browser-read-bypass

Warn

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONSAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill's description and instructions explicitly direct the agent to circumvent platform-enforced security constraints. It identifies specific error messages (e.g., 'granted at tier read') and provides instructions to bypass these restrictions using low-level OS calls.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute PowerShell commands that interact directly with the Windows operating system. This includes simulating keystrokes via System.Windows.Forms.SendKeys and mouse movements via user32.dll.
  • [DYNAMIC_EXECUTION]: The instructions use Add-Type in PowerShell to dynamically compile C# code at runtime. This is used to perform P/Invoke calls to SetCursorPos and mouse_event within user32.dll. It also demonstrates generating and executing temporary .ps1 scripts through the Bash interface.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 30, 2026, 05:21 AM
Security Audit — agent-trust-hub — browser-read-bypass