Guardrails
Framework for validating and controlling LLM outputs in AI apps
Guardrails is an open source Python framework designed to help developers build more reliable and controlled applications powered by large language models. It provides mechanisms for validating and constraining both the inputs sent to a model and the outputs generated by it, helping reduce risks such as harmful content, prompt injection, or inaccurate responses. Guardrails works by applying configurable guards that intercept and evaluate interactions with the model before results are...