The Agent Skills Directory

[SAFE]: The skill implements a robust 6-step workflow for evaluating corporate credit applicants against banking regulations. It includes internal 'veto' redlines (A1-A6) to prevent unauthorized lending and utilizes a local Python validation script (scripts/validate_admission_scan.py) that relies only on standard libraries. No external downloads, credential exposure, or prompt injections were detected. Data handling practices include explicit desensitization rules (placeholders for IDs and account numbers) and adherence to official Chinese banking regulatory guidelines. The technical components, including the audit trail and test cases, are consistent with its stated purpose of assisting credit managers in pre-screening loan applications.

admission-rules-scan