dotcom-secrets
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze external content, specifically competitor funnels via user-provided URLs or descriptions. This creates a surface for indirect prompt injection where hidden instructions in the target URL could influence the agent's behavior.
- Ingestion points: Analyzing competitor funnels from a URL or description in the 'Prompt Examples' and 'Instructions' sections.
- Boundary markers: The skill lacks explicit delimiters or instructions for the agent to treat external content solely as data.
- Capability inventory: The skill itself does not define direct execution capabilities or network tools, though the agent's environment may provide them.
- Sanitization: No sanitization or validation of the external input is described.
- [METADATA_POISONING]: The 'Skill Boundaries' section contains misleading metadata. It claims the skill is for 'audio production workflows' and 'audio engineering,' which contradicts the rest of the skill focused on sales funnels. This inconsistency suggests a template error or misleading documentation regarding the skill's actual scope and limitations.
Audit Metadata