promotional-campaigns
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill presents an ingestion surface for untrusted external data.
- Ingestion points: Competitor research from the Meta Ad Library, social media platforms (TikTok, Instagram, YouTube), and community threads (Reddit), along with user-provided data exports for performance measurement.
- Boundary markers: The skill mandates the use of
references/data-input-integrity.mdto verify the completeness and validity of data before any computation occurs. - Capability inventory: The skill has the ability to read and write to the local
.agents/product-context.mdfile and can route content to other agent skills such asemail-campaignandad-copy. - Sanitization: Integrity checks are focused on temporal consistency and data accuracy; specific sanitization for instruction-based injection in research data is not explicitly defined.
- [SAFE]: The skill manages product-specific knowledge by maintaining state in
.agents/product-context.md. This practice allows the agent to build a persistent context for a specific account or company without accessing sensitive system-level credentials, user home directories, or private configuration files. All external research activities are aligned with the skill's primary marketing research purpose.
Audit Metadata