MotherDuck

Loader

MotherDuck is a cloud data warehouse built on DuckDB, aimed at teams who want fast, serverless SQL analytics without managing infrastructure. This connector loads data from any of your Meltano data sources into a MotherDuck database, so it lands somewhere you can query it directly or build on with other tools.

Note: MotherDuck is a loader, not a data source: it doesn't have its own streams, sync type or custom queries to sync from. The facts below reflect that.

At a glance

Property	Value
Authentication	API key / token
Loads via	Standard Singer records, plus Arrow- or JSONL-encoded batch messages with no extra configuration
Target database	Must already exist: this connector does not create it for you

What it loads

Whatever streams your connected data sources send it: one table per stream, created in the schema you configure.

Prerequisites

• A MotherDuck database that already exists. This target won't create it for you.
• MotherDuck Token: your MotherDuck access token.
• MotherDuck Connection String: e.g. md:my_database, pointing at the database above.

Setup

In MotherDuck

1. Create the database you want to load into, if it doesn't already exist.
2. Generate a MotherDuck token.

In Meltano Cloud

1. Enter your MotherDuck Connection String (e.g. md:my_database) and Token.
2. Set Default Target Schema: the schema new tables are created in. Defaults to analytics.

Available streams

This target doesn't discover streams itself: it loads whatever streams the connected extractor(s) send it, creating one table per stream in your configured target schema.

Advanced configuration

• Batch size: Batch Size Rows controls how many rows load into DuckDB at a time (default 100,000).
• Flush behaviour: Flush All Streams flushes every stream after each batch, rather than only the one currently loading.
• Schema flattening: Data Flattening Max Level controls how deeply nested JSON is expanded into columns (0 = no flattening, the default).
• Metadata columns: Add Metadata Columns adds Singer's standard tracking columns to each loaded table.
• Deletes: Hard Delete performs an actual row delete instead of a soft delete when a source sends delete records.
• Validation: Validate Records validates each record against its schema before loading (off by default).
• Primary keys: Primary Key Required (on by default) rejects a stream that doesn't declare a primary key, since the target needs one to know how to upsert rows.
• Temporary files: Temporary Directory sets where temp files are written while loading.

Settings

Setting	Type	Description
connection_string	string	MotherDuck connection string, e.g. md:my_database.
token	string	MotherDuck access token.
default_target_schema	string	Schema new tables are created in. Defaults to analytics.
batch_size_rows	integer	Number of rows to load at a time. Default 100000.
flush_all_streams	boolean	Flushes every stream after each batch, rather than only the one currently loading.
add_metadata_columns	boolean	Adds Singer's standard tracking columns to each loaded table.
hard_delete	boolean	Performs an actual row delete instead of a soft delete when a source sends delete records.
data_flattening_max_level	integer	Maximum depth to flatten nested JSON into columns. 0 (default) disables flattening.
primary_key_required	boolean	Rejects a stream that doesn't declare a primary key. Enabled by default.
validate_records	boolean	Validates each record against its schema before loading. Off by default.
temp_dir	string	Directory used for temporary files while loading.

Need help?

If a stream or field you need isn't listed here or the connector doesn't work as expected, file it through the usual Meltano support channel.