detecting-data-anomalies
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues detected. The skill provides documentation and standard boilerplate code for data analysis. It does not attempt to access sensitive files, communicate over the network, or execute remote code.- [COMMAND_EXECUTION]: The skill requests access to the Bash tool to run Python scripts. The provided scripts perform benign operations like file metadata inspection (size) and basic JSON parsing to validate structure.- [PROMPT_INJECTION]: An indirect prompt injection surface exists as the skill is designed to process external datasets. However, the provided scripts do not perform unsafe interpolation or execution of data content, and the risk is mitigated by the agent's internal safety guardrails.
Audit Metadata