Guardrails
Guardrails overview
Protect your LLM application in real time by blocking unsafe inputs and outputs
Policies
Define your checks once in Opik and reference them by name from your application
Guards
Prompt injection model
Opik Guard, our purpose-built detection model, evaluated on attacks it has never seen
Custom models
Fine-tune a language model for a check the built-in guards do not cover
Guardrails server
Install, configure, and run the backend that powers the model-based guards