impl

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the [Name] placeholder directly in multiple PowerShell commands (e.g., .\scripts\build-and-install.ps1 -PluginName [Name]). Since this value is provided as user input during the skill trigger, it represents a command injection surface if the agent does not sanitize the input before shell execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from various project files including parameter-spec.md, ui-spec.md, and style-guide.md. This content is then interpolated into generated C++ source code and HTML/JavaScript files.
  • Ingestion points: $PluginPath/status.json, .ideas/parameter-spec.md, Design/v[N]-ui-spec.md, and Design/v[N]-style-guide.md.
  • Boundary markers: Absent; there are no delimiters or instructions to ignore embedded malicious content in the ingested files.
  • Capability inventory: The skill can write to the local filesystem (Source/ directory) and execute local PowerShell scripts.
  • Sanitization: Absent; the skill does not mention escaping or validating the content extracted from design documents before generating code.
  • [PRIVILEGE_ESCALATION]: The skill instructs the agent to execute PowerShell scripts using the -ExecutionPolicy Bypass flag. While common in development workflows, this technique is used to circumvent local security restrictions on script execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 01:17 AM