NeMo Guardrails
Easily add programmable guardrails to LLM-based conversational systems
About
Found in: Yigtwxx/Awesome-RAG-Production
Details
NVIDIA NeMo Guardrails is an open-source toolkit that enables developers to add programmable guardrails to LLM-based conversational systems. Guardrails are specific rules that control LLM outputs, such as avoiding certain topics, following predefined dialog paths, using specific language styles, or extracting structured data. The toolkit supports building trustworthy and safe applications by allowing developers to define behavior on specific topics and prevent unwanted discussions. It also facilitates secure integration of LLMs with other services (tools) and enables controllable dialog through predefined conversational paths. The library is compatible with Python 3.10-3.13, is installable via pip, and is available under the Apache 2.0 license. Official documentation is hosted at docs.nvidia.com/nemo/guardrails.