data-research
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external domains, social media, and search results to generate structured prospect summaries intended for outreach.
- Ingestion points: External web content, job postings, and search engine results (SERP) as specified in the research pass instructions.
- Boundary markers: Absent; there are no instructions provided to delimit retrieved content or to warn the agent to ignore instructions embedded within the ingested data.
- Capability inventory: The skill uses automated crawlers and scraping APIs to read external content and generates summary text for direct use in outbound communication.
- Sanitization: No methods for sanitizing or validating the retrieved external content are described before its inclusion in the output CSV.
- [DATA_EXFILTRATION]: The skill performs automated network requests to non-whitelisted domains to gather company information.
- Evidence: The research methodology described in SKILL.md involves performing SERP queries and site content crawls against arbitrary company domains.
Audit Metadata