guardrails
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides defensive code templates and documentation for building safety guardrails. It does not contain executable malicious code, unauthorized network operations, or persistence mechanisms.
- [PROMPT_INJECTION]: While the skill contains common prompt injection strings (e.g., 'ignore previous instructions'), these are explicitly used within regex patterns for the purpose of building a defensive detection guardrail.
- [DATA_EXPOSURE]: The skill includes regex patterns for sensitive data (SSN, credit cards, etc.), but these are used for identification and redaction in the context of an output filter, promoting safe data handling practices.
- [COMMAND_EXECUTION]: The provided Python snippets demonstrate standard logical flows for data validation and do not use unsafe functions like eval(), exec(), or subprocess calls.
Audit Metadata