Dethernety in two minutes

If you have two minutes, this video is the shortest introduction to Dethernety. It models Google's Online Boutique, a public demo application with twelve pods, six trust zones and one GKE cluster.

The model comes from the repository. Dethereal, an open-source plugin for AI coding agents, reads eleven kinds of source, including Terraform, Helm charts, CI pipelines and database schemas. It proposes elements, and you confirm them before anything is written. Dethereal runs in your coding agent, which sends what it reads to the model provider you configure.

When the repository changes, a re-run models only what changed, so the threat model can follow each release.

Findings come from classes. A class is a reusable type, such as an API gateway, with its own questions and policies. The policies turn your answers into exposures, and each exposure traces back to the rule and the answers behind it.

The platform runs on your own infrastructure. If you would rather draw the model yourself, the browser editor works on the same model.

Read the quick start to try it.