Security researchers have identified malicious code in dozens of packages distributed through Red Hat's official @redhat-cloud-services namespace on npm after attackers gained unauthorized access to the repository.
The incident was first reported by researchers at Aikido Security, who found that software packages published through the trusted Red Hat namespace had been modified to include malware capable of collecting credentials from developer environments. Because the affected namespace is used for legitimate Red Hat cloud-related packages, developers may have installed the compromised versions without suspecting unauthorized changes.
According to researchers, more than 30 package versions were affected. Several remained available for download when the activity was initially disclosed, creating a risk for organizations that automatically pull dependencies into development workflows.
Technical analysis showed that the malicious code was designed to run during package installation. This means exposure could occur as soon as a package is installed, even if the software itself is never executed inside an application.
Researchers found that the malware searched infected systems for authentication data commonly used by developers and cloud administrators. The targeted information reportedly included GitHub Actions secrets, npm access tokens, Kubernetes credentials, Vault secrets, and other cloud-service authentication material that could provide access to source code repositories, deployment environments, and internal infrastructure.
The malware also contained mechanisms intended to expand the compromise beyond the initial victim. If credentials with sufficient privileges were discovered, the malicious code could attempt to publish altered packages through repositories or accounts available to the infected environment. This behavior could allow attackers to use one compromised system as a stepping stone into additional software projects.
Investigators further observed that stolen information was encrypted before being transmitted from infected systems. Reports indicate that the malware included backup methods for data exfiltration, including the ability to use compromised GitHub repositories if its primary communication channel became unavailable.
Researchers noted signs that the incident may have involved CI/CD infrastructure. Continuous Integration and Continuous Delivery systems automate software building, testing, and deployment, making them attractive targets because a compromise can provide access to multiple projects simultaneously. Evidence reviewed by researchers suggested that GitHub Actions OpenID Connect workflows may have been involved in publishing the affected packages.
The exact method used to gain access to the Red Hat namespace remains under investigation. Researchers have not publicly attributed the initial compromise to a specific technique, although they believe unauthorized access to publishing credentials likely played a role.
Security firms examining the incident linked the malware to a variant of "Shai-Hulud," a credential-stealing program that has appeared in recent software supply-chain investigations. Researchers noted that code associated with the malware has circulated publicly, increasing the likelihood that similar attacks could be adopted by multiple threat actors.
Following notification of the issue, Red Hat removed the affected packages and began an internal investigation. In a public statement, the company said the compromised packages were intended for internal development purposes and were not distributed to customers through Red Hat production services. The company also stated that it had not identified evidence of impact to customer environments, partner systems, or production infrastructure at the time of its investigation.
Security experts recommend that any organization or developer who installed affected package versions review their systems immediately. Response measures should include rotating credentials, examining CI/CD environments for unauthorized activity, reviewing repository permissions, and checking software dependencies for indicators associated with the compromise.
The incident illustrates a recurring challenge in modern software development: trust placed in widely used package repositories can become a point of failure when an attacker gains access to a legitimate publishing channel. When that occurs, malicious code can reach downstream users through routine software updates rather than through traditional intrusion methods.
