Kubernetes ClusterRole Privilege Escalation via DataVolume Clone

In containerized-data-importer (CDI), the aggregated cdi.kubevirt.io:view ClusterRole, intended to provide read-only access to CDI resources, includes a rule granting create on the datavolumes/source subresource. CDI's DataVolume clone authorization accepts this permission as sufficient to authorize cloning the contents of any PVC the caller can name, without requiring write access to the source namespace. A user or service account bound to the view role, commonly granted cluster-wide via ClusterRoleBinding, who also has ordinary write access (edit/admin) to any single namespace, can use this to exfiltrate the contents of any PVC in the cluster into a namespace they control, bypassing namespace isolation and the read-only guarantee of the view role.

Casky was already ahead

This CVE exploits attack patterns that Casky's 283matched skills already investigate — long before this vulnerability was disclosed. Claude's reasoning model maps these techniques to MITRE ATT&CK, so practitioners who ran these skills have already seen the threat behaviour in their findings.

Analysis

CVE-2026-17527 exposes a critical authorization bypass in KubeVirt's Container Data Importer (CDI). The aggregated cdi.kubevirt.io:view ClusterRole, designed to provide read-only access, inadvertently grants create permissions on the datavolumes/source subresource. This permission is then misused by CDI's DataVolume clone authorization logic, which fails to validate that the caller has write access to the source namespace—allowing attackers to clone and exfiltrate any PVC they can name. The vulnerability is particularly dangerous because the view role is commonly distributed cluster-wide via ClusterRoleBindings, meaning numerous users and service accounts inherit this dangerous permission without explicit intent. Organizations running KubeVirt deployments are at risk of unauthorized data access and lateral movement within their Kubernetes infrastructure.

Casky's 283 matching skills leverage Claude AI's extended reasoning to detect the attack patterns mapped to MITRE Privilege Escalation (TA0004) and Credential Access (TA0006). Practitioners using Casky would observe findings related to excessive RBAC permissions—specifically flagging ClusterRole aggregations that grant create/write-like actions under read-only role names, and detecting authorization logic that accepts insufficient permissions for sensitive operations like data cloning. The platform would highlight service account activity patterns showing DataVolume clone requests originating from accounts bound to the view role, combined with cross-namespace PVC access attempts. These behavioral indicators, analyzed through Claude's reasoning engine, surface the gap between intended (read-only) and actual (data exfiltration) capabilities, enabling practitioners to identify and remediate over-privileged RBAC configurations before exploitation occurs.

Live Threat IntelligenceComing soon

Casky Risk Score

—

EPSS Probability

—

Shodan Exposure

—

GreyNoise Signal

—

Composite risk scoring from EPSS, CISA KEV, Shodan, and GreyNoise — 21 security APIs correlated into a single Casky Risk Score. Coming in Casky Pro. Join early access →

283 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-17527.

abusing-dpapi-for-credential-access

red teaming · high

Run

abusing-shadow-credentials-for-privesc

red teaming · high

Run

Access with Stolen Session Cookie

identity access management · low

Run

MITRE ATT&CK Techniques

TA0004 TA0006

Investigate the attack patterns behind CVE-2026-17527

Casky has 283 skills that investigate the attack patterns behind CVE-2026-17527. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn