credential-harvest-landing-page-copy
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to research and incorporate target branding language, recurring phrases, and department names. This ingestion of external, untrusted content provides a surface for indirect prompt injection attacks.
- Ingestion points: Step 2 of the workflow in
SKILL.mddirects the agent to research target branding and naming conventions. - Boundary markers: The instructions lack boundary markers or explicit guidance to treat the research data as strictly passive content, which may lead the agent to follow instructions embedded in the researched text.
- Capability inventory: The skill is focused on text generation and does not have capabilities for command execution, file system access (beyond local reference reading), or network operations.
- Sanitization: No sanitization or validation steps are defined for the data retrieved during the research phase.
Audit Metadata