offer-factory
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external URLs or documents to generate business reasoning. This data enters the agent's context without sanitization or markers to distinguish it from instructions.
- Ingestion points: Phase 1 fetches site data via WebFetch or Apify and reads Google Docs via read_file_content.
- Boundary markers: None detected; the skill does not wrap ingested data in delimiters.
- Capability inventory: The skill can write output to Google Sheets and generate cold outreach hooks.
- Sanitization: No sanitization of the fetched input is performed.
- Mitigation: The skill requires user confirmation before writing results to Google Sheets, providing a manual oversight step.
Audit Metadata