codex-windows-fast-patch-skill

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This skill intentionally tampers with local application binaries (MSIX/ASAR), injects/forces privileged request parameters (service_tier=priority), creates and trusts a development code-signing certificate to sign and reinstall modified packages, uses PowerShell ExecutionPolicy bypasses, and auto-updates from GitHub — behaviors that enable authorization bypass, persistence and supply-chain risk and can be readily abused as a backdoor or to escalate access even though no explicit data-exfiltration code is shown.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill documentation states that update-skill-from-github.ps1 is run automatically at runtime and requires network access to https://github.com (and suggests running git pull), which means remote repository content can be fetched and used to overwrite/execute local skill scripts at runtime — a runtime remote-code update dependency on GitHub.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). This skill explicitly instructs modifying and reinstalling system MSIX packages (unpack/repack/sign), running PowerShell with ExecutionPolicy Bypass, creating and importing a code-signing certificate into the machine trust store, and editing system/registry-level manifests—actions that alter the host system state and bypass security controls.