the wizard

I build AI agents and test them for made-up answers and tricks that leak data.

I build AI agents that do real work, with the security inside: tools that reach only what they need, your OK before risky actions, and tests you can rerun. I also test the harnesses agents run on, and publish every finding.

Work with me See a sample check-up

Research

All research

Independent tests of AI agent harnesses: who the model listens to, which attacks get through, which defences hold.

Series to read in order, and posts from real builds.

Projects

All projects

Open source, simulations to play, live tools, a model and benchmarks.