ads
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists entirely of instructional text, checklists, and templates designed for marketing professionals. No executable scripts or high-privilege commands are present within the skill files.
- [PROMPT_INJECTION]: No malicious prompt injection or jailbreak patterns were detected. The skill contains a theoretical surface for indirect prompt injection through its ingestion of project context files like '.agents/product-marketing.md'. Ingestion point: SKILL.md. Boundary markers: Absent. Capability inventory: None in the provided files; references external ad platform tool integrations. Sanitization: Absent.
- [EXTERNAL_DOWNLOADS]: Reference documentation includes standard tracking pixel code snippets for platforms like Google, Meta, LinkedIn, and TikTok. These reference official domains from well-known services and are provided for user guidance rather than agent-side execution.
- [DATA_EXFILTRATION]: The skill's request for project context files is a benign grounding technique used to gather marketing requirements and does not involve searching for sensitive credentials or system files.
Audit Metadata