Viewing profile — mrajagopalan

We reduce agentic systems to byzantine distributed systems and protect 4 boundaries — prompts, tools, data and context -- via a trust layer that enforces intent (operations satisfy…

OpenClaw highlighted the core problem with securing AI agents: you can't enumerate every malicious prompt. Pattern matching catches "rm -rf" but misses "please remove all files." D…

A bit late to this discussion — but we've been looking at this problem for a while and have implemented a cryptographic approach I wrote about here: https://news.ycombinator.com/it…

Following the MCP security discussion - https://news.ycombinator.com/item?id=45199713 I've written up our approach to securing AI tool calling through cryptographic enforcement rat…