influencer-marketing
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external sources like scraped text, reviews, and emails.
- Ingestion points: SKILL.md (scraped page text, reviews, comments, emails, third-party exports).
- Boundary markers: The skill contains a strong defensive instruction requiring the agent to treat such data as untrusted and to not follow embedded instructions.
- Capability inventory: None; the skill is documentation-based and does not include scripts or tools for automated execution.
- Sanitization: Explicitly instructs the agent to ignore instructions within the data.
- [SAFE]: The skill accesses local project context files but includes robust rules for authorization and evidence labeling, which reduces the risk of unintended behavior.
Audit Metadata