JONATHAN

Hey everyone and hi product hunt,

I want to share what I ve been working on. The current trend in AI agent security is using another LLM as a "guardrail" to check if a prompt is malicious. I think this is fundamentally broken it's slow, probabilistic, and gets bypassed by indirect prompt injections.

AI agents are the ultimate "Confused Deputy." They hold high-privilege API keys, but their reasoning can be hijacked by untrusted data (like a Jira ticket or an email). The real damage doesn't happen at the prompt layer; it happens at execution.

So I built Relay a runtime security proxy that sits between the agent's decision and the actual tool execution.

Building Relay a runtime security layer for AI agents because my own agent exposed live database credentials with zero warning. No checkpoint, no policy, nothing. Just decided and acted.

Curious if others have had close calls. What's the most dangerous thing an agent tried to do in your setup whether it got stopped or not?