Checked as of 4 November 2026.

cPGuard is a security agent that some hosting providers install on the servers behind shared and managed hosting plans. As a website owner you rarely see it. That is why this incident matters: the people who must act are mostly your provider’s administrators, and your job is to make sure they have.

On 3 November 2026 the vendor, OPSSHIELD, published a security incident disclosure describing how an attacker used the cPGuard update channel to push a malicious update to customer servers. It replaces the vendor’s earlier notice on the same incident. The Hacker News reported the discovery on 3 November; that report cites the earlier guidance, so the current vendor disclosure is the document that should drive decisions.

What the vendor’s investigation says

Everything below comes from OPSSHIELD’s own account of its investigation, not from independent forensics.

The attacker broke into the vendor’s billing and licensing server on 1 November, took the key used to publish updates, and uploaded a fake cPGuard update. Agents installed it between 02:08 and 03:20 UTC on 2 November, which is 09:08 to 10:20 Vietnam time. The fake update ran as root. It copied local password hashes, root SSH keys, the MySQL root password where one was stored, root shell history and, on some servers, cloud storage keys. It also left a hidden root shell and a scheduled task behind. The earlier notice identified the malicious update versions as 5.88.12 through 5.88.19 inclusive.

What changed in the latest disclosure

The earlier notice pointed administrators at their update logs. The latest disclosure sets a stricter rule: if cPGuard was installed and online during that window, assume the server’s root-level secrets are known to the attacker. Three points follow.

  • A log match proves installation, not execution. A matching version in the update log confirms the fake update was installed. It does not by itself prove the script ran or that data left the server. The vendor’s guidance is to treat the server as exposed anyway.
  • Missing logs are not clearance. Neither is a clean current version. A server that was online in the window is in scope even if nothing suspicious shows today.
  • Cleanup does not revoke secrets. The vendor says version 5.88.26 removes the hidden root shell and a follow-up update removes the scheduled task. Install the latest release, not merely one newer than the malicious range. Even then, the copied secrets stay valid until someone rotates them.

Where the work happens

Rotating root passwords and SSH keys is server-level work. The privileges you hold on a shared or managed plan vary, but root and SSH operations are normally managed by the provider, and you should not run disruptive operations on infrastructure you do not administer. What you can do is ask precise questions, keep the answers, and secure what is yours.

Checklist for your hosting provider

Ask your provider to confirm each point in writing.

  1. Product. Is cPGuard installed on the server hosting my site, or on any server used to manage it?
  2. Window. Was that server online between 02:08 and 03:20 UTC on 2 November? If yes, is it treated as exposed regardless of what the logs show?
  3. Use. Which cPGuard versions were installed around that time, and did any fall in the 5.88.12 to 5.88.19 range?
  4. Cleanup. Is cPGuard on the latest release, and are the leftover hidden shell and scheduled task confirmed gone? The disclosure lists the exact file paths.
  5. Rotation. Have root and local passwords been reset, root SSH keys replaced and old keys removed from every server that trusted them, database passwords and secrets in root shell history rotated, cloud storage keys replaced, and the OPSSHIELD billing password reset with two-factor login enabled?
  6. Recovery. Did OPSSHIELD notify them that the attacker logged in to a server by hand? The vendor advises rebuilding such servers from a clean backup taken before 1 November. For other servers, how was integrity judged? Expect a case-specific answer, not a blanket assurance.
  7. Evidence. Were logs and copies of affected files preserved before cleanup?

Your own steps

  • Change your credentials from a trusted device, not through a session on the possibly exposed server, and protect the email and phone used for account recovery.
  • Ask your provider what else may have been reachable. The vendor’s list of collected files covers password hashes, root SSH keys, the MySQL root password where stored, shell history and some cloud storage keys. Root-level access can reach more than that list, so as an extra precaution the provider should assess and rotate other server and application credentials that may have been reachable, including the database password your site uses, where their assessment supports it.
  • Confirm you have a tested off-server backup. Our guide to backup rules and restore testing explains what a usable one looks like.
  • Watch for unusual activity, not only downtime. See monitoring beyond uptime.
  • If your site already shows signs of tampering, follow what to do when your website is hacked and preserve evidence before anything is cleaned.

There is no universal all-clear here. Whether your server was involved, and how far the exposure reaches, depends on answers only your provider has. Treat those answers as evidence for your next decision, such as whether to rotate more credentials or move the site, not as a security clearance.