A significant cybersecurity breach has compromised thousands of GitHub accounts, allowing attackers to inject malware into numerous repositories. The malicious activity, which began on November 10, 2026, targets confidential information and cloud credentials stored within GitHub Actions, a widely-used tool for software development.
Extent of the Attack
The breach has affected 4,837 accounts, resulting in unauthorized pull requests in 61,764 repositories. So far, 1,806 repositories have been confirmed to have executed the malware, leading to the theft of sensitive credentials. Notably, 50 of the compromised accounts belong to users who publish popular packages across various platforms like npm and PyPI, escalating the risk to users who depend on these packages.
Security experts at Aikido have identified the malware and have noted that the primary goal of the attackers is to obtain continuous integration (CI) secrets rather than launching a widespread supply chain attack. However, the malware has already impacted one Packagist package and a Go module due to automated publishing systems.
Methods of Infiltration
The attackers utilized stolen GitHub tokens to manipulate repositories the tokens could access. Initially, they opened pull requests with the title “docs: dependency audit notes” from a branch named wfx-docs to verify write access. Over 61,839 such pull requests were identified, with many targeting repositories not owned by the compromised accounts.
Most of these pull requests remain open, indicating that many repository owners are unaware of the breach. The malware is deployed in several ways: by adding extensive lines of code to existing workflows, embedding hidden scripts, or using new workflow files on temporary branches. The attackers then swiftly remove traces of their activity, making detection challenging.
Impact and Recommendations
The malware, referred to as “wfx,” is designed to extract GitHub Actions secrets, runtime tokens, and credentials for cloud services and AI tools. It further attempts to escalate access by exchanging GitHub tokens for cloud service sessions, creating persistent access keys, and bypassing security measures like branch protection.
To mitigate the effects of this attack, it is crucial to search for suspicious branches, pull request titles, and commit messages. Affected users should revoke compromised tokens, rotate credentials, and review audit logs for unauthorized changes. Enhancing sign-in security and monitoring outbound traffic are also recommended practices.
Aikido advises treating certain packages and module versions as potentially compromised. Specifically, the Packagist package digitalelvis/neuronai-studio 3.14.3 should be considered hostile, and any systems that interacted with it should be scrutinized for compromise. While Go module users may not be directly affected, it is recommended to remove affected versions as a precaution.
Staying vigilant and informed about the latest threats is critical for preventing and responding to such incidents. Organizations should integrate threat intelligence to enhance their security operations and prevent future breaches.
