google-ads-audit
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to the processing of external data from Google Ads.
- Ingestion points: Data from Google Ads campaigns, conversion actions, and user search terms are ingested into the context via
mcp__cogny__google_ads__tool_execute_gaqlinSKILL.md(Steps 2, 3, 4, 5, and 6). - Boundary markers: Absent. There are no explicit instructions or delimiters defined to separate the incoming Google Ads data from the system prompt or to prevent execution of instructions contained within text fields.
- Capability inventory: The skill is capable of recording findings using
mcp__cogny__create_findingand has access toWebFetchandWebSearchtools. - Sanitization: Absent. There is no parsing, validation, or escaping specified for the fields retrieved from the external API before processing.
Audit Metadata