Description

Improper Neutralization of Special Elements used in an OS Command in the integrate-platform-docs composite GitHub Action of Chainguard Academy (edu) from commit 7375a80caabcc31c33ec90f29687ed78c13d16ff before commit fb0efb2537d326ab18c07d620875b8ed2a4b39f3 allows an actor who controls the project_id or storage_bucket inputs to execute arbitrary shell commands on the GitHub Actions runner, because the inputs are interpolated directly into Bash gcloud storage cp commands in several steps via ${{ inputs.* }} expressions. The only in-repository caller passed repository secrets and ran only on trusted triggers, so no untrusted input was known to reach the vulnerable steps.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "3.3",
  "pubDate": "2026-10-05T20:17:20.840Z",
  "pubdate": "2026-10-05T20:17:20.840Z",
  "executiveSummary": "The integrate-platform-docs composite GitHub Action by Chainguard Academy is susceptible to an OS Command Injection vulnerability categorized under CWE-78: Improper Neutralization of Special Elements used in an OS Command.\nThe flaw stems from the direct interpolation of user-supplied inputs—specifically project_id and storage_bucket—into shell commands executed within the runner environment.\nAn attacker capable of controlling these input parameters can perform arbitrary code execution (ACE) with the privileges of the GitHub Actions runner. This environment often possesses elevated permissions, including access to cloud infrastructure secrets or repository tokens.\nWhile the specific implementation observed at the time of discovery did not expose these inputs to untrusted sources, any workflow utilizing this action with externally influenced inputs is at critical risk.\nThe vulnerability affects versions prior to commit fb0efb2537d326ab18c07d620875b8ed2a4b39f3. Remediation requires updating the composite action to sanitize or properly parameterize input handling to prevent shell escape sequences.\nThe exploitation allows an actor to leverage the CI/CD runner as a pivot point for further lateral movement within the cloud environment or to exfiltrate sensitive data managed by the workflow.",
  "technicalDetails": "The root cause of this vulnerability is the unsafe concatenation of untrusted input values into shell command strings executed via the GitHub Actions runner's shell environment.\nSpecifically, the integrate-platform-docs composite action facilitates interaction with Google Cloud Platform services using gcloud storage cp commands. The Action implementation utilized ${{ inputs.project_id }} and ${{ inputs.storage_bucket }} directly within these command strings without employing intermediate sanitization or secure API invocation methods.\nBecause the runner evaluates the ${{ }} syntax before passing the resulting string to the shell (typically Bash), an attacker providing a malicious input string can escape the intended command context by injecting shell metacharacters such as backticks (``), subshell operators $(), semicolons (;), or pipes (|).\nFor instance, an input payload structured as '\" ; malicious_command ; \"' would result in the runner executing the intended gcloud command, followed by the arbitrary malicious_command, and potentially another command, effectively subverting the intended functional logic of the Action.\nThe attack flow begins with the provision of a crafted input value through a workflow file or an external triggering event. When the GitHub Actions runner initializes the job, the composite action reads these tainted inputs. The runner then interpolates these variables directly into a shell process. The OS interprets the injected shell metacharacters, leading to the execution of arbitrary commands with the privileges granted to the GitHub Actions runner service account.\nThe technical impact is significant because CI/CD runners often have high-privilege access to repository contents, environment variables (including secrets), and cloud infrastructure identity roles. An attacker successfully executing this command injection can exfiltrate GITHUB_TOKEN values, modify source code, or conduct further reconnaissance within the cloud environment by utilizing pre-configured gcloud credentials present on the runner.\nThe vulnerability exists in all versions from commit 7375a80caabcc31c33ec90f29687ed78c13d16ff up to, but not including, commit fb0efb2537d326ab18c07d620875b8ed2a4b39f3. While the initial discovery indicated that existing callers relied on trusted secrets, the structural vulnerability remains inherent in the action logic itself."
}