algo-risk-credit

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's instructions and reference files describe standard credit risk modeling procedures without any malicious patterns. The Python scripts in references/reject-inference.md and references/woe-binning.md perform domain-appropriate statistical operations.
  • [COMMAND_EXECUTION]: No dangerous system commands, subprocess calls, or privilege escalation patterns were identified. The code is restricted to data processing using well-known, legitimate libraries.
  • [EXTERNAL_DOWNLOADS]: The skill does not attempt to download external content or execute remote scripts. It references reputable open-source libraries (scorecardpy, optbinning) as recommendations for production use.
  • [DATA_EXFILTRATION]: There is no evidence of network operations or unauthorized data access. The skill focuses on the local analysis of borrower characteristics provided within the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:10 AM