Building things nobody asked for, then open sourcing them so everyone can have things they didn't ask for.
Mostly interested in AI security, poking around the OpenClaw wild west, and writing small apps that have no business being as useful as they are.
Automation enthusiast in the sense that I'll spend 6 hours automating something that takes 4 minutes.
Deploying for fun, not profit, which is handy because that's usually how it works out anyway.....
London, UK
SkillShield — The security-scored directory for AI agent skills
https://skillshield.dev/
I got paranoid and scanned 33,746 AI agent extensions. Turns out 12% are actual malware, credential stealers, reverse shells, data exfiltration. Not "potentially suspicious." Real attacks waiting to happen.
The problem: AI agents are getting powerful (browsers, code execution, file access), but we're plugging in random "skills" without checking what they do. It's like installing browser extensions from a stranger in a trenchcoat.