learning-translation-quality
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE] (SAFE): No access to sensitive files or network exfiltration paths identified.
- [REMOTE_CODE_EXECUTION] (SAFE): No dynamic execution (eval/exec) or remote script downloads detected.
- [PROMPT_INJECTION] (SAFE): No instructions found that attempt to bypass safety guidelines or override agent behavior.
- [INDIRECT_PROMPT_INJECTION] (SAFE): 1. Ingestion points: untrusted data enters via source_content in skill.py. 2. Boundary markers: Absent. 3. Capability inventory: The skill has no dangerous capabilities (no subprocess, network, or file-write ops). 4. Sanitization: Absent. The surface is not exploitable due to a lack of capabilities.
Audit Metadata