ad-test-designer
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: A thorough review of the skill instructions and referenced documents revealed no malicious code or security violations.
- [PROMPT_INJECTION]: The skill implements a security boundary for external data to mitigate indirect prompt injection. Ingestion point: User-provided results CSV exports. Boundary markers: Explicit instructions to treat CSV text as data values rather than commands. Capability inventory: Local file writing to the memory directory. Sanitization: Instructional guidance referencing a central security policy.
- [REMOTE_CODE_EXECUTION]: The instructions in both the main skill and the test design guide explicitly forbid the agent from using programming libraries or executing code (such as Python or scipy) for statistical analysis, mandating manual calculation and documentation.
- [DATA_EXFILTRATION]: No network operations, such as curl or wget, were identified. The skill's data handling is restricted to reading user-provided exports and writing to a local memory directory following user confirmation.
- [NO_CODE]: The skill is entirely composed of markdown instructions and does not include any executable scripts, binaries, or automated code components.
Audit Metadata