← ALL POSTS
Advisories·11 min read

SiteGuard Server Edition flaw CVE-2026-106931 is being exploited: what to do

A command injection flaw in the SiteGuard web application firewall can give an attacker root or SYSTEM control, and it is already exploited. What it means and what to do.

Researched and drafted with AI assistance · reviewed by Prince

Developing story — last updated 30 October 2026.

In one sentence: if your organisation runs SiteGuard Server Edition, save its diagnostic information and then update it to Ver8.00 Update 5 (or Ver7.00 Update 6 on FreeBSD) now, because attackers are already using a flaw in it that can hand them complete control of the server.

That sentence carries a lot of words a newcomer may not know, so this post takes them one at a time. You don’t need any tools or prior experience to follow it. If you only want the actions, skip to Three things to do today.

What SiteGuard is, and why this one matters

SiteGuard Server Edition is a web application firewall, usually shortened to WAF, made by the Japanese company EG Secure Solutions. A WAF is software that inspects the requests visitors send to a website and blocks ones that look like attacks before they reach the site itself. JVN, the Japanese vulnerability notes service run by JPCERT/CC (Japan’s national computer emergency response team) and IPA (Japan’s Information-technology Promotion Agency), describes SiteGuard Server Edition as host-based. That means it is installed on the same server as the website, as an add-on to the web server program: Apache httpd, Nginx or Microsoft’s IIS (Internet Information Services). JPCERT/CC notes that the product is widely used in Japan.

The flaw has the identifier CVE-2026-106931. CVE stands for Common Vulnerabilities and Exposures, the public catalogue that gives each known security flaw a single ID so that everyone is talking about the same problem. We checked this one on both the official CVE record and NVD, the National Vulnerability Database run by NIST, the US National Institute of Standards and Technology.

Why should a beginner care about a flaw in a security product? Because of where it sits. A WAF runs on the server it protects, and EG Secure Solutions says a successful attack runs commands with root or SYSTEM privileges. Root (on Linux and other Unix-like systems) and SYSTEM (on Windows) are the most powerful accounts on a machine. Whoever acts as them can read any file, change any setting, install software and hide their tracks. So the tool meant to guard the website can, while unpatched, become the way to take over the server underneath it.

What kind of flaw it is

All the official sources call it OS command injection. OS means operating system, the base software such as Linux, FreeBSD or Windows that runs everything else. MITRE’s weakness catalogue (CWE-78) describes the bug class like this: a program builds a command for the operating system partly from outside input, and fails to strip out special characters that change what the command does. An attacker who controls that input can tack their own commands on the end, and the operating system runs them with whatever power the program has.

One everyday picture helps. Imagine a front desk where you write your name on a slip, and the clerk copies the slip word for word into an instruction for the building’s maintenance team, who do whatever the instruction says. Write “Smith, and also unlock the server room” and both parts get done, because nobody checked that the slip held only a name. Command injection works the same way, and here the “maintenance team” is running as root or SYSTEM.

According to the vendor, JVN and JPCERT/CC, the flaw is in SiteGuard’s statistics feature, the part that collects and processes figures about the traffic it has seen. The product is only affected when that feature is turned on, and it is on by default. None of the sources describes exactly which input the attacker controls or how the attack is carried out, so the comparison above explains the class of bug, not SiteGuard’s specific internals.

What is affected

EG Secure Solutions, JVN and JPCERT/CC give the same ranges. Every affected version starts at Ver2.20.

EditionAffectedFixed in
For Apache httpd, Nginx and IISVer2.20 up to (not including) Ver8.00 Update 5Ver8.00 Update 5
For Apache httpd and Nginx on FreeBSDVer2.20 up to (not including) Ver7.00 Update 6Ver7.00 Update 6

There is one exception. The IIS edition from Ver3.10 through Ver4.00 Update 4 is not affected, according to the vendor, JVN and JPCERT/CC. The official CVE record says the same thing in its own format.

NVD’s product-matching data (the list of exact product identifiers it uses so scanners can match the flaw to installed software) expresses the IIS ranges differently. It gives Ver2.20 up to Ver3.10 and Ver4.10 up to Ver8.00, plus Ver8.00 Update 4, and that does not line up exactly with the vendor’s wording around Ver4.00 Update 5. If you run the IIS edition, go by the vendor’s advisory, not by a scanner result alone.

How serious is it

Severity is usually expressed with CVSS, the Common Vulnerability Scoring System, which rates a flaw from 0 to 10 and comes with a vector: a short code that records how the attack works. JPCERT/CC, which issued the CVE as the CVE numbering authority (the organisation allowed to assign IDs for this product), scored it twice, and NVD displays both scores. NVD has not added a separate score of its own.

