Skippr Docs

Appearance

How to Configure MotherDuck as a Warehouse Destination ​

May 2026

MotherDuck warehouse setup is compact, but it still benefits from a deliberate database and schema choice before the first load lands.

Short Answer ​

Configure MotherDuck by setting warehouse.kind to motherduck, then supplying motherduck_token, database, and schema. The cleanest setup keeps the token in MOTHERDUCK_TOKEN and points Skippr at a database and schema that are meant to hold raw landing tables rather than mixing ingestion with unrelated analyst-owned objects.

A practical first configuration is a dedicated database such as my_database and schema main, with a token that can create or write the necessary objects there. Because the destination surface is small, most of the value comes from choosing a namespace that will still make sense after more sources and CDC-managed tables accumulate.

Why Teams Struggle with This ​

MotherDuck is easy to connect, so teams sometimes skip the design questions that matter operationally. A token may authenticate successfully and still be attached to the wrong database, or a shared schema may become noisy once multiple integrations start landing tables into it.

  • The token should be injected through MOTHERDUCK_TOKEN, not stored inline.
  • The selected database and schema must already exist or be writable by that token.
  • The runner needs outbound network access to MotherDuck.
  • A casual first schema choice can create cleanup work later when more pipelines arrive.

How Skippr Handles It ​

Skippr keeps the MotherDuck destination intentionally small: token, database, and schema. That makes it a strong option for teams that want a low-friction warehouse setup but still care about a repeatable connector contract and the ability to grow into CDC final-state behavior later.

The first working run should be easy to validate. Skippr can authenticate, create or target the expected tables, and keep all landed objects inside a namespace that is obvious to both platform engineers and analysts.

  • Direct support for MotherDuck token-based authentication.
  • Config centered on database and schema rather than a larger warehouse control plane.
  • Write access checks that map closely to the permissions in the docs.
  • A simple landing zone that can later host CDC-managed final-state tables.

What the First Useful Version Looks Like ​

Choose the database and schema as if they will hold many integrations, because that is usually what happens once the first MotherDuck load proves useful.

If the connector authenticates but writes fail, check schema-level rights first before assuming the token itself is invalid.