GitHub Repositories Targeted in Command Injection Campaign, Highlighting CI/CD Supply Chain Risks

A recent command injection campaign has targeted 61 public GitHub repositories, with attackers leveraging 83 malicious issues, pull requests, and comments to execute commands within GitHub Actions workflows. Radware researchers identified that these injected commands were successfully executed in at least two GitHub Actions workflows, demonstrating a significant vulnerability in the software supply chain. The campaign, which saw a rapid succession of malicious items posted within minutes, indicates a coordinated effort to exploit CI/CD systems. This incident is highly significant for DevSecOps practitioners because it illustrates how attackers are increasingly focusing on the CI/CD pipeline as a critical attack surface. The ability to inject commands through seemingly benign actions like comments and pull requests means that traditional perimeter defenses are insufficient. This type of attack can lead to unauthorized code execution, data exfiltration, and credential theft, directly impacting the integrity and security of deployed applications. Organizations that rely heavily on GitHub Actions or similar CI/CD platforms are directly affected, as their automated workflows can become conduits for malicious activity. This development fits into the broader trend of escalating software supply chain attacks. Over the past few years, there has been a noticeable shift from targeting production environments directly to compromising earlier stages of the software development lifecycle, such as source code repositories and build pipelines. This is further exacerbated by the increasing adoption of AI in development, which, while accelerating processes, also expands the attack surface and introduces new security gaps like prompt injection and credential exfiltration. The focus on CI/CD systems is a natural progression for attackers seeking to embed malicious code or backdoors into legitimate software before it reaches end-users. Previous incidents, such as GhostAction credential theft and the tj-actions supply chain attack, have also highlighted the risks to CI/CD secrets. In practice, this means practitioners must adopt a more proactive and stringent security posture for their CI/CD pipelines. Concrete implications include the necessity for rigorous input validation on all user-generated content that interacts with CI/CD workflows, such as pull request titles and comments. Maintainers should pass untrusted titles through environment variables instead of directly inserting them into shell scripts, avoid `eval` commands, and limit workflow permissions to the absolute minimum required. Furthermore, organizations should separate notification jobs from privileged release workflows to contain the blast radius of any successful injection. Regular auditing of workflow permissions, secrets, and outbound traffic is crucial, and any evidence of execution should trigger an immediate review of exposed credentials and audit logs. The overarching takeaway is that every interaction within the development pipeline, no matter how small, must be treated as a potential attack vector.