domains / families / security / nemo-guardrails

nemo-guardrails

separate

NVIDIA's runtime safety framework for LLM applications. Features jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, toxicity detection. Uses Colang 2.0 DSL for programmable rails. Production-ready, runs on T4 GPU.

When to use

• Need runtime safety checks
• Want programmable safety rules
• Need multiple safety mechanisms (jailbreak, hallucination, PII)
• Building production LLM applications
• Need low-latency filtering (runs on T4)
• **Jailbreak detection**: Pattern matching + LLM
• **Self-check I/O**: LLM-based validation
• **Fact-checking**: Retrieval + verification

When not to use

• standalone moderation model
• simple API-based filtering
• Google's toxicity detection
• training-time safety

Skill page

Source