prototype
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions involve executing project-level tasks using tools such as pnpm, python, and bun to run the generated prototypes.
- [DYNAMIC_EXECUTION]: The skill generates executable artifacts, including standalone HTML files with embedded JavaScript and React components, designed to be run immediately for design exploration.
- [INDIRECT_PROMPT_INJECTION]: The UI prototype branch instructs the agent to build a component that switches layouts based on a URL parameter (?variant=). This creates a vector for untrusted data to influence the generated code's behavior.
- Ingestion points: The switcher component reads from URLSearchParams in the browser context as described in UI.md.
- Boundary markers: No specific delimiters or security warnings are defined to separate user input from the rest of the generated application code.
- Capability inventory: The skill allows for writing files to the local file system and running shell commands via the project's task runners.
- Sanitization: The instructions do not require the generated code to validate or sanitize the URL parameter values.
Audit Metadata