anomaly-detector
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process metrics from external advertising platforms (Google Ads, Meta, LinkedIn, etc.), which represents a theoretical attack surface for indirect prompt injection if those sources were compromised.
- Ingestion points: Hourly and daily campaign metrics (spend, clicks, impressions) retrieved via GAQL and Marketing APIs as described in SKILL.md.
- Boundary markers: The provided workflows and Python scripts do not implement explicit boundary markers or delimiters for the external data.
- Capability inventory: The skill's scripts are limited to mathematical calculations (Z-score, IQR, pacing) and do not include file-writing, arbitrary command execution, or network-sending capabilities.
- Sanitization: No explicit sanitization or validation logic is provided for the metric inputs beyond casting to floating-point numbers in the Python implementations.
Audit Metadata