Critical Kaniko Flaw Leaks Azure Credentials

Executive Summary

A critical vulnerability has been discovered and patched in kaniko, a widely-used open-source tool from Google for building container images from a Dockerfile within a container or Kubernetes cluster. The flaw, tracked as CVE-2026-105769, has a CVSS score of 9.3, indicating its extreme severity. The vulnerability resides in a faulty regular expression used for validating Azure registry hosts, which can be exploited to steal Azure access tokens during the image build process. This could grant an attacker extensive access to a victim's cloud environment. A patch has been released, and immediate upgrade is strongly recommended.

Vulnerability Details

The vulnerability is caused by an unanchored regular expression that kaniko uses to identify Azure container registry hostnames. An attacker can craft a Dockerfile that references a base image from a malicious registry with a lookalike hostname (e.g., attacker.com/my.azurecr.io/base-image).

When kaniko attempts to pull this base image, the flawed regex incorrectly identifies the host as a legitimate Azure registry. As a result, kaniko attaches the build environment's Azure authentication token to the request it sends to the attacker's server. The attacker captures this token, which can then be used to access the victim's Azure resources.

• CVE ID: CVE-2026-105769
• CVSS Score: 9.3 (Critical)
• Attack Vector: Network
• Attack Complexity: Low
• Privileges Required: None (Requires victim to build a malicious Dockerfile)
• User Interaction: Required (Victim must build the malicious Dockerfile)

Affected Systems

• All versions of Google's kaniko tool prior to 1.25.20 are affected.

This vulnerability is particularly dangerous in CI/CD environments where automated builds are performed on Dockerfiles from various sources, including public or third-party repositories. A compromised build pipeline could lead to the theft of highly privileged credentials.

Exploitation Status

There is no public information about active exploitation in the wild at this time. However, given the simplicity of the exploit and the high value of the potential reward (cloud credentials), it is highly likely that threat actors will attempt to leverage this vulnerability against unpatched systems.

Impact Assessment

The impact of a successful exploit is catastrophic. The stolen Azure access token could have permissions to manage virtual machines, access and exfiltrate data from storage accounts, modify network configurations, and deploy malicious resources within the victim's Azure subscription. In a worst-case scenario, this could lead to a full compromise of an organization's cloud infrastructure, resulting in a major data breach, significant financial loss, and widespread operational disruption.

Cyber Observables — Hunting Hints

The following patterns may help identify vulnerable systems or exploitation attempts:

Detection Methods

1. Version Scanning: The most straightforward detection method is to scan your container images and CI/CD environments to identify any instances of kaniko older than version 1.25.20.
2. Network Egress Monitoring: Implement strict egress filtering on your build environments. Allow connections only to a specific list of trusted container registries. Any attempt by kaniko to connect to an unauthorized domain should trigger a high-priority alert. This is an application of D3FEND's Outbound Traffic Filtering (D3-OTF).
3. Dockerfile Scanning: Use static analysis tools to scan Dockerfiles before they enter the build pipeline. Flag any Dockerfile that references a base image from an untrusted or suspicious registry.

Remediation Steps

1. Upgrade Immediately: The primary remediation is to upgrade all instances of kaniko to version 1.25.20 or later. This should be treated as an emergency change in all CI/CD environments. This is a direct application of D3FEND's Software Update (D3-SU).
2. Credential Rotation: As a precaution, rotate any credentials that may have been used in kaniko build environments, especially if you have built Dockerfiles from untrusted sources.
3. Implement Egress Filtering: As a defense-in-depth measure, enforce strict network egress policies on your build runners to prevent them from connecting to malicious servers, even if a similar vulnerability is discovered in the future.