Ice Phi is a high-performance security gateway and SDK ecosystem (Python & TypeScript) that blocks prompt injections, jailbreaks, and malicious LLM payloads at the edge with sub-millisecond latency. Integrate drop-in middleware into your Express, Fastify, or Python app in 3 lines of code to protect your AI agents and chat endpoints from sophisticated attacks in production.

Hey Product Hunt! I'm the builder behind Ice Phi. If you're building LLM apps or AI agents right now, you already know that prompt injections and jailbreaks are a constant headache. Most guardrails are either too slow, hard to set up, or require rewriting your backend. I built Ice Phi to fix that with drop-in Python and TypeScript SDKs (@ice_phi/icephi-ts on npm and icephi-python on PyPI) that plug straight into your app as lightweight middleware. You can grab a free API key and test your own jailbreak payloads against the sandbox in seconds. I'd love to hear your feedback, see if it breaks under your specific use cases, or answer any questions about how we handle the latency overhead! Let me know what you think.

Hey Product Hunt! I built Ice Phi - Enterprise Shield to stop LLM prompt injections and jailbreaks without ruining chat latency, combining a high-performance edge security gateway with native TypeScript (@ice_phi/icephi-ts) and Python (icephi-python) SDKs that drop right into Express or FastAPI in just three lines of code. Grab a test key in our developer portal to run your own injection attempts against our live sandbox today, and let me know in the comments how your team is currently handling AI security guardrails!

Ice Phi - Enterprise Shield

Prompt injection API shield

Prompt injection API shield

Forum Threads

Why we built Ice Phi (and how we solved LLM guardrail latency)

Forum Threads

Why we built Ice Phi (and how we solved LLM guardrail latency)