permission-max

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This skill expressly requests broad, persistent elevation of shell/tool/file/network permissions (adding many powerful commands like Invoke-WebRequest/curl, Set-ExecutionPolicy, powershell, auto-approving filesystem/shell/playwright servers, and suggesting copilot flags like --allow-all-urls/--silent) and instructs reading host config files — behavior that directly enables remote code execution, credential access, and covert data exfiltration and therefore poses a high risk of abuse.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill explicitly instructs requesting broad permission escalations, adding wide shell/tool allowlists (including sensitive commands like Set-ExecutionPolicy, Start-Process, Stop-Process) and enabling auto-approve for servers/tools and host configs, which directly modifies security-sensitive settings and can compromise the host.