prototype
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The UI prototyping branch (UI.md) utilizes URL search parameters to determine which component to render. This creates a surface where external, untrusted input influences the application's execution path.
- Ingestion points: The 'variant' parameter is retrieved from searchParams within the generated UI code (UI.md).
- Boundary markers: The instructions do not provide specific boundary markers or warnings to ignore instructions embedded within these parameters.
- Capability inventory: The agent is instructed to generate UI components and routing logic (UI.md), and to modify project configuration files (SKILL.md).
- Sanitization: The instructions do not specify validation or sanitization for the URL parameter before using it in conditional rendering logic.
- [COMMAND_EXECUTION]: The skill directs the agent to modify existing project task runners (e.g., package.json, Makefile, or pyproject.toml) to add execution commands for the generated prototypes. While standard for developer tools, this represents a modification of the project's build and execution environment.
Audit Metadata