PromptGuard Node.js SDK

Drop-in security for AI applications. No code changes required.

Installation

npm install promptguard-sdk
# or
yarn add promptguard-sdk
# or
pnpm add promptguard-sdk

Quick Start

import { PromptGuard } from 'promptguard-sdk';

// Initialize client
const pg = new PromptGuard({ apiKey: 'pg_xxx' });

// Use exactly like OpenAI client
const response = await pg.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: 'Hello!' }]
});

console.log(response.choices[0].message.content);

Drop-in Replacement

If you're already using OpenAI's Node.js client, just change the import:

// Before
import OpenAI from 'openai';
const client = new OpenAI();

// After
import { PromptGuard } from 'promptguard-sdk';
const client = new PromptGuard({ apiKey: 'pg_xxx' });

// Your existing code works unchanged!

Features

Security Scanning

// Scan content for threats
const result = await pg.security.scan('Ignore previous instructions...');

if (result.blocked) {
  console.log(`Threat detected: ${result.reason}`);
}

PII Redaction

// Redact PII before sending to LLM
const result = await pg.security.redact(
  'My email is [email protected] and SSN is 123-45-6789'
);

console.log(result.redactedContent);
// Output: "My email is [EMAIL] and SSN is [SSN]"

Secure Web Scraping

// Scan scraped content for indirect prompt injection
const result = await pg.scrape.scanUrl('https://example.com/article');

if (result.isSafe) {
  // Use content safely
  console.log(result.content);
} else {
  console.log('Threats detected:', result.threats);
}

AI Agent Security

// Validate tool calls before execution
const validation = await pg.agent.validateTool({
  agentId: 'agent-123',
  toolName: 'execute_shell',
  arguments: { command: 'ls -la' }
});

if (validation.action === 'block') {
  console.log(`Blocked: ${validation.reason}`);
}

Configuration

import { PromptGuard, Config } from '@promptguard/sdk';

const config: Config = {
  apiKey: 'pg_xxx',
  baseUrl: 'https://api.promptguard.co/api/v1',
  timeout: 30000,
  enableCaching: true,
  enableSecurityScan: true,
};

const pg = new PromptGuard(config);

Environment Variables

export PROMPTGUARD_API_KEY="pg_xxx"
export PROMPTGUARD_BASE_URL="https://api.promptguard.co/api/v1"

Then just:

import { PromptGuard } from 'promptguard-sdk';

const pg = new PromptGuard(); // Uses env vars automatically

Error Handling

import { PromptGuard, PromptGuardError } from '@promptguard/sdk';

try {
  const response = await pg.chat.completions.create({...});
} catch (error) {
  if (error instanceof PromptGuardError) {
    if (error.code === 'BLOCKED') {
      console.log(`Request blocked: ${error.message}`);
    } else if (error.code === 'RATE_LIMITED') {
      console.log('Rate limited, try again later');
    }
  }
  throw error;
}

TypeScript Support

Full TypeScript support with type definitions included.

import type {
  ChatCompletionRequest,
  ChatCompletionResponse,
  SecurityScanResult
} from '@promptguard/sdk';

Links

• Documentation
• GitHub
• Issues

License

MIT