Our Thinking / Field Notes / The Security Watch

EVOLVING SECURITY ADVISORY / CISCO EXPRESSWAY SERIES CVE-2026-76466 HARD-CODED CREDENTIALS REMOTE CODE EXECUTION

Cisco Expressway Clusters at Risk: Hard-Coded Credentials Flaw Exploited

CVE-2026-76466 lets an unauthenticated attacker run root commands on clustered Cisco Expressway-C and Expressway-E and is now on CISA's KEV list. Who is affected, what to restrict today and how to reach X15.5.20.

What matters now

Current position: Critical (CVSS 3.1 base score 9.8). Unauthenticated remote command execution as root on clustered Expressway-C and Expressway-E. No workaround.

Affected versions: Clustered Cisco Expressway-C and Expressway-E (virtual and appliance) on X15.5.3 and earlier and X14 and earlier. Fixed in X15.5.20. Standalone devices are not affected by this CVE.

Observed exploitation: Listed in CISA KEV on 20 November 2026 (federal due date 23 November, forensic triage required, ransomware use unknown). F6 reported in-the-wild exploitation against one organisation. Cisco's advisory, last revised 21 October, still states PSIRT is not aware of malicious use.

Recommended next action: Restrict TCP 4371 and 4372 to cluster peers, preserve logs and snapshots, check for compromise, then upgrade all clustered nodes to X15.5.20 or later. X14 and earlier must migrate to a fixed release.

Identifiers: CVE-2026-76466

Immediate guidance

If you run Cisco Expressway-C or Expressway-E as a cluster on X15.5.3 or earlier, or on any X14 or earlier release, treat CVE-2026-76466 as urgent. Cisco rates it Critical (CVSS 9.8) and lists no workaround. CISA added it to its Known Exploited Vulnerabilities catalogue on 20 November 2026. Today: limit TCP 4371 and 4372 to your own cluster peers, preserve logs and snapshots, check for signs of compromise, then upgrade every node to X15.5.20 or later. X14 and earlier have no fix on their own branch and must migrate. Standalone Expressways are outside this CVE but still need the same upgrade for a separate October hardening advisory.

Current risk and what to do first

CVE-2026-76466 is a critical flaw in the clustering database service of Cisco Expressway. Cisco classes it as use of hard-coded credentials (CWE-798) and says an unauthenticated remote attacker with network access could run operating system commands as root. Root on a collaboration edge device is about as bad as it sounds, because that device sits between your internal network and the outside world.

This is no longer a theoretical risk. CISA added the CVE to the Known Exploited Vulnerabilities (KEV) catalogue on 20 November 2026 with a federal due date of 23 November, and the entry is marked as requiring forensic triage. The security firm F6 has also reported attacks against a real organisation. Cisco's advisory has not been revised since it was published on 21 October and still says Cisco PSIRT is not aware of malicious use, so the vendor page is behind the newer evidence. Act on the newer evidence.

  • Clustered Expressway-C or Expressway-E on X15.5.3 or earlier, or X14 and earlier: affected, patch or migrate to X15.5.20 or later.
  • Standalone Expressway: not affected by this CVE according to Cisco, but still within a separate October hardening advisory fixed by the same release.
  • No workaround exists. Limiting who can reach the cluster ports reduces exposure but does not remove the flaw.
  • If a node was reachable from untrusted networks before you restrict it, assume it needs checking for compromise, not just patching.

Which systems are affected and what fixes them

Cisco says the flaw affects Expressway-C and Expressway-E, virtual and appliance, but only when the device is running a vulnerable release and is configured in a cluster. Cisco's fixed-release table lists X15.5.3 and earlier as needing X15.5.20, and tells X14 and earlier customers to migrate to a fixed release. The table does not spell out releases between X15.5.3 and X15.5.20, so the safe working rule is that anything below X15.5.20 should be upgraded.

Cisco also published a separate hardening advisory on 21 October (updated 27 October) covering CVE-2026-76490 to CVE-2026-76497, rated High with a top score of 8.6. It applies regardless of configuration, has no workaround and is also fixed in X15.5.20. It does not cover CVE-2026-76466, but one upgrade deals with both, so there is no reason to stage them separately.

What the flaw is

Cisco attributes the issue to persistent cookies used by the cluster database service, combined with insufficient client certificate validation when TLS Verify Mode is set to Enabled for clustering. An attacker holding those cookies can connect to the database service and invoke specific remote procedure calls that end in command execution as root.

Cisco's advisory does not name the network ports. Independent technical write-ups and F6's guidance point to TCP 4371 and 4372 as the cluster database traffic, and F6 recommends blocking them except from trusted cluster node addresses. Treat those port numbers as well supported but not stated by Cisco, and check them against your own cluster design before writing firewall rules.

A public proof-of-concept exploit appeared on GitHub in early November, which is what turns a serious flaw into an everyday one. Cisco has not said how the cookies are obtained or whether they are the same across devices.

What is known about exploitation

Two separate pieces of evidence now exist, and they should be kept apart. CISA's KEV listing is the authoritative signal that exploitation has been observed. It does not name victims, sectors or attackers, and shows known ransomware use as unknown.

