portfolio-story-architecture
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill directs the agent to access external URLs to inspect media for design references.
- Evidence: Multiple links in references/cases.md pointing to goodcase.ai, twimg.com, and x.com.
- Context: These network operations are intended for visual style verification.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests 'Case evidence' including prompt excerpts and external media content.
- Ingestion points: The references/cases.md file contains prompt snippets and external links.
- Boundary markers: A 'Safety and attribution' section and a 'reference contract' workflow are defined in SKILL.md.
- Capability inventory: The agent is instructed to use browser, image, or video tools.
- Sanitization: No explicit filtering or sanitization of external media content is specified.
Audit Metadata