Building security tools for AI agents.

Barrikade is building one security layer to find AI agents, connect them to an owner, catch unsafe behaviour and show exactly what happened.

Where it started

Barrikade Core began as an undergraduate research project developed with design input from Jentic. It began with prompt-injection detection and grew into a working-alpha system that can watch a full agent session, spot when it moves off task and record the response.

What exists today

Barrikade Lens is an open-source tool that finds agents across endpoints, repositories, CI and Kubernetes. Barrikade Core is a working-alpha runtime security engine. The identity layer and the connection between all four stages are still being built.

What comes next

Barrikade is now connecting Lens, Core and the identity layer into one lifecycle. The next work focuses on helping those stages share context while keeping each maturity boundary clear.