> ## Content Index
> Fetch the complete content index at: https://f4n6.co.uk/llms.txt
> Use this file to discover other available public pages before exploring further.

# cPGuard supply chain compromise exfiltrates root keys from 2,950 servers
- URL: https://f4n6.co.uk/security-feed/cpguard-supply-chain-compromise-exfiltrates-root-keys-from-2-950-servers/
- Published: 2026-11-04T12:06:33.000Z
- Updated: 2026-11-04T12:06:33.000Z
- Author: Jeff Davies
- Tags: #security-feed

## 1\. Executive summary

An attacker compromised the update infrastructure of cPGuard, a security suite for Linux web hosting servers made by OPSShield, and pushed malicious updates (versions 5.88.12 through 5.88.19) to about 2,950 customer servers, where they ran as root between 02:04 and 02:09 UTC on 2 November 2026\. The malicious script exfiltrated root SSH keys, password hashes, the MySQL root password where present, root shell history and cloud storage keys, and left a hidden root shell at /var/tmp/.cpg91511\_sh with files in /var/tmp/.i93677/. OPSShield also lost the name, company, postal address, email and bcrypt-hashed password of about 7,900 billing customers. No CVE has been assigned and no severity score is recorded for this item; the risk to EMEA financial services rests on the class of data taken, not on a scored vulnerability. Any client running cPGuard on hosting or infrastructure it controls should treat root credentials on those servers as exposed until the checks in §4 are complete.

## 2\. Regulatory framing

No specific DORA/NIS2 article is directly engaged by this item.

## 3\. Technical analysis & attack chain

The chain below is drawn from OPSShield's account as reported by Help Net Security. OPSShield has not established how the attacker obtained the update key or how commands were executed on the billing and licensing server, so those two steps remain open.

1. On 1 November 2026, the attacker probed OPSShield's billing portal, registered a customer account under a false identity, ordered a free cPGuard trial and installed it to study how the agent communicates with the licensing system.
2. An hour and a half later the attacker sent crafted requests to the billing portal's callback pages. Minutes after that they gained the ability to run commands on the billing and licensing server.
3. From that server the attacker collected client account records: name, company, postal address, email and bcrypt-hashed password for about 7,900 customers. They also made copies of three of OPSShield's cloud servers.
4. On 2 November 2026 between 02:04 and 02:09 UTC, the attacker uploaded fake cPGuard updates, versions 5.88.12 through 5.88.19, to OPSShield's update storage.
5. Over the next 70+ minutes, cPGuard agents on about 2,950 customer servers downloaded and installed the fake update and ran it as root.
6. The fake update was a short script that collected sensitive files and sent them to a server controlled by the attacker. In an earlier report OPSShield said the destination was OPSShield's own licensing server, where the data was stored under a compromised account for later retrieval.
7. The script left two components behind: a hidden root shell at /var/tmp/.cpg91511\_sh and files in /var/tmp/.i93677/. OPSShield's follow-up update removes an i93677 scheduled task.

Data taken from customer servers: password hashes, root SSH keys, the MySQL root password where present, root shell history (which may contain typed passwords and tokens) and cloud storage keys. The combination matters more than any single item. Root SSH keys and shell history give an intruder a route back onto the server and onto every system that trusts those keys, and the cloud storage keys extend that reach to object storage.

The update mechanism is the vulnerability. cPGuard agents pull updates on an interval and execute them as root, so control of the update storage is equivalent to root on every server that checks in during the window. OPSShield states that servers not online during the window did not receive the fake update. No CVE has been assigned and NVD has not scored this item.

OPSShield has ruled out SSH, the control panel, FTP and API tokens as the route into the billing and licensing server, and says its logs do not record the exact request that gave access. It has patched two possible entry points and is reviewing related vendor components. The initial-access vector for the update storage key is not stated in public reporting.

OPSShield's own commentary attributes the speed of the operation (about ten hours from first probe to fake update) to attackers using AI tools to scan for weak points, read unfamiliar code and write custom tools. That is the vendor's assessment, not an independently corroborated finding. No threat actor has been named by any source, and no attribution is made here.

## 4\. Mitigation & containment

P1, within 24 hours:

- Determine whether cPGuard versions 5.88.12 through 5.88.19 were installed on any server. Check the installed version on each host and the agent update logs for the window 02:04 to 02:09 UTC on 2 November 2026\. Servers that were offline during the window did not receive the fake update, per OPSShield.
- On any affected server, update to cPGuard v5.88.26 or later, then apply the follow-up update that removes the i93677 scheduled task. Confirm that /var/tmp/.cpg91511\_sh and /var/tmp/.i93677/ are gone and that no i93677 scheduled task remains.
- Treat root and all local account credentials on affected servers as exposed. Reset passwords for root and every other password-enabled local Linux account from a trusted recovery environment.
- Rotate database passwords, S3 access keys, DigitalOcean tokens, and any password, API key or token recorded in root's shell history, plus any keys in /root/.s3cfg.
- Replace SSH private keys stored under /root/.ssh and remove the corresponding old public keys from every system that trusts them.
- Reset the OPSShield billing portal password and enable two-factor login.
- For servers where OPSShield has directly confirmed that the attacker logged in by hand, rebuild from a clean backup taken before 1 November 2026.

P2, within 72 hours:

- Review affected servers for other signs of unauthorised change: unfamiliar users, SSH keys, services and outbound connections.
- Hunt for use of the stolen cloud credentials against S3, DigitalOcean and any other cloud account whose keys sat on an affected host.

P3, within 7 days:

- Where cPGuard is used on infrastructure that supports payment or customer-facing services, confirm that the update path is authenticated and that agents do not execute updates as root without signature verification. OPSShield has not published a signed-update mechanism, so this is a control to verify locally, not a vendor fix to apply.

