Rork to production: make the app dependable without erasing the idea

Rork prototypes tend to reveal their first cracks around mobile lifecycle changes rather than the happy path: a token expires after backgrounding, a retry posts an order twice, or a network drop leaves the screen claiming success. We also inspect client-held service keys, optimistic writes without idempotency, and push or deep-link paths that were only exercised on one developer device. File uploads and webhook retries deserve their own failure tests.

For Rork, production-ready means identities survive reinstall and refresh, data has an owned schema with migrations and backups, and builds come from a CI workflow with signed artifacts rather than a laptop export. Store readiness adds privacy disclosures, permission rationale, crash symbolication, and a tested upgrade path. A polished Rork screen is useful evidence of product fit, not operational proof.

We harden Rork when its navigation and domain model still express the product customers should receive. A targeted rewrite is justified where generated state leaks into billing, ownership rules are missing, or native capabilities are boxed into brittle workarounds. The decision comes from a flow-by-flow risk map: retain stable presentation, replace the few paths whose failure would corrupt money, identity, or trust.

First value is one real customer completing the primary job, receiving a durable result, and giving the team a traceable support path. A coherent Rork build can reach that checkpoint after the critical journey has a production identity, resilient writes, analytics, and a rollbackable binary. We sequence the work around that journey instead of polishing secondary screens or promising a calendar date before the runtime is known.