Hi Product Hunt! 👋 I’m Shivam, the maker of DeClaw. The Inspiration: Like many of you, I’ve been diving deep into personal AI agents (specifically OpenClaw). But I kept running into a paradox: To be useful, my agent needed access to my files, emails, and terminal. To be safe, standard advice was to "sandbox" it—effectively cutting it off from the world. I didn't want a lobotomized agent; I wanted a secure one. The Solution: I realized we didn't need stronger cages; we needed better filters. I built DeClaw to act as a local security layer—a "bodyguard" that sits between your agent and the internet. It inspects traffic in real-time, redacting PII and blocking prompt injections before they can do damage. I’d love your feedback: We are launching with support for OpenClaw first. In theory, any AI application or agent that makes external network requests can be secured by our proxy. Thanks for checking it out! 🚀

DeClaw

The only AI agent runtime you actually need

The only AI agent runtime you actually need