OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Security & Safety

Guardrails — NVIDIA-NeMo

NeMo Guardrails is an open-source Python library that lets developers add programmable controls between an LLM-based application and its model. A configuration can shape topics, response behavior and dialog paths, as well as guide input and output handling.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Guardrails — NVIDIA-NeMo — OpenViST project profile in AI Security & Safety.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it while building or adapting an LLM-based application when you need programmable conversation rules between application code and the model.

Who it is for

  • Developers building LLM-powered applications who need configurable conversation behavior.
  • Teams building document question-answering, domain-specific assistants or custom LLM endpoints.
  • Developers who want to add a guardrails layer around LangChain chains.

Documented uses

  • Define how an assistant handles selected topics or user requests.
  • Guide conversations along predefined dialog paths, including workflows such as authentication or support.
  • Add fact-checking and output moderation to document question-answering applications.
  • Apply controls to user input, including rejecting or altering it.
  • Route model requests through a configured layer using synchronous or asynchronous Python calls.
Before you choose

Requirements & limits

The project is described as open source. The README documents pip installation and requires Python 3.10, 3.11, 3.12 or 3.13.

  • It is a library and integration layer, so use requires an LLM-based application and a guardrails configuration.
  • The repository documents a Python API and a guardrails server as integration routes.
  • The README identifies an optional LangChain integration and several supported model families; consult the official guides for configuration details.
  • The project aims to help address jailbreaks and prompt injection, but configured controls are not a security guarantee.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Route an application request through the Python API

Load a guardrails configuration and make an LLM request through an LLMRails instance.

Before you start

  • The NeMo Guardrails Python package.
  • A guardrails configuration available at a path.
  • An LLM-based application and model configuration.

Steps

  1. Load the configuration with RailsConfig.from_path("PATH/TO/CONFIG").

  2. Create an instance with LLMRails(config).

  3. Call rails.generate with a messages list, or use the documented asynchronous generate_async method.

Expected outcome

The application submits its request through the guardrails layer, which returns a completion in a chat-style message format.

Guide limitations

  • The example assumes a valid configuration and model setup; those details must be prepared separately.
  • The example does not establish that a particular configuration or model integration will work in every environment.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
12 October 2026
Original website title
NVIDIA-NeMo/Guardrails · Official homepage title · retained in its source language
How we review sourcesSuggest a correction