Skip to navigation

NVIDIA NeMo Guardrails Library

View as Markdown

The NeMo Guardrails library is an open-source Python package for adding programmable guardrails to LLM-based applications. It intercepts inputs and outputs, applies configurable safety checks, and blocks or modifies content based on defined policies.

About the NeMo Guardrails Library

Learn about the library and its capabilities in the following sections.

Overview

Add programmable guardrails to LLM applications with this open-source Python library.

Concept
Guardrail Types

Apply input, retrieval, dialog, execution, and output rails to protect LLM applications.

Concept
How It Works

High level explanation of how Guardrails works.

Concept
Supported LLMs

Connect to NVIDIA NIM, OpenAI, Azure, Anthropic, HuggingFace, and LangChain providers.

Reference

Get Started

Use the following guides to get started with the NeMo Guardrails library and browse the catalog of pre-built guardrails.

Installation

Install NeMo Guardrails with pip, configure your environment, and verify the installation.

Get Started
Tutorials

Follow hands-on tutorials to deploy Nemotron Content Safety, Nemotron Topic Control, and Nemotron Jailbreak Detect

Tutorial
Guardrail Catalog

Reference for pre-built guardrails including content safety, jailbreak detection, topic control, PII handling, agentic security, and third party APIs.

Reference

Next Steps

Once you’ve completed the get-started tutorials, explore the following areas to deepen your understanding.

About Configuring Guardrails

Configure YAML files, Colang flows, custom actions, and other components to control LLM behavior.

Concept
About Running Guardrailed Inference

Run guardrailed inference using the Python API or Guardrails API server.

Get Started
Evaluate

Measure accuracy and performance of dialog, fact-checking, moderation, and hallucination rails.

How To
Logging

Debug guardrails with verbose mode, explain method, and generation log options.

How To
Observability

Use OpenTelemetry to trace requests, forward Python logs, and emit metrics for end-to-end visibility into guardrails.

How To
Deployment Options

Deploy guardrails using the local API server, Docker containers, or production microservices.

How To
LangChain Frameworks

Integrate NeMo Guardrails with LangChain chains, runnables, and LangGraph workflows.

How To