google-ads-write
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external API responses, specifically when fetching optimization recommendations, which creates a potential surface for indirect prompt injection from compromised or adversarial API sources.
- Ingestion points: Google Ads API responses for optimization recommendations and campaign entity queries as described in
SKILL.md. - Boundary markers: The mandatory Confirm-Execute-Postcheck (CEP) protocol requires explicit user confirmation (e.g., 'CONFIRM', 'yes') before any mutation is executed.
- Capability inventory: The skill possesses capabilities to modify campaign statuses, update budgets, adjust bids, and create new ads or campaigns.
- Sanitization: The skill enforces character limit validation for ad creation and requires all new ads and campaigns to be created in a
PAUSEDstate for manual review.
Audit Metadata