Security and governance

Make security boundaries part of the product model.

CXRove is designed around tenant isolation, server-owned authority, least privilege, explicit release identity, secret-safe operations, and fail-closed availability. This page describes principles, not certifications.

Direct answer

What security model does CXRove use?

CXRove is designed around organization isolation, server-derived authority, least-privilege workloads, explicit release identity, privacy-safe telemetry, and fail-closed operations. Security claims remain tied to their exact scope and evidence; the product does not turn a design principle into an unsupported certification or service guarantee.

Keep organization authority server-owned

Tenant context should come from authenticated, active organization and membership state and be enforced in application code and PostgreSQL. Supplied models, headers, and job payloads do not grant access.

Separate runtime and administrative powers

Web, worker, support, migration, provider, and recovery workloads need distinct least-privilege identities. Ordinary runtime processes should not inherit migration or secret authority.

Bind behavior to an immutable release

A production change should identify exact source, artifact, configuration, schema, public content, and selector state. Admission, activation, readback, compensation, and rollback are different steps.

Keep operational evidence privacy-safe

Use safe handles, booleans, counts, status categories, latency, saturation, and release identities. Exclude credentials, bodies, query strings, transcripts, and unrelated host inventory.

No certification claim is made

CXRove does not claim a third-party security, privacy, healthcare, or payment certification; a fixed retention period; a service-level commitment; or a completed independent audit from this page.

Common questions

Answers for a practical evaluation.

Is CXRove certified for a compliance framework?

No certification is claimed. Any future statement must identify its exact scope, evidence, version, and limitations.

How is tenant isolation intended to work?

Through trusted authority, application scoping, PostgreSQL row-level controls, workload-role separation, and fail-closed connection hygiene.

What production proof remains separate?

Host, credentials, providers, traffic, monitoring, backup and restore, incident operations, and independent assessment require live evidence.

Next action

Turn a customer conversation into a completed next step.

Choose an Agent capacity, define the first workflow, and decide what the Agent may know, do, and hand to a person.