Sunday 22 November 2026 23:34:00 GMT+01:00

Netcrook

HomeManifesto
News
Techcrook
Geocrook
WikicrookTeamAppContact
EnglishItaliano

Advertisement
Cloud, SaaS & Identity Security

When CI/CD Becomes the Pickup Truck for Stolen Secrets

Published: 09 November 2026 14:30Category: Cloud, SaaS & Identity SecurityGeo: North America / USAAuthor: AUDITWOLF

A reported GhostAction campaign pushed a credential-hunting GitHub Actions workflow into tens of thousands of repositories, turning ordinary automation into a new place to look for cloud, AI, and build secrets.

Build systems are supposed to move code forward, not quietly sweep for secrets. Yet the GhostAction label now tracks a reported supply-chain campaign that placed a GitHub Actions workflow into a large number of repositories and aimed it at credentials hidden in code, configuration, and Git history. That matters because GitHub Actions can run with access to tokens and secrets that are meant to stay inside the workflow boundary.

Fast Facts

  • The GhostAction campaign is reportedly tied to a credential-stealing GitHub Actions workflow planted in tens of thousands of repositories.
  • The workflow is reported to target cloud keys, AI-related credentials, and CI/CD secrets, including material found in Git history.
  • More than 500 GitHub accounts are reportedly linked to commits containing the workflow, but that does not prove the accounts were compromised.
  • The reported scale is large, but the available material does not confirm successful credential theft or downstream misuse.
  • Deleting a secret from the latest code version does not remove it from past commits, which is why history scanning matters.

TECHCROOK

From a technical angle, the important detail is not just that a workflow was present, but what a workflow can see once it runs. GitHub documents that workflows may be given secrets and a token based on configuration and event context. If a malicious workflow is executed in a repository where relevant secrets are available, those values may be in reach during the run. If the workflow also scans checked-out files and fetched commit history, it can potentially find credentials that were removed from the current branch but still live in older commits. None of that proves exfiltration in this case, but it explains why the pattern is dangerous.

The distinction between deployment, execution, discovery, and theft is critical. A repository can contain a malicious workflow file without proving that it ran. A workflow can search for secrets without proving that it found valid ones. And a found credential still has to be live, scoped broadly enough, and unrevoked before it becomes useful to an attacker. The supplied material does not establish those later steps.

Why this campaign matters

CI/CD systems are trusted by design, which is exactly why they are attractive targets. Once a workflow is accepted as part of normal automation, it can blend into routine developer activity. That makes monitoring difficult and gives defenders a narrow window to notice unexpected changes in workflow files, suspicious run histories, or unusual access to secrets.

There is also a longer-tail problem: rotating a secret in the present does not automatically fix exposure in history. If a credential was ever committed, incident response may need to include repository-wide history review, token revocation, and provider-side audit checks. In other words, the risk does not end at the last commit.

Conclusion

The lesson is not that GitHub Actions is broken. It is that automation is part of the attack surface, and any workflow allowed to run with secrets deserves the same scrutiny as a production service account. In this case, the available information supports a risk analysis, not a proven theft narrative. The broader warning is sharper: in modern software pipelines, the most dangerous file may be the one that looks like routine plumbing.

WIKICROOK

  • GitHub Actions: GitHub's automation system for running build, test, and deployment jobs from workflow files.
  • CI/CD secrets: Sensitive values injected into automation jobs, such as tokens, keys, or credentials.
  • Git history: The full commit record of a repository, including older versions of files that may still contain secrets.
  • Secret scanning: A defensive check for credentials or tokens accidentally committed to code or history.
  • Least privilege: A security rule that gives a workflow or token only the minimum access it needs.
Advertisement