cac-roas-benchmark
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown, reference documentation, and evaluation data. No executable code (Python, JavaScript, or shell scripts) is included in the skill package.
- [INDIRECT_PROMPT_INJECTION]: The skill instructions specify that the agent should process external data sources such as CSV files, spreadsheets, and data warehouse extracts. While this creates a theoretical surface for indirect prompt injection if malicious data is provided at runtime, the skill lacks the capabilities (e.g., shell access, network operations, or persistent file system modifications) to act as a significant attack vector.
- Ingestion points: User-provided exports including CSVs and spreadsheets mentioned in
SKILL.md. - Boundary markers: Not explicitly defined for raw data, though structured report templates are provided.
- Capability inventory: The skill contains no subprocess calls, network operations, or dynamic code execution capabilities.
- Sanitization: No explicit sanitization logic is described for the ingested data.
Audit Metadata