The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on a TypeScript wrapper (scripts/ui_architect_wrapper.ts) that executes PowerShell scripts on the host environment using child_process.spawn. This provides a direct path for executing shell commands.
[PRIVILEGE_ESCALATION]: The wrapper script (scripts/ui_architect_wrapper.ts) explicitly bypasses standard security restrictions by applying the -ExecutionPolicy RemoteSigned flag when spawning PowerShell. This enables the execution of local scripts that might otherwise be blocked by the system's security policy.
[DYNAMIC_EXECUTION]: The skill employs several dynamic execution techniques:
In scripts/build_wpf.ps1, the use of [System.Windows.Markup.XamlReader]::Load() to parse UI definitions is a known vector for XAML injection, which can lead to arbitrary code execution if the XAML content is manipulated.
The scripts/create_winforms.ps1 script dynamically generates new PowerShell scripts based on input parameters and writes them to the file system using Set-Content.
The TUI logic in scripts/design_tui.ps1 uses the call operator & to execute scriptblocks provided in menu item definitions.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of external UI configurations and XAML templates.
Ingestion points: Untrusted data enters the skill through the XamlPath, Controls hashtables, and MenuItems parameters.
Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore embedded commands within the processed data.
Capability inventory: The skill possesses capabilities for file system modification and arbitrary command execution via PowerShell.
Sanitization: The scripts do not perform sanitization of the XAML content or the generated script strings to prevent command or object injection.

powershell-ui-architect