Cybersecurity

CVE-2026-76466 Cisco Expressway PoC Exploited in the Wild: 9.8 Root RCE

Published Updated By Shivam Sharma6 min read
Cisco logo on a light background with a warning about the critical CVE-2026-76466 remote code execution flaw

A public exploit is now available for the CVE-2026-76466 Cisco Expressway PoC, a critical flaw that lets an unauthenticated attacker run commands as root on clustered Cisco Expressway servers. Researcher Nicolas Krassas, known as dinosn, published the exploit and a fleet detector on GitHub, according to GBHackers. Netcrook notes the publication has not been independently confirmed in its review. Cisco rated the flaw 9.8 out of 10, published its fix on 21 October, and says there is no workaround.

Update, 21 November 2026: CISA added CVE-2026-76466 to its Known Exploited Vulnerabilities catalog on 20 November, based on evidence of active exploitation. Federal agencies must remediate it by 23 November 2026, and CISA lists known ransomware campaign use as “Unknown”.

Update — 19 November: first confirmed exploitation in the wild

The flaw is now confirmed exploited in the wild. Cybersecurity firm F6 investigated the first case: attackers breached two internet-exposed Expressway X15.5.3 devices at a Russian organisation, which F6 did not name, then pivoted to two internal cluster nodes. After gaining root, they prepared custom Python scripts for lateral movement — scanning the network, locating SMB resources, interacting with SQL Server and WinRM, and searching GitLab repositories — aiming to exfiltrate credentials. The scripts’ comments and fix descriptions show signs of AI-assistant use. F6 did not attribute the attack to any actor, and did not confirm that data was actually exfiltrated.

Detection: F6 points to the Erlang system log on Expressway devices. For organisations that cannot patch immediately, F6 recommends blocking TCP ports 4371–4372 at the firewall, except from trusted cluster nodes — the same ports this article already flags as the interim mitigation. Cisco’s advisory has not been revised since 21 October and still states Cisco is unaware of malicious use; SecurityWeek has asked Cisco for comment.

The timeline is now complete: Cisco published the fix on 21 October; public exploit code appeared six days later; a polished PoC followed 13 days after that; and the first confirmed breach surfaced 28 days after the patch. For a flaw with no workaround, unpatched clustered gateways had under a month from patch to observed exploitation. CISA added CVE-2026-76466 to its Known Exploited Vulnerabilities catalog on 20 November 2026, based on evidence of active exploitation.

What the CVE-2026-76466 Cisco Expressway PoC actually does

GitHub’s documentation locates the flaw in the cluster-manager module, where a hard-coded maintenance identity, wombat@, and a matching hard-coded cookie, wombat, are compiled into the software and identical on every affected installation. An attacker opens a TLS 1.3 connection to TCP port 4372, completes the Erlang handshake as wombat@127.0.0.1, answers the challenge with the hard-coded cookie, then calls rpc:call to run os:cmd, as the README describes the sequence. In the researcher’s test, the command id -u returned 0, meaning root. The researcher reports validation only on Expressway-C X15.3.2 in a two-node cluster with TLS Permissive mode; TLS Enforcing mode, Expressway-E, X14 releases and the fixed X15.5.20 build were not tested.

Only clustered deployments running a vulnerable release are affected: Expressway-C and Expressway-E, virtual and appliance, when configured in a cluster. Expressway is the system that lets remote workers and outside callers reach a company’s Cisco phone and video system, according to CyberExperts. Cisco’s advisory, published 21 October, says an unauthenticated remote attacker with network access can execute arbitrary operating system commands as root, and that there are no workarounds. The flaw, tracked as CWE-798 for hard-coded credentials, was reported to Cisco by Stefan Aleksic of Jane Street.

Cisco’s advisory describes the mechanism as persistent cookies plus insufficient client-certificate validation when TLS Verify Mode is set to Enabled for clustering; an attacker with the cookies connects to the database service and invokes specific RPC commands. One scope question is unresolved: the researcher tested TLS Permissive mode, the factory default, while Cisco describes the flaw as occurring when TLS Verify Mode is set to Enabled. Whether the two wordings describe the same setting across releases is not established in the sources read. The exploit script has probe, ping and uid modes, and ping mode authenticates only, without running commands.

The same repository also includes a detector for fleet-scale triage. It holds exploit, detector and docs directories. The flaw needs network access to the affected device, which is why restricting ports 4371 and 4372 to cluster peers matters even before the upgrade. Port 4372 is the documented exploit path; 4371 is the cluster-recovery port, included as additional hardening.

The CVE-2026-76466 Cisco Expressway PoC is not necessarily the first public exploit code for this flaw: AcidPeak first noticed public exploit code on 27 October, nine days before the Krassas commits on 5 November, so the 12 November news is GBHackers reporting the Krassas exploit to a wider audience. Cisco’s advisory was part of a wider 21 October patch batch, and a separate Expressway hardening release groups eight more flaws, up to 8.6, fixed in the same X15.5.20 release, according to CyberExperts.

How do I protect a clustered Expressway?

Upgrade clustered Expressway-C and Expressway-E on X15.5.3 and earlier to X15.5.20, and migrate X14 and earlier to a fixed release, because Cisco says there is no workaround. Until the upgrade is done, restrict TCP ports 4371 and 4372 to authorised cluster peers only as interim mitigation, and alert on connections from any other address.

  1. Inventory every clustered Expressway node, record its release, and confirm whether port 4372 is reachable from outside the cluster.
  2. Upgrade clustered Expressway on X15.5.3 and earlier to X15.5.20, and migrate X14 and earlier to a fixed release.
  3. Restrict TCP ports 4371 and 4372 to authorised cluster peers only, and alert on connections from any other address.
  4. Run the repository’s ping mode against each node, since it authenticates only and runs no commands.
  5. Watch CISA’s Known Exploited Vulnerabilities catalog, which still did not list the flaw as of 18 November.

What to watch next

The window between fix and public exploit code was short. Cisco published the fix on 21 October; AcidPeak first noticed public exploit code six days later, on 27 October; the Krassas commits followed on 5 November; and GBHackers reported the exploit to a wider audience on 12 November, 22 days after the patch. A 7 November ZoomEye search returned 39,423 Cisco Expressway fingerprints on the internet, a fingerprint count, not a count of vulnerable devices: it cannot tell whether a device is in the vulnerable configuration, so it does not bound the vulnerable population. The search ran 17 days after Cisco published the fix. AcidPeak gives an EPSS estimate of 0.42% over 30 days. Cisco knew of no malicious exploitation when it published the advisory, and Netcrook’s 12 November analysis found no public evidence in the material it reviewed confirming successful exploitation. That changed on 19 November: F6 investigated the first confirmed in-the-wild exploitation (see the update above), so the ‘actively exploited’ question is no longer unverified. Feedly’s ‘yes’ is no longer just an automated assessment — but the scale beyond this single incident is unknown, and CISA’s catalog still does not list the flaw as of 18 November.