promptshield-firewall

Real-time LLM prompt injection firewall — detect and block jailbreaks, system prompt extraction, and adversarial attacks before they reach your model.

Installation

pip install promptshield-firewall

Quick Start

from promptshield import PromptShield

shield = PromptShield(api_key="ps_your_key")

Simple boolean check

if not shield.is_safe(user_input): return {"error": "Request blocked by PromptShield."}

Safe to send to your LLM

response = your_llm.chat(user_input)

Full Response

result = shield.scan(user_input) print(result["verdict"]) # BLOCKED / SUSPICIOUS / SAFE print(result["risk_score"]) # 0-100 print(result["attack_types"]) # ["jailbreak", "role_hijacking"] print(result["explanation"]) # Human-readable analysis print(result["confidence"]) # 0.0-1.0

Async Support

result = await shield.async_scan(user_input)

What It Catches

• DAN jailbreaks and 300+ known templates
• System prompt extraction attempts
• Role hijacking attacks
• Indirect injection via documents/URLs
• Context manipulation and memory poisoning
• Data exfiltration attempts

Fail-Safe Design

If PromptShield is unreachable, the SDK defaults to SAFE so your application never goes down.

Pricing

Free: $0 — 50 calls/month Pro: $99/mo — 25,000 calls/month Business: $499/mo — 500,000 calls/month Enterprise: Custom — Unlimited

Get Your Free API Key

https://promptshield.live/docs — no credit card required.

Links

• Website: https://promptshield.live
• API Docs: https://promptshield.live/docs
• GitHub: https://github.com/kehinde-code/promptshield

Author

Kehinde Omosor — Founder & CEO, PromptShield Inc. kehinde@promptshield.live