Agent R&D · Security AI · Evaluation
Drop an agent into a sandboxed enterprise. Let it run.
Security agents need realistic networks before you trust them in front of a customer. Running an agent on a customer's network is a non-starter. Toy environments don't reveal real-world behaviour. And without a fair baseline, you can't honestly compare one agent against another.
Ludus gives every agent its own clean, isolated range. Pivot, escalate, run exploits — fully sandboxed. Snapshot, run, revert, run again. Reproducible networks mean apples-to-apples evaluation, every time.
Can't say enough good things about #ludus by @badsectorlabs. I've used a lot of Cyber Range management tools and none come close to how refined and reliable ludus is. Kudos to the team 👏👏👏
Apr 29, 2024 · on XOpen ↗
OutcomeA controlled adversary sandbox. Repeatable scoring, real behaviour.