ux-researcher-designer
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script
scripts/persona_generator.pyto synthesize user personas from research data. This is an expected functional component of the toolkit intended for automated data processing. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted research data (such as interview transcripts, survey responses, and support tickets), which constitutes an indirect prompt injection attack surface.
- Ingestion points: User research data is ingested through interviews, recordings, analytics, and support tickets, then structured into JSON format for processing as specified in the core workflows.
- Boundary markers: The skill mandates identifying the privacy posture and source data availability before artifact production, and it includes formal 'Validation Gates' to review artifact quality.
- Capability inventory: The skill's primary capabilities include running the local persona generation script and producing formatted design artifacts in Markdown.
- Sanitization: Robust data handling instructions are provided in
references/privacy-validation-and-troubleshooting.md, including the mandatory use of participant IDs, redaction of direct identifiers, and masking of sensitive on-screen information to prevent PII exposure.
Audit Metadata