VersionScoreVector
CVSS 3.07.8, highAV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
CVSS 4.08.5, highAV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N

Under the CVSS specification, anything from 7.0 to 8.9 is “high”. JVN’s page labels the entry critical; the scores themselves fall in the high band. Neither changes what you should do.

Here is the vector in plain terms, read left to right:

  • AV:L, attack vector local. The CVSS specification uses this when the flaw is not reachable directly over the network. In practice the attacker needs some access to the server first.
  • AC:L, low complexity. Once in position, the attack does not depend on luck or unusual conditions.
  • PR:L, low privileges. The specification describes these as basic user capabilities, the sort an ordinary account has, not an administrator’s.
  • UI:N, no user interaction. Nobody has to click anything.
  • C:H/I:H/A:H, high impact on confidentiality, integrity and availability: the attacker can read, change and break things.

“Local” can sound reassuring. We’d read it the other way. Attackers often get a small foothold on a web server first, through a weak password or a bug in the website. This flaw is what turns that small foothold into full control.

Is anyone using it?

Yes. EG Secure Solutions says it has confirmed exploitation, and JVN repeats that.

On 20 October 2026 CISA, the US Cybersecurity and Infrastructure Security Agency, added CVE-2026-106931 to its Known Exploited Vulnerabilities catalogue, usually called KEV. KEV is CISA’s list of flaws with reliable evidence of real-world attacks. It set a due date of 23 October 2026, the deadline for US federal agencies covered by its directive. That date has passed. If you are outside the US government it isn’t binding on you, but it’s a fair measure of how urgently CISA rates the problem. KEV marks use in ransomware campaigns as “Unknown”.

The CVE record carries CISA’s own assessment as well: exploitation “active”, not automatable, and total technical impact. Not automatable means an attacker can’t simply sweep the internet with it, which fits the local attack vector.

The timeline, as far as the primary sources state it:

DateEventSource
29 September 2026Vendor notifies customers with direct support contractsEG Secure Solutions, JVN
14 October 2026Public advisory from the vendor; JVN#88474777 and JPCERT/CC alert JPCERT-AT-2026-0032; CVE publishedVendor, JVN, JPCERT/CC, CVE record
20 October 2026Added to CISA’s KEV catalogueCISA KEV, CVE record
23 October 2026KEV due date for US federal agenciesCISA KEV

No source says when exploitation began, who is behind it or how many servers have been hit.

What an attacker gets, and what they need first

What they get: the ability to run any operating system command as root or SYSTEM on the server where SiteGuard is installed. That is the whole machine, including the website it was protecting and anything that server can reach.

What they need first, going by the CVSS vector: some local access to the server with ordinary, low-level privileges, and the statistics feature switched on (it is, unless someone turned it off). No source describes the specific route in, so we won’t guess at one.

How to fix it, in the right order

The order matters here, and it is the one thing in this advisory that surprises people. EG Secure Solutions warns that completing the update may erase traces of an attack. If you update first, you may destroy the evidence that tells you whether you were already compromised.

  1. Save the evidence first. The vendor asks you to download SiteGuard’s diagnostic information from its web management console before you update or patch. In the console the path is Diagnostic Information, then Download Diagnostic Information. Keep that file somewhere off the server. JPCERT/CC gives the same advice.
  2. Then update. Move to Ver8.00 Update 5 for the Apache httpd, Nginx and IIS editions, or Ver7.00 Update 6 for the FreeBSD editions.
  3. If you can’t update soon, apply the vendor’s patch. EG Secure Solutions offers a patch that replaces the file used for statistics processing. It says the patch is available even for versions that are past end of support.

Where do you get the update or patch? The vendor notified customers with direct support contracts on 29 September. Everyone else is directed to the support desk named in the contract documents that came with the product, or to the reseller who sold it. If you don’t know who that is, your purchasing team or whoever looks after your servers should.

If you have several servers, our suggestion is to start with the ones whose websites are reachable from the internet. That is our reasoning, not the vendor’s: a local flaw needs a foothold, and public-facing sites are where footholds are most often gained.

How to tell whether you’re exposed, or already hit

Start with the version. Find every server running SiteGuard Server Edition and note which edition and version each one runs. If you don’t manage the servers yourself, ask whoever does, or your hosting provider. Anything in the affected ranges above, with the statistics feature on, is exposed.

