Vigil is an open-source security scanner that detects prompt injections, jailbreaks, and other potential threats to Large Language Models (LLMs). Prompt injection arises when an attacker successfully influences an LLM using specially designed inputs.

Source: Help Net Security