A security model built on separation and ownership

The control plane owns identity, projects, specifications, generation jobs, artifacts, validation evidence, and deployment records. The backend it generates for you is a separate service: it runs your business logic, on infrastructure you choose, from source you can read.

That separation is a security property, not just an architecture diagram. Platform credentials are not injected into your generated deployment, and your generated runtime does not need access back into the control plane to serve its API.

Sign-in and session management are handled by a dedicated authentication provider rather than home-grown password storage — the privacy policy names the processors involved. Inside the platform, projects, agents, specs, artifacts, and logs are scoped to your tenant, and requests for another tenant’s resources behave as if those resources do not exist.

Quota checks run before expensive side effects. A request that would exceed your plan is refused up front instead of queued and billed.

You retain the rights to the specs you write and the code Axtra generates for you, and your content is not used to train shared models. The artifact is downloadable as source, so an audit never depends on our description of what the code does — you can read it.

Credentials for your deployment target belong to your environment. Deploy preflight verifies configuration and reports what is missing; it does not harvest working credentials into marketing claims, and it never returns a synthetic success URL.

If you believe you have found a vulnerability in Axtra Intellion, email hello@axtl.dev with enough detail to reproduce it. Please do not test against other tenants’ data; a report with your own workspace as the subject is always enough.