Repolode
Submit
guardrails-ai

guardrails-ai/guardrails

A Python framework that adds input and output guards to LLM applications and generates validated structured data from LLM outputs.

View on GitHubWebsite
Stars
7,540
Forks
727
Open issues
74
Last commit
6 days ago

What is it?

Guardrails is a Python framework for making LLM-based applications more reliable. It runs input and output guards that detect, quantify and mitigate specific risks, built from validators that can be combined into a single guard. Validators are distributed through Guardrails Hub, and each can be installed as a separate package with pip.

The framework also helps generate structured data from LLMs. You define the expected output as a Pydantic model, and Guardrails produces output matching that schema, using function calling where the model supports it and schema-in-prompt instructions where it does not.

Guardrails can run as a standalone Flask service exposing a REST API, and it offers an OpenAI-compatible endpoint so existing OpenAI SDK clients can route calls through guards. The project is licensed under Apache-2.0.

Key features

  • 01Input and output guards that check LLM inputs and responses against configurable validators.
  • 02Pre-built validators from Guardrails Hub, such as regex matching, competitor checks and toxic language detection.
  • 03Structured data generation from LLMs using Pydantic models.
  • 04Function calling and prompt-based schema instructions for different model capabilities.
  • 05Standalone server with a REST API and an OpenAI-compatible endpoint.

Use cases

  • Filtering toxic or off-topic content in a chatbot before it reaches users.
  • Preventing an assistant from naming competitors or leaking restricted information.
  • Getting LLM responses in a fixed JSON structure for downstream code.
  • Adding a validation layer to an existing OpenAI-based application without rewriting client code.

Quick start

  1. 1

    Install the guardrails-ai package

    pip install guardrails-ai
  2. 2

    Configure the Guardrails Hub CLI

    guardrails configure
  3. 3

    Install a validator from Guardrails Hub

    pip install guardrails-ai-regex-match
  4. 4

    Create a Guard with the installed validator and run it, as shown in the README

Things to know

  • Guardrails is discontinuing its hosted remote inferencing, with a planned cutoff on August 25, 2026.
  • Validators are moving to separate PyPI packages that must be installed individually.

Topics

#llm#guardrails#validation#structured-output#pydantic#openai#python#ai#foundation-model#gpt-3

FAQ

What is guardrails?
A Python framework that adds input and output guards to LLM applications and generates validated structured data from LLM outputs. Guardrails is a Python framework for making LLM-based applications more reliable. It runs input and output guards that detect, quantify and mitigate specific risks, built from validators that can be combined into a single guard. Validators are distributed through Guardrails Hub, and each can be installed as a separate package with pip.
Is guardrails free to use? What license does it have?
guardrails is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is guardrails still maintained?
Actively maintained. The last commit to guardrails was on Oct 15, 2026.
What can I use guardrails for?
Filtering toxic or off-topic content in a chatbot before it reaches users. Preventing an assistant from naming competitors or leaking restricted information. Getting LLM responses in a fixed JSON structure for downstream code. Adding a validation layer to an existing OpenAI-based application without rewriting client code.
How do I install guardrails?
Install the guardrails-ai package: pip install guardrails-ai Configure the Guardrails Hub CLI: guardrails configure Install a validator from Guardrails Hub: pip install guardrails-ai-regex-match Create a Guard with the installed validator and run it, as shown in the README
Where can I get guardrails?
The source code and installation instructions are on GitHub: https://github.com/guardrails-ai/guardrails

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects