ux-architect
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use common shell utilities like cat and grep to extract business goals and task lists from files within the ai/memory-bank/ directory. These operations are limited to the project's local context and are used for requirement analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates an attack surface by reading untrusted project data from external files. 1. Ingestion points: ai/memory-bank/site-setup.md and tasks/*.md. 2. Boundary markers: None explicitly mentioned in the cat/grep workflow. 3. Capability inventory: File system reading (cat) and generation of web code (CSS/JS/HTML). 4. Sanitization: None defined for the ingested project text. However, the risk is minimal given the technical architect use case.
- [SAFE]: The provided JavaScript ThemeManager class and the CSS variable architecture are standard, benign templates for modern web development. No suspicious network requests or privilege escalation attempts were found.
Audit Metadata