/

CVE-2026-11325

Severity:

HIGH RISK

(8.8)

OS Command Injection (CWE-78) in Cloudflare Pages Action

CVE-2026-11325

Awaiting Analysis

Vulnerability type

Category:

OS Command Injection

CWE:

CWE-78, CWE-1104

Impact:

Remote Code Execution and CI Secret Exposure

Severity:

HIGH RISK

(8.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A remote code execution vulnerability in the archived Cloudflare Pages Action may allow attacker-controlled workflow inputs or configurations to execute operating system commands during GitHub Actions runs. Successful exploitation could expose CI secrets, including deployment and repository tokens, when vulnerable workflow configurations invoke the action.

Affected context

The issue affects workflows that use the deprecated cloudflare/pages-action repository, including those referencing its moving v1 tag or other published versions. Risk depends on the workflow configuration, how action inputs are supplied, and whether sensitive credentials such as CLOUDFLARE_API_TOKEN or GITHUB_TOKEN are available to the job.

Exploit preconditions

  • Attack surface: GitHub Actions workflows invoking cloudflare/pages-action
  • Required access: Ability to influence relevant workflow inputs, repository content, or another trusted input consumed by the action
  • Trigger point: Execution of the affected action within a CI job
  • Why exploitation is possible: Insufficiently constrained input may reach an operating system command execution path, allowing attacker-controlled data to be interpreted by the shell or command runner

Why this happens

The vulnerability is consistent with CWE-78: Improper Neutralization of Special Elements used in an OS Command, where untrusted or insufficiently validated data reaches a system command interpreter. The use of an archived, unmaintained action also maps to CWE-1104: Use of Unmaintained Third-Party Components, which increases exposure because the repository is not expected to receive security fixes.

Potential impact

  • An attacker may execute arbitrary commands in a GitHub Actions runner
  • CI environment variables and workflow secrets could be exposed
  • Deployment credentials or repository tokens could be misused
  • Source code, build artifacts, or deployment targets could be modified
  • A compromised workflow could enable supply-chain tampering or lateral movement

General mitigation guidance

Mitigation generally focuses on removing the vulnerable action from workflow trust paths and reducing the amount of attacker-controlled data that can reach command execution.

Common defensive patterns include:

  • Migrate workflows from deprecated actions to a maintained, security-reviewed alternative
  • Avoid shell command construction from untrusted workflow inputs
  • Prefer argument-safe process APIs over shell evaluation
  • Restrict workflow permissions using least-privilege GITHUB_TOKEN settings
  • Scope deployment credentials to the minimum required environment and operation
  • Pin trusted action references according to organizational supply-chain policy
  • Rotate secrets that may have been exposed by affected workflow runs
  • Add approval boundaries for deployments originating from untrusted branches or pull requests

How teams detect issues like this

Organizations can identify this risk through:

  • Repository-wide searches for cloudflare/pages-action, including moving tags and indirect reusable workflows
  • Static analysis for shell execution, string interpolation, and command construction involving action inputs
  • Code review checks for untrusted pull-request data entering deployment or shell steps
  • CI policy rules that block deprecated or unmaintained actions
  • Audit-log review for unusual workflow executions, token use, deployments, or secret access
  • Automated dependency and workflow scanners that detect vulnerable or archived third-party actions

Sources

SourceLink
github.comgithub.com/cloudflare/wrangler-action

Frequently asked

questions

Exploitation can be triggered through GitHub Actions, but practical reachability depends on whether an attacker can influence workflow inputs, repository content, or another input consumed by the action.

Not necessarily direct administrative access. The required access depends on workflow triggers and trust boundaries; permissions to submit or modify content processed by a vulnerable workflow may be sufficient in some configurations.

Remove the deprecated action from affected workflows, replace it with a maintained alternative, prevent untrusted data from reaching shell commands, and rotate credentials that may have been exposed.

Enforce least-privilege workflow permissions, review action and shell-input boundaries, use trusted pinned dependencies, scan workflows for command-injection patterns, and prohibit unmaintained actions through CI policy.

Find out what's already
exploitable in your codebase.