Product / UX / Safety & Guardrails

NeMo Guardrails

NVIDIA's open-source toolkit for adding programmable guardrails around LLM applications. It wraps input/output moderation, dialog control, retrieval filtering, and tool-use control through a small policy language (Colang), and works across multiple LLM providers. Deployable as a Python library, CLI, Docker, or standalone server.

NeMo Guardrails, Apache-2.0, Python; ~6.6K GitHub stars (June 2026). Fully open source and self-hostable. Verified live June 2026.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public(full Python implementation)
self-host: yes(library/CLI/Docker/server)
service: none
core-gated: ungated

Fully open source: Apache-2.0, complete source, self-hostable with no proprietary tier.

https://github.com/NVIDIA/NeMo-Guardrails recorded 2026-06-29
Apache-2.0; full Python source; self-hostable as library/CLI/Docker/server; ~6.6k stars

Adoption

3 low confidence

3.0

~6.6K GitHub stars (June 2026); stars-only proxy, capped per methodology; broadly referenced guardrail framework.

https://github.com/NVIDIA/NeMo-Guardrails recorded 2026-06-29
~6.6k stars (June 2026)

Capability

4 medium confidence

4.0

One of the most feature-complete open guardrail orchestration layers; the Colang policy language adds flexibility beyond simple scanners.

https://github.com/NVIDIA/NeMo-Guardrails recorded 2026-06-29
I/O moderation, dialog control, retrieval filtering, tool-use control

Unchanged since 2026-08-01 (last edited, not re-checked)