guardrails-ai/guardrails

Project overview

By providing a framework to mitigate unmitigated inputs and outputs alongside structured data generation, guardrails-ai/guardrails directly addresses the risks and unreliability inherent in large language model applications.

Deployment

Refer to project documentation

License

Apache-2.0

Best for

• Developers and AI engineers building Python-based applications who need to detect, quantify, and mitigate risks in LLM inputs and outputs.

Key capabilities

• Users can create their own validators and contribute them to Guardrails Hub.
• The framework can be set up as a standalone service served by Flask, allowing interaction via a REST API.

Limitations and risks

• Guardrails can be used with Python and JavaScript; support for other languages is still being worked on.
• The framework requires an external LLM (proprietary or open-source) to function and generate structured data.

Getting started

• Install easily via pip and follow clear CLI examples in the README to get started. First steps include running 'pip install guardrails-ai', 'guardrails configure', and 'guardrails hub install hub://guardrails/regex_match'.

Evidence and sources

• GitHub project description: Adding guardrails to large language models.
• README: Guardrails runs Input/Output Guards in your application that detect, quantify and mitigate the presence of specific types of risks.
• README: We just launched Guardrails Index -- the first of its kind benchmark comparing the performance and latency of 24 guardrails across 6 most common categories!
• README: Function calling: For LLMs that support function calling, we generate structured data using the function call syntax.
• README: pip install guardrails-ai

Reviewed at: 2026-07-18T06:36:04Z