrive.
Check: does your vulnerability policy say what to do when exploitability cannot be validated safely?
6. Owner and escalation path per asset
A prioritized list is only useful if each line has a name next to it. The slowest patches during any disclosure are on assets where the first task is finding out which team, agency, or subsidiary operates the host.
Check: does each internet-facing asset in your inventory have an owner, and does that owner have a ticketing path?
In the first 24 hours
7. Start the clock at the rumor, not the advisory
CVE identifiers for php2shell circulated on October 20. Amazon’s threat intelligence team saw exploitation attempts four days before the patch. The vendor advisory arrived October 21. The response that started at the advisory was already four days behind the attacker.
Check: who in your team watches for pre-advisory signals, and what triggers a response before the official notice?
8. Produce a wide list first, then narrow it
IONIX’s first php2shell list was every asset running PHP, whatever version. Version-narrowed and feature-checked lists followed within 48 hours of the fix. A wide list on day zero that a team can start working beats a precise list on day three.
Check: can you produce the wide list within hours of a rumor, from data you already hold, without launching a new scan?
9. Apply the vendor’s partial mitigations, and know they are partial
PHP’s advisory listed configuration workarounds and said they “do not make an installation safe.” Several early advisories across the industry were more optimistic about one of them, disabling session upload progress, and PHP’s own advisory corrected that by the afternoon: it covers one of the two CVEs and only the confirmed path. Partial mitigations buy hours on hosts that cannot be patched this week. They buy nothing on hosts nobody has found.
Check: who decides which workarounds to deploy, and who tracks which hosts have them so they get patched rather than forgotten?
10. Communicate the honest confidence level
The most useful advisory format in the php2shell week was the one that said: three lists, highest confidence first, no single list complete, no safe exploitability test possible, patch across all three. Teams that got that message knew what to do with it.
Check: does your internal communication template have a place for “what we do not know yet”?
In the week after
11. Verify the patch from the outside
A disclosed version header is evidence, not proof. Distribution backports patch without changing the version string. Load balancers rewrite headers. Re-scan after remediation and confirm the fix from the attacker’s side of the firewall, including whether the vulnerable feature is still reachable.
Check: what triggers a re-scan of each remediated asset, and who confirms closure?
12. Track the long tail to zero
Five days after the php2shell fix, 94% of measurable hosts on the internet were unpatched, and the hosting-panel and Magento segments were at 100%. Inside an enterprise, the long tail is the end-of-life hosts, the subsidiary properties, and the vendor-operated assets. They do not patch themselves, and they are where the next breach starts.
Check: does the affected-asset list stay open and tracked until the last host is patched, migrated, isolated, or decommissioned?
The checklist, in one place
| # | Item | Phase |
|---|---|---|
| 1 | External inventory built from organizational structure outward | Before |
| 2 | Technology fingerprints on every asset, kept current | Before |
| 3 | End-of-life runtimes on internet-facing hosts listed and owned | Before |
| 4 | WAF coverage audit: presence, block mode, origin lockdown | Before |
| 5 | Policy for when exploitability cannot be validated | Before |
| 6 | Owner and escalation path per asset | Before |
| 7 | Response starts at the rumor | First 24h |
| 8 | Wide list first, narrowed lists within hours | First 24h |
| 9 | Partial mitigations deployed and tracked as partial | First 24h |
| 10 | Communication includes confidence levels and unknowns | First 24h |
| 11 | Patch verified from outside after remediation | Week after |
| 12 | Long tail tracked to zero | Week after |
Items 1 through 6 are what IONIX delivers continuously: an organizational entity map, discovery and fingerprinting across the full external attack surface, end-of-life tracking, WAF Posture Management, and an owner per asset. Items 7 through 12 are what Live Exposure Defense does at each disclosure: CVE correlation within hours, tiered lists with honest confidence, mitigation guidance, re-validation, and a tracked record to closure.
If you want to see where your organization stands on items 1 through 4 before the next disclosure, the free assessment covers them. Company name only, results within 24 hours.
Request your free exposure assessment
Readiness checklist FAQ
Six things: an external asset inventory built from the organizational structure outward, technology fingerprints on every asset, a list of end-of-life runtimes on internet-facing hosts, a WAF coverage audit that checks block mode and origin lockdown, a written policy for when exploitability cannot be validated, and an owner for every asset.
Within hours of the first credible rumor, not the vendor advisory. For php2shell, exploitation attempts began four days before the patch and CVE identifiers leaked a day before. A team that produces a wide list of potentially affected internet-facing hosts from existing inventory data on day zero is ahead of most attackers.
Only for assets that are behind the WAF, with rules in block mode, and with the origin unreachable except through the WAF. IONIX’s State of the WAF report found 47% of live web applications have no blocking firewall and 29% of protected origins answer direct requests. A WAF audit before the disclosure tells you which assets have time bought and which do not.
Re-scan the asset after remediation and check the advertised version against the fixed releases, then confirm the vulnerable session upload progress feature is no longer reachable. A version header alone is evidence, not proof, because distribution backports patch without changing the version string.
