RC RANDOM CHAOS

GitHub Token Compromise Fuels Massive Cloud Credential Theft

· via Dark Reading

Original source

'Gheist' Weaponizes GitHub Tokens to Steal Cloud Secrets

Dark Reading →

Gheist weaponizes compromised GitHub tokens to inject malicious workflows into thousands of repositories. The attack chain is straightforward: using a stolen token, attackers create branches and commit a workflow that captures repository secrets as environment variables, encodes the output, and exfiltrates it through GitHub’s public API before erasing the evidence. This approach bypasses network-based defenses since the data never leaves GitHub’s legitimate infrastructure.

The campaign has affected 13,500+ malicious branches across 850 compromised accounts, with successful penetration into private enterprise repositories that typically assume GitHub’s public-repository safeguards provide protection. Access tokens came primarily from infostealer malware and credentials disclosed in earlier supply chain breaches. Wiz observed attackers using stolen tokens from multiple accounts and automation systems to clone private repositories, extract cloud credentials from the same organizations, and enumerate AWS Secrets Manager—all within hours. Researchers estimate approximately a dozen organizations experienced serious impact, and the overlap with other abuse campaigns (malware distribution, GitHub Actions resource theft) suggests tokens are circulating and being reused among multiple threat actors.

Defense requires treating developer credentials with the same rigor as production secrets. Organizations should replace long-lived personal access tokens with fine-grained, short-lived alternatives or GitHub Apps, mandate code review for environment changes, and hunt for specific indicators: malformed “ci-sync-*” branches containing only a .github/workflows/sync-check.yml file with “ci: sync config” commit messages, workflow runs deleted within minutes, and the “gheist” user agent in audit logs.

Read the full article

Continue reading at Dark Reading →

This is an AI-generated summary. Read the original for the full story.