Guardrailz SDK

Official JavaScript SDK for Guardrailz — enterprise-grade guardrails for AI applications.

Protect your LLM inputs and outputs against:

• Prompt injection
• PII & sensitive data leaks
• Unsafe or non-compliant content
• Custom policy violations

Installation

npm install @guardrailz/sdk
# or
yarn add @guardrailz/sdk

Quick Start

import { GuardrailsClient } from '@guardrailz/sdk';

const client = new GuardrailsClient({
  baseUrl: 'https://guardrailz.vercel.app',
  apiKey: process.env.GUARDRAILZ_API_KEY,
});

const result = await client.validate({
  text: 'Hello world',
  profileName: 'default',
  validationType: 'input',
});

console.log(result);

API

new GuardrailsClient(config)

interface GuardrailsSDKConfig {
  baseUrl: string;
  apiKey?: string;
  timeoutMs?: number;
  retries?: number;
}

client.validate(request)

interface ValidateInputRequest {
  text: string;
  profileName: string;
  validationType: 'input' | 'output' | 'tool';
}

Returns detailed guardrail execution results including:

• Pass / fail status
• Execution time
• Guardrail-level decisions

Documentation

Full docs available at:
https://guardrailz.vercel.app/docs

License

MIT © Guardrailz