guardrails-safety-filter-builder
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill serves as a defensive implementation guide. It utilizes reputable libraries for PII management and follows industry best practices for LLM safety.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes an attack surface by processing user-supplied input, but effectively mitigates this via a multi-layered guardrail pipeline.\n
- Ingestion points: user_input in the apply_guardrails function (SKILL.md).\n
- Boundary markers: Logically implemented through the detect_prompt_injection and check_policy gates.\n
- Capability inventory: Limited to LLM-based text generation and classification.\n
- Sanitization: Uses regex patterns for injection detection and the presidio library for PII redaction.
Audit Metadata