Hackers Compromise 4,800 GitHub Accounts in Massive Credential-Theft Campaign

Hackers compromise thousands of GitHub accounts in a campaign dubbed wfx-loot that targets cloud credentials hidden in GitHub Actions secrets. Beginning around 12:45 UTC on November 10, the attacker uses stolen credentials from 4,837 compromised accounts to probe 61,764 repositories for write access by opening pull requests disguised as dependency audit notes. Researchers at Aikido confirm the threat actor successfully steals secrets from at least 1,835 targets.

The campaign reaches high-profile victims, including Fortune 100 organizations, the US General Services Administration, the French government, Nextcloud, and WebdriverIO. About 50 of the compromised accounts appear to hold publishing rights on major open-source ecosystems such as npm, PyPI, Go, crates, and Packagist, raising concerns about broader supply chain risk. Roughly 92% of the malicious pull requests remain open, suggesting most repository owners do not notice the activity.

The attackers deliver a 1,500-line credential-stealing shell script by modifying existing GitHub Actions workflows, adding a workflow on a short-lived branch, or planting an npm install hook. The script harvests GitHub tokens, AWS, Azure, and Google Cloud credentials, Git, SSH, Docker, Kubernetes, and Terraform secrets, and even AI service credentials for OpenAI, Anthropic, Gemini, and other developer tools. It also scans environment variables, .env files, private keys, shell histories, process memory, and workflow artifacts, while supporting cloud privilege escalation through GitHub OIDC token exchange.

Read More at the original source →