credit-related-party-detection
Pass
Audited by Gen Agent Trust Hub on Jun 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or security vulnerabilities were identified in the skill logic or associated files. The skill is designed for highly regulated financial auditing tasks.
- [COMMAND_EXECUTION]: The skill executes a local validation script (scripts/validate_related_party.py) during the initial data verification step. Analysis of this script confirms it is benign, using only standard Python libraries (re, sys, datetime) to perform data format checks and field validation. It does not perform network requests, file system writes, or system-level changes.
- [PROMPT_INJECTION]: The instructional content focuses on enforcing auditing standards and regulatory compliance. The 'Anti-model laziness' instructions are protective measures to ensure the AI performs thorough analysis of financial data and does not skip required investigative steps. There are no attempts to bypass AI safety guidelines or manipulate agent behavior maliciously.
- [DATA_EXFILTRATION]: While the skill handles sensitive financial and corporate data, it does not include any mechanisms for unauthorized data transmission. Data sources mentioned such as 'ECIF', 'Tianyancha', and 'Qichacha' are standard and legitimate for the intended financial domain and are used as references for the AI rather than targets for exfiltration code.
Audit Metadata