portfolio-story-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes design case evidence from
references/cases.mdto influence its output, creating a surface for indirect prompt injection. - Ingestion points: Data is read from
references/cases.mdwhich contains prompts and descriptions from external design cases. - Boundary markers: The skill requires the agent to select exactly one 'anchor Case' and generate a 'reference contract' defining what to Preserve, Replace, and Avoid, which helps bound the influence of the external content.
- Capability inventory: The skill utilizes the agent's ability to access external URLs via image, video, or browser tools for media inspection.
- Sanitization: The 'Reference Contract' workflow serves as a logical filter to ensure specific traits are mapped rather than blindly following instructions embedded in the examples.
- [SAFE]: The skill includes explicit safety instructions that prohibit publishing, purchasing, or making account changes without approval. It also contains attribution requirements to prevent creator impersonation.
Audit Metadata