No fixed release is named for the billing portal or licensing server compromise beyond OPSShield's own rebuild; that remediation is internal to the vendor.

## 5\. Indicators of compromise

OPSShield has published a list of indicators that includes some found on its own servers. That list is not reproduced in the reporting available here, so the atomic indicators below are limited to the artefacts named in the text. No network indicators are given in the source.

| type           | value                  | confidence | source                              |
| -------------- | ---------------------- | ---------- | ----------------------------------- |
| file path      | /var/tmp/.cpg91511\_sh | high       | Help Net Security, citing OPSShield |
| file path      | /var/tmp/.i93677/      | high       | Help Net Security, citing OPSShield |
| scheduled task | i93677                 | high       | Help Net Security, citing OPSShield |
| file path      | /root/.s3cfg           | high       | Help Net Security, citing OPSShield |
| file path      | /root/.ssh             | high       | Help Net Security, citing OPSShield |

```iocs
filepath  /var/tmp/.cpg91511_sh
filepath  /var/tmp/.i93677/
schtask  i93677
filepath  /root/.s3cfg
filepath  /root/.ssh

```

Behavioural indicators

| behaviour                                                                                                | where to observe                                  | confidence                                           |
| -------------------------------------------------------------------------------------------------------- | ------------------------------------------------- | ---------------------------------------------------- |
| cPGuard agent installs version 5.88.12 through 5.88.19 outside a scheduled maintenance window            | cPGuard agent update log, package manager history | high                                                 |
| Outbound connection from a host running cPGuard to OPSShield's licensing server carrying collected files | Host egress logs, firewall logs                   | medium, single-sourced to OPSShield's earlier report |
| Creation of a hidden shell at /var/tmp/.cpg91511\_sh                                                     | File integrity monitoring, EDR file events        | high                                                 |
| Creation of an i93677 scheduled task                                                                     | cron, systemd timer, or at-job inventory          | high                                                 |

## 6\. Detection

The source gives file paths and a scheduled-task name but no distinctive content strings from the malicious script, so a YARA rule built on those paths would match on filename alone. The Sigma rule below targets the observable behaviour: creation of the hidden shell and the scheduled task.

```yaml
title: cPGuard malicious update artefacts
id: 8f3c2a71-4b6e-4d2a-9c1e-cpguard0001
status: experimental
description: Detects creation of the hidden root shell and i93677 scheduled task left by the malicious cPGuard update (versions 5.88.12 through 5.88.19)
author: Adverse Trace
date: 2026-11-04
references:

  - https://www.helpnetsecurity.com/2026/11/04/cpguard-update-supply-chain-compromised/
logsource:
  product: linux
  category: file_event
detection:
  selection_shell:
    TargetFilename|startswith: '/var/tmp/.cpg91511_sh'
  selection_dir:
    TargetFilename|startswith: '/var/tmp/.i93677/'
  condition: 1 of selection_*
falsepositives:

  - Unlikely; these paths are specific to the malicious update
level: critical

```

```yaml
title: cPGuard i93677 scheduled task creation
id: 2d9b7e14-6a3f-4c8b-b5e2-cpguard0002
status: experimental
description: Detects creation of a scheduled task named i93677 associated with the malicious cPGuard update
author: Adverse Trace
date: 2026-11-04
references:

  - https://www.helpnetsecurity.com/2026/11/04/cpguard-update-supply-chain-compromised/
logsource:
  product: linux
  category: process_creation
detection:
  selection:
    CommandLine|contains: 'i93677'
  condition: selection
falsepositives:

  - Administrative scripts that reference the string i93677; verify against the file paths above
level: high

```

## 7\. Sources

- Help Net Security, "Poisoned cPGuard update hits thousands of web hosting servers", https://www.helpnetsecurity.com/2026/11/04/cpguard-update-supply-chain-compromised/, 2026-11-04
- The Hacker News, "cPGuard Update Channel Hijacked to Backdoor Linux Servers and Collect Password Hashes", https://thehackernews.com/2026/11/cpguard-update-channel-hijacked-to.html
- SecurityWeek, "Malicious cPGuard Updates Served in OpsShield Supply Chain Attack", https://www.securityweek.com/malicious-cpguard-updates-served-in-opsshield-supply-chain-attack/

The technical detail in §3 and §4 rests on Help Net Security's report of OPSShield's advisory. The Hacker News and SecurityWeek items corroborate the event and the fixed version (5.88.25 and later per The Hacker News, 5.88.26 per Help Net Security) but add no independent technical detail. The account is effectively single-sourced to the vendor; verify the fixed version against OPSShield's own advisory before relying on it.

## 8\. Adverse Trace position

Adverse Trace assesses this as a high-impact supply chain compromise for any client running cPGuard on infrastructure it controls, because the malicious update executed as root and took exactly the credentials that allow an intruder to return: root SSH keys, password hashes, shell history and cloud storage keys. No CVE has been assigned and no severity score is recorded, so the assessment rests on the data class, not a CVSS rating. No actor is named in any source and no attribution is made. Clients that do not run cPGuard have no exposure from this event; clients that do should complete the P1 checks in §4 before treating any affected host as trusted. Adverse Trace will monitor OPSShield's published indicator list and update this advisory if it becomes available, and will re-issue if the initial-access vector or any actor attribution is established.

---

[Read the original source →](https://www.helpnetsecurity.com/2026/11/04/cpguard-update-supply-chain-compromised/?ref=f4n6.co.uk)

*Published via PulseTrace — Adverse Trace threat intelligence.*