email-outreach-prep
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, creating a potential surface for indirect prompt injection from untrusted content.\n
- Ingestion points: External data is retrieved via
linkedin_web_v2_user_profileandtavily_search(SKILL.md).\n - Boundary markers: The instructions do not specify explicit delimiters or isolation markers for external content interpolated into prompts.\n
- Capability inventory: The skill is limited to research activities and lacks tools for persistent storage, file system modification, or direct network exfiltration.\n
- Sanitization: No specific input validation or sanitization procedures are defined for the ingested research data.\n- [SAFE]: The skill operates as a legitimate lead generation and research utility. It utilizes a managed tool gateway for interactions with external APIs and strictly follows a read-only methodology. No evidence of prompt injection, data exfiltration, or obfuscation was identified within the instructions or referenced files.
Audit Metadata