csv-to-json-validator
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [SAFE]: The skill documentation serves as a design specification and does not contain any executable scripts, remote downloads, or malicious logic.
- [PROMPT_INJECTION]: The skill provides explicit instructions to prevent indirect prompt injection via CSV files by ensuring formula-injection payloads (e.g., '=SUM(...)') are treated as literal strings and not evaluated.
- [DATA_EXFILTRATION]: The skill identifies and mitigates potential directory traversal risks (CWE-22) by instructing agents to reject input paths containing '..' or absolute paths outside the workspace.
Audit Metadata