Clone the whole Supabase project.

Install with the one-liner (or npm install -g twinstack if you have Node and psql), then run twinstack clone or twinstack ui. The engine works through the pipeline: schema → data → auth users → storage buckets → storage files → edge functions → config. The end report lists the few values Supabase keeps unreadable so you can paste them on the target.

Yes — this is the differentiator. Database backups only contain the metadata rows; the actual blobs live in S3-backed Storage. Twinstack walks every bucket, streams each object source → target with bounded concurrency, and tracks completion per file so a dropped connection at file 9,000 of 12,800 resumes at file 9,001.

Twinstack does a data-only migration of auth.users, auth.identities, and related tables with ON CONFLICT DO NOTHING so the target's pre-seeded auth schema doesn't collide. The encrypted_password column (bcrypt) transfers, so users log in on the copy with their existing passwords. Sessions and refresh tokens do not transfer — the JWT secret is per project, so a one-time re-login is expected.

Yes. Twinstack runs supabase functions download against the source for each function, then deploys to the target with supabase functions deploy, preserving per-function config (verify-JWT, import map, entrypoint). Function source code transfers; secret values do not — Supabase makes those write-only on purpose. The checklist tells you exactly which ones to set on the target.

Run twinstack env add prod-to-staging once to define the pair, then twinstack drift prod-to-staging for a read-only summary of what has changed since the last sync, and twinstack sync prod-to-staging to re-clone on demand. Storage files resume per-object, so subsequent syncs are much faster than the first clone.

Free covers projects under 100 MB, with 1 run to try Twinstack end to end. The full feature set is included on free — no degraded engine, same auth user migration, same storage transfer, same secrets checklist. You upgrade when your project size crosses 100 MB or you've used your 1 free run. Unlimited is a flat $10/year, no per-gigabyte metering.

Yes — they live in the schema dump alongside tables, and come along automatically. Custom-role passwords are not in any backup (security), so Twinstack lists the roles for you to reset on the target.

Those errors happen when a naive pg_dump includes the auth or storage schemas — both are pre-seeded on every new project. Twinstack avoids the issue entirely by dumping only public and your custom schemas, and migrating the auth schema data-only with ON CONFLICT DO NOTHING.

The CLI runs on your machine and talks directly to Supabase — credentials never reach our servers, because the default mode has no server. The engine is open source: read what it does with your keys before you run it. After a run you can revoke every token from your Supabase dashboard in one click. The control plane only sees the project size you report — never the data.

The Mac/Linux script checks for Node 18+ and pg_dump/psql, and uses Homebrew or your distro's package manager to install only what's missing. The Windows script uses winget for the same. Then it runs npx twinstack@latest ui which downloads the CLI on demand. Re-running is idempotent — it skips anything already installed.