Follow us on Google News
A large-scale cyberattack has targeted more than 61,000 GitHub repositories in an attempt to steal sensitive credentials, cloud access keys and artificial intelligence service secrets.
Cybersecurity researchers confirmed that attackers successfully obtained confidential information from at least 1,835 targets through malicious GitHub Actions workflows.
The campaign, identified as wfx-loot, involved thousands of compromised GitHub accounts and affected repositories associated with major corporations, government organisations and open-source software projects. Investigators found that the attackers attempted to exploit trusted development systems to gain access to valuable credentials.
How Did Hackers Target Thousands of GitHub Repositories?
The attack began around 12:45 UTC on November 10, when cybercriminals used stolen credentials from 4,837 compromised GitHub accounts to examine 61,764 repositories for possible access.
The attackers created pull requests disguised as routine dependency audit notes, attempting to introduce malicious code into repositories where compromised accounts had sufficient permissions.
According to researchers, approximately 92% of these pull requests remained open, suggesting that many legitimate repository owners had not noticed the suspicious activity.
Some compromised accounts reportedly belonged to Fortune 100 companies, the US General Services Administration, the French government and technology organisations including Nextcloud and WebdriverIO. Researchers also identified approximately 50 accounts with publishing permissions across software ecosystems such as npm, PyPI, Go, crates and Packagist.
The attackers attempted to exploit these permissions to insert malicious workflows into trusted software development environments.
What Information Did the Attackers Steal?
Investigators identified a malicious shell script containing approximately 1,500 lines of code designed to collect credentials and confidential development information.
The programme targeted GitHub authentication tokens, workflow variables and credentials associated with major cloud platforms, including Amazon Web Services, Microsoft Azure and Google Cloud.
It also searched for developer credentials linked to Git, SSH, Docker, Kubernetes, Terraform and software package management services.
Significantly, the attackers targeted API credentials belonging to artificial intelligence platforms such as OpenAI, Anthropic, Google’s Gemini, Hugging Face, Cursor and GitHub Copilot.
These credentials could potentially allow unauthorised access to AI services and development tools. The malicious software also attempted to collect private keys, environment files, stored secrets and authentication information from affected development systems.
Researchers found that the programme could exploit existing cloud permissions to obtain additional access and establish persistent credentials.
After collecting the information, the attackers compressed it into files and transmitted it to infrastructure under their control. They then attempted to remove traces of their activity by reversing changes and deleting workflow records, sometimes completing the process in less than a minute.
What Are the Security Risks?
The investigation revealed that malicious code also reached software packages through automated publishing processes.
Affected releases included a Packagist package and two versions of a Go module, although researchers believe the primary objective was stealing continuous integration credentials rather than distributing compromised packages. The campaign showed similarities to the earlier GhostAction operation identified in 2025.
Security researchers have advised repository owners to examine unfamiliar pull requests, suspicious workflow changes and gaps in execution records. Where malicious activity is confirmed, organisations should immediately replace exposed credentials and review cloud permissions.
The incident highlights how compromised developer accounts can provide cybercriminals with access to multiple interconnected systems, placing cloud infrastructure, software supply chains and AI services at risk.
