Guardrails AI

It is a comprehensive tool designed to fortify the security of Large Language Models (LLMs). By offering sanitization, detection of harmful language, prevention of data leakage, and resistance against prompt injection attacks, it ensures that your interactions with LLMs remain safe and secure.

It is a robust static analysis framework for validating that LLM-generated structured output is safe. It currently supports SQL.

It is an open-source toolkit for monitoring Large Language Models (LLMs). It extracts signals from prompts & responses, ensuring safety & security.

It is a self-hardening prompt injection detector. It is designed to protect AI applications from prompt injection (PI) attacks through a multi-stage defense.