Disabled GitHub Actions Reactivated, Resuming Malware Attacks

Two compromised GitHub Actions disabled since May came back online without cleanup on September 16, restarting Mini Shai-Hulud credential theft in downstream CI workflows.

@sharedot

Disabled GitHub Actions Reactivated, Resuming Malware Attacks

By @sharedot · · 6 pages

  • Programming
  • Devops
  • Supply Chain Security
  • Github Actions

Two compromised GitHub Actions disabled since May came back online without cleanup on September 16, restarting Mini Shai-Hulud credential theft in downstream CI workflows.

A disabled attack simply resumed

In May 2026, attackers injected credential-stealing code into two popular GitHub Actions — actions-cool/issues-helper and actions-cool/maintain-one-comment. The compromise was caught and the repositories disabled, which halted the campaign by breaking workflow fetches.

According to Rescana, on September 16, 2026 the repositories were re-enabled without the malicious release tags being removed, so any downstream workflow referencing them by tag immediately began executing the malware again — no new code, infrastructure, or attacker action required.

The payload harvested environment variables, secrets, and tokens from the CI/CD context and sent them to the attacker-controlled domain t.m-kosche[.]com, which Rescana says has been used for credential harvesting across multiple Mini Shai-Hulud campaigns.

The mutable tag trap

The surprising core of the incident is persistence: the attack survived its own discovery.

Rescana's analysis explains that the malicious code lived in mutable release tags such as v2.2.1, and any project referencing the action by tag rather than a pinned commit SHA would download and run the compromised code the moment the repository came back.

This maps to MITRE ATT&CK techniques T1195.002 (compromise of software dependencies and development tools), T1557 for credential harvesting, and T1041 (exfiltration over a C2 channel).

The Mini Shai-Hulud cluster, previously linked to npm compromises in the @antv ecosystem, effectively weaponized GitHub's default tag-mutability behavior to achieve long-term supply chain persistence with zero ongoing effort.

A credential-first attack wave

The reactivation fits a broader pattern Qualys documents across 2025–2026 campaigns: install-time or build-time code reaching developer and CI credentials, then pivoting into cloud infrastructure.

Qualys reports that the May 19 Mini Shai-Hulud wave compromised 639 package versions across 323 packages in the @antv ecosystem, including echarts-for-react with 1.1 million weekly downloads, targeting secrets such as GITHUB_TOKEN, AWS access keys, KUBECONFIG, and VAULT_TOKEN.

Qualys also describes related campaigns — TeamPCP compromising trusted tooling like Trivy and KICS, and typosquatting npm packages harvesting pipeline secrets — arguing that once malware runs in a CI environment, every credential that runner held must be treated as exposed, and package removal alone is not containment.

What teams should do now

Rescana recommends auditing every repository and workflow for references to actions-cool/issues-helper and actions-cool/maintain-one-comment, treating any tag-based usage (v2.2.1 or later) as compromised, and either removing the actions or pinning to a known-clean commit SHA predating May 18, 2026.

Teams should rotate all secrets and tokens the affected workflows could access, review run histories for successful runs following failures after September 16, and audit repository histories for unexpected post-reactivation commits.

Rescana and Qualys converge on the same preventive lessons: pin third-party actions to immutable commit SHAs, minimize long-lived credentials in builds, prefer short-lived OIDC-based credentials, restrict install-time scripts, and monitor cloud audit logs for the pivot that follows credential theft.

Sources

  1. rescana.com › Active Exploitation Alert: Compromised GitHub Actions Reactivated, Resuming Mini Shai-Hulud Malware Supply Chain Attacks in CI/CD Workflows
  2. blog.qualys.com › The Developer is the New Perimeter: How Supply Chain Attacks Are Becoming Cloud Breaches

More on Programming