Back to Articles

SOC

Patched Is Not Clean: Ivanti Connect Secure, NetScaler and Compromise Assessment After the Fix

12 min read

In autumn 2026 Ivanti Connect Secure and Citrix NetScaler were exploited as zero-days, and CISA's KEV entries required forensic triage, not just patching. Why the update does not undo what an attacker took, what to check on the appliance and around it, which secrets to rotate and how to prove the fix.

ivanti
netscaler

Between August and October 2026, two of the most common remote access gateways in European companies, Ivanti Connect Secure and Citrix NetScaler, were exploited before or right after their vendors published fixes. In each case the CISA Known Exploited Vulnerabilities (KEV) catalog did something worth noticing: alongside "apply the update", it pointed to forensic triage requirements. The message is that an update closes the hole but does not tell you whether someone already came through it. This article reconstructs the autumn chain, explains why a patch does not remove what an attacker took, and lists what to check, what to rotate and how to prove the fix. It is part of our coverage of exploited edge devices, summarised in our analysis of the 2026 KEV entries.

Key Takeaways

  • Ivanti Connect Secure CVE-2026-104935 (unauthenticated command execution) was exploited from at least 1 October and added to KEV on 13 October 2026; two more chained zero-days followed into KEV on 30 October.
  • The attacker left no malware or web shell, re-exploited the device for each command and tampered with logs, so Ivanti's Integrity Checker Tool could miss it.
  • NetScaler had an authentication bypass in KEV on 9 September and two more exploited flaws on 27 September, with KEV notes requiring forensic triage.
  • What survives the patch is what the attacker took: passwords, session tokens, certificates, API keys and access to the network behind the appliance.
  • Proving the fix from outside is necessary but not sufficient. It confirms the hole is closed, not that nobody used it.

The October chain on Ivanti Connect Secure

On 13 October 2026 Ivanti published an advisory for CVE-2026-104935, an authentication bypass by spoofing (CWE-290) that lets a remote, unauthenticated attacker execute arbitrary commands, and CISA added it to KEV the same day. According to Rapid7, Ivanti stated that exploitation had occurred before disclosure against a limited number of customers, and that reliable indicators of compromise were not available.

Volexity's investigation, reported by BleepingComputer, showed that the actor it tracks as UTA0576 had chained the flaw with two other vulnerabilities, starting at least on 3 October, to gain unauthenticated root-level code execution. Defused reported exploitation of CVE-2026-104935 from 1 October. Ivanti fixed the two additional flaws at the end of the month:

CVEWhat it is (per CISA KEV)Added to KEV
CVE-2026-104935Authentication bypass allowing a remote unauthenticated attacker to execute arbitrary commands13 October 2026
CVE-2026-111357Missing authentication for a critical function, allowing a local authenticated attacker to escalate privileges30 October 2026
CVE-2026-118527Predictable identifiers allowing an authenticated attacker to upload files with attacker-controlled content to a known location30 October 2026

Two details matter for defenders. First, per Volexity, the attacker "deployed no malware or web shell": it re-exploited the device for every command it ran and tampered with logs to cover its tracks, evading Ivanti's Integrity Checker Tool. Second, CISA's alert states that these vulnerabilities were used for credential theft and data exfiltration, and that the Integrity Checker Tool may not detect exploitation of CVE-2026-104935 because the attacker's commands execute in memory. BleepingComputer cites Shadowserver tracking more than 5,000 Connect Secure instances exposed to the internet.

NetScaler in late summer

Citrix NetScaler ADC and Gateway had their own sequence. CVE-2026-19490, an authentication bypass through an alternate path (CWE-288) affecting appliances configured as a Gateway (SSL VPN, ICA Proxy, CVPN, RDP Proxy) or AAA virtual server, was published in a Citrix bulletin on 19 August 2026 with a CVSS 4.0 score of 9.3 and entered KEV on 9 September. On 27 September CISA added CVE-2026-88771, an input validation flaw allowing an unauthenticated attacker to execute arbitrary commands, and CVE-2026-88772, a memory buffer flaw that could allow remote code execution or denial of service. On 4 October it added CVE-2026-88779, a memory buffer flaw that could allow denial of service.

The KEV notes for the September flaws are explicit: "Running the provided IOCs in the NetScaler console may help identify indicators of exploitation. Customers must conduct forensic triage" under CISA's Binding Operational Directive 26-04. The directive binds US federal agencies, not European companies, but the reasoning applies to anyone: if a gateway was exposed and vulnerable during a period of active exploitation, you have to look for evidence, not just update it.

Why the patch does not remove what the attacker took

An update replaces vulnerable code. It does not change the state the attacker created or carried away. On a remote access gateway that state is particularly valuable:

  • Credentials. The appliance handles user logins and often stores service account credentials for LDAP or RADIUS. If they were read, they still work after the patch.
  • Sessions and tokens. Active VPN or SSO sessions captured during the exploitation window may remain valid until they expire or are revoked.
  • Certificates and keys. Private keys on the appliance can be used to impersonate it or decrypt traffic.
  • Configuration changes. New local accounts, modified authentication policies or added routes survive an in-place upgrade unless someone looks for them.
  • Footholds behind the appliance. If the attacker used the gateway to reach internal servers, the compromise now lives elsewhere: on a domain controller, a file server or a jump host.

