Connections
Este conteúdo não está disponível em sua língua ainda.
Connections live in the backend’s local database; passwords are stored in the OS keyring (Windows Credential Manager, macOS Keychain or the Linux secret service) and never leave it.
Switching connections
Section titled “Switching connections”The connection switcher in the header (Ctrl+Alt+O) lists recent
connections first, then folders. Type to filter by name or host. Each entry has
actions to connect, edit, duplicate or remove the connection. Connections and
folders are sorted alphabetically, ignoring case and accents, with numbers in
natural order (db2 before db10). New connection opens the connection
dialog.
Creating a connection
Section titled “Creating a connection”The connection dialog walks through:
- Database type: PostgreSQL, MySQL, MariaDB, SQL Server, Oracle, MongoDB, S3, generic JDBC or ODBC.
- Connection details: host, port, database or service, user and password, or a JDBC URL / ODBC connection string.
- Options: folder, environment, SSL, MCP access and which schemas to load.
Test connection shows the result in a banner before you save. Closing the dialog with unsaved changes asks for confirmation.
Per-database fields and limitations are listed in Database support. S3 and DuckLake settings are described in Local analysis.
Folders
Section titled “Folders”Open Manage connections to create and rename folders, and to move connections in bulk: select them, or drag whole rows onto a folder.
Environments
Section titled “Environments”Tag a connection with an environment so you always know where you are. The badge appears in the header, in the switcher and on editor tabs, in the environment’s color.
Development, Staging and Production exist by default. