FAQ

PromptShield — scan AI apps for prompt-injection and jailbreak risk before you ship.

What does PromptShield do?

Describe your AI app and optionally paste untrusted user input. You get deterministic pattern hits, jailbreak signal flags, and a hardening checklist (ruleset prompt-injection@2026-07-19.2).

Is this a penetration test?

No. It is decision-support for LLM/agent product teams. It is not a formal pentest, red-team engagement, or compliance attestation.

Do you guarantee 100% detection?

No. OWASP notes that fool-proof prevention of prompt injection is unclear given how generative models work. PromptShield does not guarantee 100% coverage or that you will never miss a bypass.

What standards do you align with?

We reference OWASP LLM01:2025 Prompt Injection and the OWASP Prompt Injection Prevention Cheat Sheet as industry guidance — not as a certification seal.

What happens if AI is unavailable?

Fail-closed: demo is labeled; missing configuration and upstream failures return errors. Never silent mock as live AI.

How does pricing work?

$29/month or $290/year via /api/checkout with monthly or yearly cycle. Enterprise can include BYOK/DPA.

Policies & contact

Privacy · Terms · lixingliangsy@163.com