OPSSHIELD says an attacker used compromised credentials to distribute a malicious script through cPGuard’s update infrastructure on November 2 at 02:06 UTC. Its November 3 advisory identifies 5.88.12 through 5.88.19, inclusive, as affected updates. cPGuard is a Linux web-server security suite used by hosting providers and server administrators.
Evidence status: a real update-channel compromise reported by the affected vendor. OPSSHIELD confirms a backdoor, collection of sensitive files and attacker retrieval of some collected data. It is still determining what left each customer server and how the distribution credentials were obtained.
Check historical updates, not just today’s version
The vendor says 5.88.25 or later removes the known backdoor and staged files. A server already running that release may nevertheless have installed an affected update earlier.
Check cPGuard’s update history in /opt/cpguard/logs/install_log for versions 5.88.12–5.88.19. The advisory includes the exact read-only check and examples. An affected entry confirms installation, rather than proving script execution or data transfer. OPSSHIELD instructs administrators to treat such servers as potentially compromised.
The reverse inference is unsafe too. A missing or incomplete log cannot establish that a server was unaffected. The absence of known malicious files after cleanup also cannot show whether the script previously ran.
Cleanup leaves copied secrets usable
The script targeted Linux account records and password hashes, root’s database configuration, shell history, and SSH authorization and known-hosts files. It uploaded collected material to OPSSHIELD’s own licensing server through a compromised account. The attacker retrieved data from there.
Follow the advisory’s recovery sequence:
- Update every cPGuard server and verify 5.88.25 or later. Use the vendor’s documented update procedure; contact support if the corrective release cannot be installed.
- Reset potentially affected local passwords from a trusted recovery environment. Include root and other password-enabled Linux accounts. Change reused passwords elsewhere from a trusted device.
- Rotate database and cloud credentials that may have been exposed. Include secrets recorded in root’s shell history, such as storage keys or cloud-provider tokens.
- Replace SSH private keys stored under
/root/.ssh. Remove their old public keys from every system that trusts them, as the vendor recommends.
Password hashes are not plaintext passwords, but attackers can test guesses against them offline. Strong passwords reduce that risk without removing the need for resets. Database configuration and shell history can contain directly usable secrets.
OPSSHIELD is also preparing an update to remove previously installed support-access keys. That precaution is separate from invalidating customer credentials.
Keep the scope estimates separate
The advisory’s preliminary estimate describes 32 collected files retrieved from the licensing server, each corresponding to a customer server. It explicitly says this is not the total potentially affected population.
Bill Toulas’s reporting for BleepingComputer separately quotes a vendor representative identifying 2,946 affected servers and a copied billing database. Its prose also describes “32 files from each customer,” which differs from the advisory’s wording. These reports do not provide a reconciled account of collection, transfer and attacker retrieval. Use the affected-version history for response decisions, rather than treating either figure as a complete per-server exposure assessment.
OPSSHIELD currently recommends updating and rotating credentials rather than routinely rebuilding every operating system. It calls that assessment provisional. Additional compromise indicators or an inability to establish server integrity warrant an individual assessment, potentially including a rebuild.
Caveat: OPSSHIELD is investigating its own breach; its cleanup and rebuild guidance remains provisional, and per-server data loss is unresolved.