In the Ivanti case the attacker left nothing on the device itself. That is the hardest scenario for an integrity check, and the reason the useful evidence is often outside the appliance: in authentication logs, directory changes and network traffic.

What to check

On the appliance

  • Run the vendor's integrity tools and IOC scripts, but do not treat a clean result as proof. CISA warns that Ivanti's tool may miss in-memory activity.
  • Compare the running configuration with a known-good backup from before the exploitation window: local accounts, authentication servers, policies, scripts, scheduled tasks.
  • Look for gaps or inconsistencies in local logs. Log tampering was part of the Ivanti activity.

Around the appliance

  • Off-box logs. Syslog, SIEM and NetScaler Console copies cannot be edited from the compromised device. Review them for the exploitation window, from the start of exploitation to the moment you patched.
  • Outbound connections from the appliance to unfamiliar addresses.
  • VPN logins after the window from unusual locations or for accounts that rarely use remote access.
  • Directory changes: new accounts, group membership changes, use of service accounts whose credentials were stored on the gateway.
  • Internal hosts the gateway can reach, starting with authentication servers, jump hosts and sensitive systems.

Citrix's guidance for a suspected compromise adds evidence preservation first: snapshot virtual instances, record time and NTP settings before isolation, preserve remote syslog and NetScaler Console logs, and keep in mind that evidence requirements may outweigh the wish to restore service quickly.

Rotating secrets on the appliance

If compromise is suspected, the vendors' recovery guidance converges on the same steps. For Ivanti, CISA describes backing up the configuration, performing a factory reset and upgrading to the latest version (rebuilding cloud and virtual instances), restoring the configuration, then:

  • resetting passwords for all accounts, including domain, administrator and local accounts;
  • revoking and reissuing stored certificates;
  • resetting stored API keys and license server credentials.

For NetScaler, Citrix lists:

  • changing on their source systems all service account passwords and secrets stored on the appliance, such as LDAP service accounts, RADIUS shared secrets, OAuth tokens, API keys and SNMP community names;
  • changing the passwords of user accounts that may have authenticated through the appliance;
  • revoking certificates and private keys stored on it;
  • rebuilding the instance, upgrading firmware before restoring a configuration backup that predates the compromise, then changing local passwords, rotating key encryption keys and replacing restored certificates;
  • monitoring the rebuilt system closely for at least 90 days, and never exposing the management interface to the internet.

Revoke active sessions as well, so that tokens captured during the window stop working.

Proving the fix from outside

After the update, confirm from the internet that the vulnerable version or behaviour is gone on every exposed instance, including the forgotten test gateway and the one at the subsidiary. This is where external verification helps, with one honest caveat: it proves that the hole is closed, not that nobody used it.

  • RECON External continuously discovers your internet-facing hosts and verifies findings from an isolated server. A finding that was Proven before the patch moves to Closed when the verification no longer reproduces it, with the evidence of each test. RECON confirms that a vulnerability exists or no longer exists; it does not exploit it.
  • Fireline, our autonomous external penetration test, goes further on the perimeter you authorise in a signed Authorisation to Test. After remediation, a proof-of-fix retest checks up to 50 findings from an earlier run again, with fresh evidence and a new report.

When to call an incident response team

Patch and verify on your own if the appliance was not exposed during the exploitation window, or if it was patched before exploitation began and off-box logs show nothing unusual. Call for help when:

  • the appliance was exposed and unpatched during the period of known exploitation;
  • the vendor's IOC tools return hits, or local logs show gaps;
  • you see VPN logins, account changes or outbound connections you cannot explain;
  • the KEV entry for your vulnerability flags forensic triage;
  • a regulatory clock may be running: under NIS2, the early warning to the national CSIRT is due within 24 hours of becoming aware of a significant incident, and personal data breaches have 72 hours under the GDPR.

Oversight, our SOC as a Service on the CYREBRO platform, includes post-incident analysis of verified incidents: how far the attack got, its root cause and the remediation steps, with the evidence behind every finding. With your sources connected, analysts on duty 24/7 watch for the next sign of activity instead of waiting for the next advisory. Our incident response plan guide explains how to prepare the people and decisions before you need them.

Checklist after an edge-device zero-day

  1. Identify every instance of the affected product exposed to the internet, including test and subsidiary appliances.
  2. Establish the exploitation window from the vendor, CISA and researchers, and compare it with your patch date.
  3. Preserve evidence before rebuilding: snapshots, off-box logs, configuration exports.
  4. Run vendor integrity and IOC tools, without treating a clean result as proof.
  5. Review off-box logs, VPN logins, directory changes and outbound connections for the window.
  6. Rebuild or factory-reset if compromise is suspected, and upgrade before restoring configuration.
  7. Rotate every secret the appliance stored or saw, and revoke sessions and certificates.
  8. Verify the fix from outside and keep the evidence.
  9. Check whether NIS2 or GDPR notification duties apply.
  10. Prioritise future edge-device patches by exploitation, as described in our article on KEV, EPSS and SSVC.

If you are not sure whether your gateway was reached before you patched, talk to us. Oversight can help reconstruct what happened and watch for what comes next, and RECON External and Fireline can show, with evidence, that the door is now closed.

ivanti
netscaler
vpn
compromise-assessment
kev
incident-response