LlamaGuard is a specialized 7-8B parameter model from Meta designed specifically to classify the safety of LLM prompts and responses across specific harm categories.

How does this skill help with Claude Code?

This skill provides Claude with the domain-specific patterns, code snippets, and optimization best practices necessary to integrate LlamaGuard moderation into your AI development workflow.

Which safety categories does LlamaGuard cover?

It filters content across six main categories: Violence & Hate, Sexual Content, Guns & Weapons, Regulated Substances, Suicide & Self-Harm, and Criminal Planning.

Can LlamaGuard be used for high-throughput production?

Yes, by using vLLM or tensor parallelism, LlamaGuard can achieve throughput of 50-100 requests per second on enterprise GPUs like the NVIDIA A100.

LlamaGuard AI Content Moderation

Name: LlamaGuard AI Content Moderation
Author: Orchestra-Research

byOrchestra-Research

•

3,983

•

Security & Testing

Moderates LLM inputs and outputs using Meta's specialized LlamaGuard models to ensure safety and policy compliance across six critical categories.

This skill provides comprehensive guidance and implementation patterns for Meta’s LlamaGuard, a specialized 7-8B parameter model designed specifically for content safety classification. It enables developers to implement robust input and output filtering across key safety categories—including violence, sexual content, and criminal planning—with 94-95% accuracy. Ideal for production-grade AI applications, the skill covers integration with vLLM for high-throughput inference, FastAPI for modular services, and NeMo Guardrails for building sophisticated, multi-layered safety frameworks.

Key Features

01Detailed workflows for input filtering and output auditing

023,983 GitHub stars

03Integration guides for NeMo Guardrails and HuggingFace

04High-performance inference patterns using vLLM and quantization

05Multi-category safety classification for prompts and responses

06Optimization strategies for latency and hardware requirements

Use Cases

01Building scalable, low-latency moderation microservices for AI platforms

02Auditing LLM-generated responses for policy violations before delivery to end-users

03Filtering malicious or unsafe user prompts before they reach a generative model

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add orchestra-research/ai-research-skills llamaguard

For use in Claude.ai and ChatGPT

Download Skill