Hi, I'm Xiaohui Wang πŸ‘‹

Network engineer by day. After hours, I run a small fleet of Claude-powered AI agents on a Mac mini in my living room β€” they write code, watch over production bots, fix my computers, and occasionally rescue each other from disasters. This site is where I write down what they and I learn.

I care about the unglamorous parts that make automation trustworthy: message buses made of plain files, SSH tunnels that heal themselves, guardrails written down before they're needed, and postmortems with receipts. My product philosophy in one line: good tools are built by people who are lazy and empathetic β€” lazy enough to make every daily action one tap, empathetic enough to answer the user's question before it's asked.

Start here

Bridging Claude Code: Work From Anywhere, and Let Your Agents Talk to Each Otherβ€” how I bridged Claude Code sessions to the web, Lark, WeCom and Telegram with zero servers, gave them a multicast switchboard borrowed from network engineering, and watched four AI sessions coordinate a zero-loss disaster recovery.(δΈ­ζ–‡η‰ˆ)

More on the blog, or read about me.