F6's report, relayed by SecurityWeek, describes an attack on an unnamed organisation in which two internet-exposed Expressway devices on X15.5.3 were compromised and the attackers then moved to two further cluster nodes on the internal network. Reported follow-on activity included Python scripts that scanned for SMB resources, interacted with SQL Server and WinRM, and searched GitLab repositories for credentials. F6 did not attribute the activity to a known group. Treat the wider picture as limited: there is no public count of victims, and no one should assume that only organisations matching this profile are being targeted.

Immediate actions, in order

These steps need administrator access to the Expressways and to the firewalls that sit around them. The upgrade in step 5 restarts each node, so book a maintenance window and expect calls and registrations handled by a node to drop while it reloads. In a cluster, upgrade one peer at a time. Steps 1 to 4 should not wait for that window.

  1. Establish whether you have a cluster and which release each node runs. Check the web interface overview page on every Expressway for the software version, and the clustering configuration page for other peer addresses. If peers are listed, the device is clustered and in scope. Do not forget lab, disaster recovery and forgotten pilot nodes.
  2. Restrict TCP 4371 and 4372 so that only your genuine cluster peers can reach them, on the network firewall and, if you use them, the Expressway's own firewall rules. Pay particular attention to any path from the internet, user networks or application networks. Afterwards, test from a machine you own that is not a cluster peer. A refused connection or a timeout is the result you want. Replace EXPRESSWAY_ADDRESS with the node's address.
    bash
    nc -vz -w 5 EXPRESSWAY_ADDRESS 4372
  3. Before changing anything else, preserve evidence from every peer. Download the diagnostic logs, take a configuration backup and, for virtual machines, take a snapshot that includes memory if your hypervisor supports it. An upgrade reloads the system and can overwrite what a responder would have wanted to see. This is our judgement rather than a Cisco requirement, and it matches the forensic triage expectation CISA attaches to this KEV entry.
  4. Look for signs of compromise on every peer, starting with the primary. Review the Erlang system logs for cluster nodes you do not recognise, admin accounts or configuration changes nobody made, and unexpected scripts or files on the device. Check firewall, proxy and flow records for connections to TCP 4372 from non-peer addresses and for Expressway addresses talking to internal SMB, SQL Server, WinRM or GitLab hosts, which a collaboration edge device would not normally do. Compare addresses against the indicators F6 published on its own site. Cisco has published no indicators, and logs may not show every attempt, so an empty result is reassuring but not proof.
  5. Upgrade every clustered node to X15.5.20 or later, or plan the migration if you are on X14 or earlier. Follow Cisco's upgrade guide for cluster order: the guidance we found says to back up first, upgrade the primary peer, then the others one at a time, and to upgrade the Expressway-E cluster before the Expressway-C cluster. Confirm that order against the release notes for your target version before starting. Allow time for each node to restart and rejoin; carry on only when it reports healthy.
  6. Verify the result. Every peer should report X15.5.20 or later, the cluster database should show as healthy on each peer, and the test from step 2 should still be refused from a non-peer machine. Keep the port restrictions in place afterwards, since cluster traffic has no reason to be reachable from anywhere else.

If you find signs of compromise

Do not rebuild straight away. Keep the evidence you collected, isolate the affected node from the cluster and the internal network where you can do so safely, and contact Cisco TAC and your incident response provider before wiping anything. Because attackers reportedly moved from edge nodes to internal cluster members and then on to other systems, scope the investigation beyond the Expressways: credentials stored on or reachable from the devices, and the internal servers they can talk to, need review. If personal data may be affected, speak to your data protection lead about reporting obligations early.

What is still unclear

Cisco has not updated its advisory to acknowledge exploitation, has not published indicators of compromise and has not named the ports or the source of the cookies. KEV confirms exploitation but not its extent. F6's report is the only detailed public account at the time of writing. Check Cisco's advisory and the KEV entry again before your change window, as both could change quickly.

Update history

We’ll keep this page updated as new information becomes available. Check back for the latest developments.

  1. 21 November 2026 initial update: CISA added CVE-2026-76466 to KEV on 20 November 2026, F6 reported exploitation against an unnamed organisation, and Cisco's advisory remained at version 1.0 from 21 October with no workaround and X15.5.20 as the fixed release. Update source

Technical terms explained
cluster
A group of Expressway nodes configured to share configuration and a database so they behave as one resilient system.
CVSS
Common Vulnerability Scoring System, a 0 to 10 scale for the technical severity of a vulnerability.
Expressway-C
The Cisco Expressway role that sits inside the internal network and connects to call control and other internal collaboration systems.
Expressway-E
The Cisco Expressway role placed at the network edge to handle traffic from the internet, such as remote workers and external partners.
forensic triage
A rapid, focused investigation of a system to decide whether it has been compromised, while preserving evidence for later analysis.
hard-coded credentials
A password, key or cookie built into software so that it is the same wherever the software is installed, which means anyone who learns it can use it.
Known Exploited Vulnerabilities (KEV)
A catalogue run by the US cyber agency CISA listing vulnerabilities that it has evidence are being exploited in real attacks.

Sources and further reading

KEEP THE CONVERSATION GOING

Need a hand with an Expressway estate?

If you would like a second pair of eyes on cluster exposure, firewall scoping or upgrade planning, get in touch.

Contact Rismor