audit-ad-account
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform local directory discovery and create a structured workspace (e.g., /creative/, /reference/) to store audit findings. This behavior is necessary for the skill's primary function of providing persistent context for advertising management.\n- [PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection via scraped ad content (captions, transcripts). 1. Ingestion points: identified in SKILL.md as captions, transcripts, and emails; 2. Boundary markers: the skill provides explicit instructions to treat this content as material for analysis, not instructions; 3. Capability inventory: tool-based data retrieval and local file system writes; 4. Sanitization: included rules to ignore embedded commands and terminate the process if the text addresses the agent directly.\n- [DATA_EXFILTRATION]: Accesses advertising account identifiers and performance metrics via Zernio MCP tools. Data is used strictly for report generation within the local environment or the user's connected Google Drive, with no evidence of unauthorized external transmission.
Audit Metadata