Guardrails & Security
Libraries and platforms that validate LLM inputs and outputs, block prompt injection and PII leaks, and red-team models and agents before they reach production.
-
Guardrails AI
Python framework that validates and structures LLM outputs using a hub of community validators.
-
NeMo Guardrails
NVIDIA toolkit for adding programmable input, output, retrieval and dialogue rails to LLM apps.
-
Llama Guard
Meta's safety classifier models for moderating LLM prompts and responses, part of Purple Llama.
-
Lakera
Runtime security for GenAI apps and agents, covering prompt injection, data leakage and red teaming.
-
garak
NVIDIA's LLM vulnerability scanner that probes models for jailbreaks, leaks and other failures.