perfect-design
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it ingests untrusted user data and uses it to direct the actions of several specialized sub-skills.
- Ingestion points: In
SKILL.md(Step 1: Listen), the skill is designed to accept URLs, screenshots, and free-text descriptions from users to diagnose project needs. - Boundary markers: There are no explicit instructions or delimiters defined to isolate user-provided content or to instruct the agent to ignore any embedded directives within that content.
- Capability inventory: The skill possesses significant routing capabilities, coordinating eight distinct phases including research, design, and site builds (e.g.,
/audience-research,/visual-identity, and/motion-web-design). - Sanitization: The instructions do not specify any validation or filtering mechanisms for the user-supplied data before it is passed to the specialized skills in the pipeline.
Audit Metadata