DuckDB
Connect a DuckDB database file for federated analysis. The file's tables and views are mounted into the query engine and can be joined with datasets from other connectors.
When to use
- An analytical dataset packaged as a single
.duckdbfile. - Results exported from another DuckDB pipeline that you want to analyze conversationally.
How the file is provided
Like SQLite, DuckDB is a single-file database, so Datyo needs access to that file:
- Object storage (managed service): the file lives in an S3-compatible bucket and is read using the storage credentials you provide. Used on datyo.ai.
- Local file path (desktop / self-hosted): the connector reads the file directly from a filesystem path. Not available on the managed service — see Managed vs. self-hosted.
Connection settings
Object-storage mode
Provide the object-storage access credentials, plus:
| Field | Required | Description |
|---|---|---|
| Bucket | Yes | The bucket that holds the database file. |
| Object path | Yes | The key of the file within the bucket, e.g. warehouse/analytics.duckdb. |
Local file path (desktop / self-hosted)
| Field | Required | Description |
|---|---|---|
| File path | Yes | Absolute path to the .duckdb file in the environment where Datyo runs. |
Capabilities
- Execution model: Federated. DuckDB tables can be joined with datasets from other connectors.
- Access: Read-only. The database is attached without modification.
- Tables and views: Both are listed for import.
- Environment: Object-storage mode works everywhere; local file path is desktop / self-hosted only.
Troubleshooting
- File not found or empty — verify the bucket and object path (or local file path) and read access to the object.
- A view is missing — re-sync; tables and views are both enumerated once the file is attached.