Skip to content

Financial Services

Five Sites Rebuilt on Patched AWS Servers in Under an Hour After PHP2Shell

UK Specialist Lender

Under 1 hour

To migrate five sites to fresh, patched AWS servers

When the critical PHP2Shell vulnerabilities were disclosed, we used our agentic tooling to move a UK lender's five sites onto fresh, patched AWS servers in under an hour.

The Challenge

In October 2026, details of two critical flaws in PHP itself leaked ahead of their planned disclosure. Known together as PHP2Shell (CVE-2026-109618 and CVE-2026-109619), both were rated 9.2 under CVSS v4.0 and could allow unauthenticated remote code execution with the privileges of the web server.

Because the flaws sat in the PHP engine rather than in any application, every PHP site was exposed, however well its own code was written. Exploitation attempts were reported before the patches shipped, and the PHP maintainers were clear that updating was the only safe option.

Our client, a UK specialist lender, ran five PHP sites. As a regulated financial business, they could not leave those sites exposed while they waited for a fix, and could not afford a slow, manual recovery once it arrived.

Our Solution

Closing the Gap Before the Patch

There was a delay between the first exploits and the patched PHP release. Rather than leave the sites running on vulnerable PHP in the meantime, we replaced them with static placeholders. With no PHP running, there was nothing for PHP2Shell to attack, and visitors still saw a holding page rather than an error.

A Clean Rebuild, Not a Patch

When the patch shipped, we didn't just update PHP on the existing servers. Because exploits were circulating before the fix, we couldn't be certain the old machines were clean. So we rebuilt: all five sites moved onto fresh servers on AWS running patched PHP, leaving nothing behind that an attacker could have touched.

Human in the Lead

We ran the migration with our own agentic tooling. AI agents carried out the provisioning, configuration and migration work for each site, so the five migrations ran together rather than one after another.

Our engineers were in the lead, not stuck in the loop. They set the plan, made the decisions and owned every change that went live, without becoming a bottleneck for each routine step. The agents did the work; people stayed in charge of it.

Results

  • No PHP exposed while the client waited for a patch, thanks to static placeholders
  • Five sites rebuilt on fresh, patched AWS servers
  • Under an hour to migrate all five once the patch was available
  • Clean infrastructure with no carry-over from the previously exposed servers

Why It Matters

Critical vulnerabilities do not wait for a convenient maintenance window. With agentic tooling led by engineers who understand what it is doing, a response that would once have taken days can be done safely within the hour.

Ready to achieve similar results?

Schedule a strategy call to discuss how we can transform your AI initiatives into operational reality.

Stay ahead with AI insights

Practical notes on operationalising AI, a few times a year. We care about your data. Read our privacy policy.