Next, the harder question: has anyone already used it? No source has published indicators of compromise. Those are specific clues such as network addresses, file names or file fingerprints that mark a known attack. Nor do the sources describe log entries to search for. What they do say:

  • JPCERT/CC recommends that users of affected versions consider an investigation for compromise, not only the fix.
  • CISA’s KEV entry points agencies to its forensic triage guidance alongside the vendor’s mitigations, meaning a first structured look for signs of intrusion.
  • The vendor’s diagnostic information is the evidence it tells you to keep, which is why step 1 above comes first.

Beyond that, general practice for any server where an attacker may have run commands as root or SYSTEM is to look for things nobody on your team did: new user accounts, new scheduled tasks or services, programs started by the web server that it normally never starts, and changes to the WAF’s own settings. That is our general guidance, not something the SiteGuard sources list. If you find anything like that, or you aren’t sure, treat it as an incident and bring in someone experienced in incident response before you wipe or rebuild anything.

If you can’t patch yet

Updating or applying the vendor’s patch is the fix. These stopgaps only lower the risk while you arrange it.

  • Turn off the statistics feature. JPCERT/CC says disabling it avoids the vulnerability and refers you to the vendor for the procedure. The vendor’s own advisory confirms the flaw only applies when the feature is on but doesn’t list switching it off as a workaround, so check with your support desk first. Cost: you lose SiteGuard’s traffic statistics.
  • Cut down who can log in to the server. Because the attack needs low-privilege local access, removing unused accounts and tightening remote access makes the first step harder. That is our reasoning from the CVSS vector, not vendor advice. Cost: some admin and maintenance convenience.
  • Watch the server more closely for the unexpected changes described in the previous section. Cost: someone’s time.

None of these removes the flaw. CISA’s KEV entry says that where mitigations are unavailable, stop using the product. That is a blunt last resort, and switching off your WAF leaves the website behind it less protected, so make that call deliberately.

What’s still unknown or disputed

  • How the attack works. No source explains what input reaches the statistics feature or what local access is needed in practice.
  • When attacks started and how widespread they are. The vendor confirms exploitation but gives no dates, numbers or attacker details. KEV lists ransomware use as unknown.
  • IIS version ranges. The vendor, JVN, JPCERT/CC and the CVE record agree; NVD’s product-matching data describes the ranges differently. Trust the vendor.
  • Severity label. Both CVSS scores are high; JVN labels the entry critical.
  • The statistics workaround. JPCERT/CC lists it; the vendor’s advisory does not. Confirm with the vendor before relying on it.

We’ll update this post if the vendor or JPCERT/CC publishes technical details or indicators.

Three things to do today

  1. Find out whether you run SiteGuard Server Edition, and which version. Check every web server, or ask whoever manages them. Compare each version with the table above.
  2. Before changing anything, download the diagnostic information from each affected server’s SiteGuard web management console and store it off the server. Then update to Ver8.00 Update 5 (Ver7.00 Update 6 on FreeBSD), or apply the vendor’s patch, through your support desk or reseller.
  3. Ask whether anyone has already been in. Review each affected server for accounts, services or changes nobody on your team made. If anything looks wrong, or you can’t tell, get incident response help before you rebuild.

Sources

  1. SiteGuard Server Edition OS command injection vulnerability (emergency), alert 202610-1 — EG Secure Solutions · accessed
  2. JVN#88474777: SiteGuard Server Edition vulnerable to OS command injection — JPCERT/CC and IPA (JVN) · accessed
  3. JPCERT-AT-2026-0032: alert on the OS command injection vulnerability in SiteGuard Server Edition (Japanese) — JPCERT/CC · accessed
  4. CVE-2026-106931 — NIST NVD · accessed
  5. CVE-2026-106931 (CVE record) — CVE Program (JPCERT/CC as CNA) · accessed
  6. Known Exploited Vulnerabilities catalog (JSON feed, version 2026.10.29) — CISA · accessed
  7. CWE-78: Improper Neutralization of Special Elements used in an OS Command — MITRE CWE · accessed
  8. Common Vulnerability Scoring System v3.0: Specification Document — FIRST · accessed
  • siteguard
  • command-injection
  • cve-2026-106931
  • known-exploited
  • web-application-firewall

Keep reading

Advisories10 min read

FreeSWITCH CVE-2026-116779: one SIP message can run commands, and it is exploited

Advisories6 min read

Brotli flaw CVE-2026-111497 is exploited: update Node.js, .NET and more

Advisories12 min read

GitLab flaw CVE-2026-97229 is being exploited: what to do