Give every agent an identity and boundary.

Security teams and autonomous-system operators

Rubian gives people and agents scoped users, homes, services, persistent state, and direct system controls.

Identity is part of the operating system.

Authenticated users can own durable workspaces, services, and persistent state.

Services remain attributable.

Service control and lifecycle records keep durable work connected to a responsible user.

Recovery begins with known state.

Versioned baselines, manifests, checksums, and isolated tests make it possible to return to a known system state.

Security work is still in progress.

Networking, process isolation, device control, and shared runtime behavior require more implementation and testing before production deployment.

How Rubian and Kestowv got here.See where Rubian runs today.

Set the limits before the failure.

An unlimited restart loop is still a failure. Before a service goes into a pilot, define how health is checked, how many retries are allowed, how long to wait and when the operator takes over.

Kestowv's pilot supports finite restart budgets, backoff and controlled drain. Test those policies deliberately. A process that starts again hasn't necessarily recovered the job it was doing.

Recovery also needs a boundary around who can act.

Reading status and changing a service are different responsibilities. An evaluation should identify who can inspect the system, who can stop or restart work, and how those changes are recorded.

Keep application data recovery, service access and the release rollback plan explicit. The current hosted pilot is for trusted workloads; it isn't a promise of hostile-tenant isolation.

Discuss a recovery-focused evaluation

Evaluating reliability in your own environment? Explore reliability services

Rubian/Kestowv Documentation

Startup and discovery, file and project organization, Forge libraries, Taskman, agent work and Kestowv pilot operations.