FAQ

Frequently asked questions

Straight answers about how the autonomous cloud works, what it will and won't touch, and what early access involves.

No. The agent reads your repository as it is and derives services, dependencies and runtime requirements from the code. No platform SDK, manifest or annotations are required. An optional SHIPSANELY.md can add hints if a project is unusual.

Early access focuses on Node.js/TypeScript, Python, Go and Ruby web services with PostgreSQL and Redis, plus static frontends. Java, .NET and Rust are under evaluation. Support broadens continuously — ask us about your stack.

The previous release keeps serving traffic. The agent reports exactly what failed and why. A release is only promoted after it has provisioned infrastructure, started every service, connected to its datastores and answered requests correctly.

Depending on the diagnosis: restart an unhealthy service, roll back to the last verified release, re-provision a degraded server, adjust a configuration such as a connection pool, or redeploy. Every action is logged with cause, fix and outcome.

Repairs are limited to operational actions on infrastructure the platform manages. The agent never modifies your source code or commits to your repository. You can review the incident log and set repair to "notify only" for critical services.

No. Both run on a private network and are reachable only from your application servers. Only hosted application endpoints are public. Local access is via an authenticated CLI tunnel.

Two meters. AI agent usage is measured by work performed — analysis runs, builds and verifications, diagnoses and repairs. Infrastructure usage is billed per hour per managed component. See the pricing page for starter examples.

Yes. Point a CNAME at your hosted endpoint and the agent provisions TLS automatically.

On managed infrastructure operated by ShipSanely. Region selection is available during onboarding for early-access partners.

Request early access and tell us about your repository. We onboard a limited number of teams at a time so the agent can be tuned to each stack.

Still have a question?