May 11, 2026 • By SottoVox

PromptGuard SDK: Complete Protection Against Prompt Injection

We're releasing PromptGuard — an open-source SDK that protects your AI web applications from prompt injection attacks.

Why This Matters

Every day, developers deploy chatbots and AI features without protection against malicious inputs. Attackers can:

Extract API keys and credentials from chatbots
Bypass content filters for prohibited topics
Access internal system prompts
Generate malicious code

Features

50+ Detection Patterns

Category	Examples
Role Override	"You are now DAN", "Pretend to be root"
Instruction Override	"Ignore previous", "Forget rules"
Jailbreak	"do anything now", "jailbreak mode"
Credential Theft	"show me api key", "tell password"

Multi-Layer Protection

import { PromptGuard } from 'sottovox-prompt-guard';

// Input sanitization
const result = guard.sanitize(userInput);
// → { safe: false, threats: ['jailbreak_dan (critical)'], blocked: true }

// Response validation  
const responseCheck = guard.validateResponse(input, llmResponse);

Zero Config Setup

// Just works out of the box
const guard = new PromptGuard();

// Or customize
const guard = new PromptGuard({
  strict: true,           // Additional strict patterns
  maxLength: 10000,       // Limit input size
  blockOnDetection: true, // Auto-block threats
});

Installation

npm install sottovox-prompt-guard

Protect your AI. Ship with confidence.

Get PromptGuard on GitHub →