nemo-guardrails

Implement NVIDIA NeMo Guardrails runtime safety with Colang 2.0 DSL.

11.5k|842|Updated Nov 3, 2025
One-click install
npx skills add https://github.com/Orchestra-Research/AI-Research-SKILLs --skill nemo-guardrails-orchestra-research
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: nemo-guardrails
Source: https://github.com/Orchestra-Research/AI-Research-SKILLs/tree/main/07-safety-alignment/nemo-guardrails
Command: npx skills add https://github.com/Orchestra-Research/AI-Research-SKILLs --skill nemo-guardrails-orchestra-research

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires nemoguardrails, and includes references (resource) and scripts (resource) components.

What problem does it solve?

This Skill safeguards LLM applications by implementing runtime safety checks, preventing harmful or unintended outputs and inputs.

Core Features & Use Cases

  • Jailbreak Detection: Prevents prompt injection attacks.
  • Input/Output Validation: Ensures user inputs and LLM outputs adhere to defined policies.
  • Hallucination Detection: Verifies factual accuracy of LLM responses.
  • PII Filtering: Masks or removes Personally Identifiable Information.
  • Toxicity Detection: Identifies and blocks toxic language.
  • Use Case: Protect a customer service chatbot from generating inappropriate responses or revealing sensitive information by implementing NeMo Guardrails to filter all user queries and bot replies.

Quick Start

Install the nemoguardrails library by running 'pip install nemoguardrails'.

Frequently Asked Questions about nemo-guardrails

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent prompt injection attacks in my LLM application?

Preventing prompt injection attacks involves implementing runtime guardrails to intercept and validate user inputs before they reach the LLM. This Skill uses NVIDIA's NeMo Guardrails to detect jailbreaks and filter malicious queries programmatically.

What is the best way to detect hallucinations in LLM responses at runtime?

Detecting LLM hallucinations at runtime requires fact-checking mechanisms that verify response accuracy against defined policies. This Skill implements NeMo Guardrails to validate LLM outputs, ensuring factual accuracy and preventing unintended responses in real-time.

Can I filter PII from LLM inputs and outputs using Colang?

Filtering PII from LLM interactions is achievable using programmable rails defined in the Colang 2.0 DSL. This Skill leverages NeMo Guardrails to automatically mask or remove Personally Identifiable Information from user queries and bot replies.

Does NeMo Guardrails work for runtime safety on NVIDIA T4 GPUs?

NeMo Guardrails operates effectively on NVIDIA T4 GPUs for low-latency runtime safety filtering. This Skill is production-ready and deploys on T4 hardware to perform real-time input validation, toxicity detection, and jailbreak prevention without significant overhead.

Why do I need runtime safety checks for my customer service chatbot?

Runtime safety checks are needed for customer service chatbots to prevent the generation of inappropriate responses and protect sensitive information. This Skill implements NeMo Guardrails to filter all user queries and bot replies, ensuring adherence to defined compliance policies.

How do I set up input and output validation policies for LLMs?

Setting up input and output validation policies requires defining programmable rules using the Colang 2.0 DSL. This Skill uses NeMo Guardrails to enforce these policies, ensuring both user inputs and LLM outputs adhere strictly to your defined safety and compliance standards.