Product / UX / Safety & Guardrails

Guardrails AI

An open framework for input/output guards on LLM applications, with a Hub of pre-built validators (PII, toxicity, prompt injection, format/structure). Wraps model calls in guards that detect risks and enforce structured output, self-hostable as a Python library.

Guardrails AI framework, Apache-2.0, ~7.1K GitHub stars (June 2026); plus Guardrails Hub (validator registry) and commercial offerings. Verified live June 2026.

Openness

4 medium confidence

4.0

license: Apache-2.0(OSI) for the framework
source: public
hub: Guardrails Hub validator registry
commercial: hosted/enterprise tier
core-gated: gated

Apache-2.0 framework that is fully self-hostable, with a Hub of validators and a commercial tier on top; classed open_core alongside other OSS-core-plus-commercial tools.

https://github.com/guardrails-ai/guardrails recorded 2026-06-29
Apache-2.0 framework; ~7.1k stars; Guardrails Hub validators; commercial offerings

Adoption

3 low confidence

3.0

~7.1K GitHub stars (June 2026); stars-only proxy, capped per methodology.

https://github.com/guardrails-ai/guardrails recorded 2026-06-29
~7.1k stars (June 2026)

Capability

4 medium confidence

4.0

Broad validator ecosystem via the Hub; strong on structured-output enforcement in addition to safety checks.

https://github.com/guardrails-ai/guardrails recorded 2026-06-29
I/O guards, validator Hub, structured-output enforcement

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