google-ads-quality-score
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, such as Google Ads CSV exports and fetched landing pages, which constitutes an attack surface for indirect instructions.
- Ingestion points: Reads keyword-level component ratings and landing page content as defined in
SKILL.md. - Boundary markers: Includes a robust security framework in
references/agent-security.mdand explicit constraints inSKILL.mdto treat retrieved text as data only. - Capability inventory: The agent utilizes tool-based web fetching and writes persistent context to
.agents/product-context.md. - Sanitization: Instructs the agent to quote potential injection attempts verbatim rather than executing them and prohibits following secondary URLs found within untrusted content.
Audit Metadata