Enoki Labs

Want to test your AI agent before an attacker does? 🛡️ Here's what a test with Enoki looks like. 🔗 You paste your agent's endpoint. No install, no code access. 🔍 Enoki learns what your agent can do. ⚔️ It attacks your agent like a real attacker would. 🚨 When something breaks, you see the exact attack and what your agent did. ✅ You fix it. Enoki runs the same attack again to confirm the fix holds. Start with one agent. The first few findings are on us. 🎁

We’re opening Enoki Labs up to all agent builders. 🚀 For the past few months, we have been testing our design partners’ agents by hand: inputs, memory, tool calls, retrieval, outputs. Today, we’re opening it up to everyone for the next two weeks. “Test your agent” is now live on our site. Connect the place where your agent lives, and Enoki starts testing it automatically. If you’re building an agent, give it a try. Link in the comments. 👇

🚀 We just launched our MCP 🚀 After building the first version of Enoki Labs, the next step was clear: bring security testing into our customers' existing workflow. And if there is one thing developers love these days, it's their coding agent. So we brought Enoki directly into it. 🎯 You can now run security tests on your AI agents directly from your coding environment. React with "MCP" if you're interested in trying our agentic security MCP and getting early access. 📩

We just launched our MCP 🚀 After building the V1 of Enoki, the next step was clear: bringing security testing into your existing workflow. And if there is one thing I discovered, it's that developers love their coding agent. So we brought Enoki directly into it. 🎯 You can now run security tests on your AI agents from your coding environment. Interested in trying it out? Send me a DM and we'll give you early access. 📩 #agentsecurity #aisecurity

Your agent talks too much - It handed a stranger your secrets. We find that before they do.

Enoki hacks your AI 🍄🟫 - Better us than a real attacker. You get the findings and the fixes.