sales-practices
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection.
- Ingestion points: The skill is intended to evaluate external data like trade logs and disclosures for compliance violations.
- Boundary markers: There are no specific instructions or delimiters provided to prevent the agent from following instructions potentially embedded in that data.
- Capability inventory: The skill specifies the use of high-privilege tools (Bash, Read, Write, Edit), which increases the potential impact of an injection.
- Sanitization: There is no logic for sanitizing or escaping the data being analyzed.
- [NO_CODE]: The skill does not contain any executable scripts, Python packages, or Node.js dependencies.
- [SAFE]: No obfuscation, data exfiltration patterns, or hardcoded credentials were found. The technical content is consistent with its stated educational and regulatory purpose.
Audit Metadata