meta-ad-policy-checker
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input and external website content.
- Ingestion points: Ad assets (copy) and Destination URLs provided by the user (SKILL.md, Phase 0 & 4).
- Boundary markers: Absent; there are no explicit delimiters instructed for the fetched content or ad copy.
- Capability inventory: The skill utilizes web-fetching tools to retrieve policy and landing page data.
- Sanitization: Absent; the skill does not specify filtering for malicious instructions within the fetched data.
- [EXTERNAL_DOWNLOADS]: Fetches policy text from Meta's official Transparency Center (transparency.meta.com) and user-provided URLs. This is necessary for the skill's primary function and targets a well-known service.
Audit Metadata