Savant Environments

A Savant Environment is a single container image holding one real security defect in one real open-source program, together with everything needed to work on it: the source, a test suite, the tooling to build and run it, and a server that exposes a small, fixed set of operations over the Model Context Protocol.

You supply the agent. We supply the environment, the task, and the grader.

Where to start

• Quickstart — get a key, log in, and pull your first environment.
• Running an environment — which way to start it, and why the development server is not the one to score with.
• Using an environment — drive it, and read the grade.
• Tags and entitlements — which versions your organization may pull.

Browse what is available in the catalog — sign in with your organization first; it is visible to customers rather than to the public.

How access works

Your organization holds grants. A grant names an environment and the tag patterns you may pull from it.

A registry proxy checks every pull against those grants. If you are not entitled to an image, the registry reports it as unknown rather than forbidden — a "forbidden" would confirm the image exists, so a denial and a typo look the same from outside.