Llamaguard

A content moderation wrapper for the LlamaGuard 7-8B model from Meta, designed to classify AI inputs and outputs for safety violations. You'd reach for this when building chat applications or AI agents that need guardrails around sensitive content categories like violence, hate speech, or self-harm. The skill handles the HuggingFace authentication and model loading boilerplate so you can focus on defining your moderation policies. Worth noting that you're running a 7-8 billion parameter model locally, so expect meaningful compute requirements. It's forked from ovachiever/droid-tings and sits in a template collection, which suggests it might be more of a starting point than a production-ready solution.

Install to Claude Code

npx -y skills add davila7/claude-code-templates --skill llamaguard --agent claude-code