nemo-guardrails

Enforce runtime safety rails for LLM applications with Colang 2.0 DSL.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/t2ance/dr-claw-plugin --skill nemo-guardrails-t2ance
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: nemo-guardrails
Source: https://github.com/t2ance/dr-claw-plugin/tree/main/plugins/ml-frontiers/skills/safety-alignment/nemo-guardrails
Command: npx skills add https://github.com/t2ance/dr-claw-plugin --skill nemo-guardrails-t2ance

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

NeMo Guardrails provides programmable runtime safety rails to enforce policy, reduce jailbreak risks, and prevent unsafe outputs in LLM applications.

Core Features & Use Cases

  • Jailbreak detection, input/output validation, and fact-checking to keep interactions safe.
  • PII filtering, toxicity detection, and multi-layer safety rails for production-grade deployments.
  • Use cases include chatbots, virtual assistants, and enterprise AI agents that require configurable safety flows via Colang 2.0 DSL.

Quick Start

Install nemoguardrails and wrap your LLM with LLMRails using a RailsConfig defined from content

Frequently Asked Questions about nemo-guardrails

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

FAQPage Schema
How do I add runtime guardrails to an LLM application for jailbreak detection and PII filtering?▼

Runtime guardrails for LLM applications are added by wrapping your model with LLMRails using a RailsConfig defined by Colang 2.0 DSL to enforce safety flows. This setup provides programmable input validation, jailbreak detection, and PII filtering for production chatbots.

What are programmable safety rails for large language models and when do I need them?▼

Programmable safety rails are runtime checks that enforce policy and prevent unsafe LLM outputs. You need them in production chatbots, virtual assistants, or enterprise AI agents requiring configurable input validation, toxicity detection, and fact-checking to reduce jailbreak risks.

How do I configure multi-layer safety flows using Colang 2.0 DSL for an LLM chatbot?▼

You configure multi-layer safety flows by defining rules in Colang 2.0 DSL within a RailsConfig, then initializing LLMRails with that configuration. This programmable approach enforces fact-checking, toxicity detection, and input validation to keep interactions safe.

Do I need to install nemoguardrails separately to set up runtime safety checks for my assistant?▼

Yes, you must install the nemoguardrails package separately to use this Skill. Setting up runtime safety checks requires installing nemoguardrails, then wrapping your LLM with LLMRails using a RailsConfig to apply the desired safety flows.

Can I use runtime guardrails for enterprise AI agents that require fact-checking and toxicity detection?▼

Yes, runtime guardrails support enterprise AI agents by providing configurable fact-checking, toxicity detection, and PII filtering. Using Colang 2.0 DSL, you can define multi-layer safety flows to enforce policy and reduce unsafe outputs in production deployments.

What is the best way to prevent jailbreaks and unsafe outputs in production LLM-powered chatbots?▼

The best way to prevent jailbreaks in production LLM chatbots is implementing programmable runtime guardrails. By configuring LLMRails with Colang 2.0 DSL, you enforce input validation, output checks, and fact-checking to systematically reduce jailbreak risks and unsafe outputs.