break
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill performs legitimate developer-oriented tasks without malicious patterns. All detected behaviors are consistent with its primary purpose of UI testing and development.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an inherent attack surface because it processes untrusted local source code (components) to generate new test files. However, this is standard behavior for developer tools.
- Ingestion points: Local project component files (e.g., JSX, TSX) are read to infer props, slots, and data structures.
- Boundary markers: No explicit delimiter-based prompt protection is used for interpolation.
- Capability inventory: The skill has file-writing capabilities used to create temporary harness pages and routes within the application.
- Sanitization: The skill mitigates risks by explicitly instructing the agent to never wire the harness to live data and to avoid importing production state, ensuring data isolation.
- [DYNAMIC_EXECUTION]: The skill generates and saves a temporary web page or route (e.g., a Next.js 'use client' page) to render components. This constitutes low-risk script generation from known templates for the purpose of visual inspection.
Audit Metadata