I’ve been building my own “OpenClaw” like thing with go-mcp and cloudflare tunnel/email relay. I can send an email to Claude and it will email me back status updates/results. Not as easy to setup as OpenClaw obviously but alt least I know exactly what code is running and what capabilities I’m giving to the LLM.
I think it's worth it to build your own miniaturized versions of OpenClaw/claw-like agents. It's easy enough to build and the confidence of it being in a language you're familiar with, small enough surface area to limit risk, etc. seems worth it imo
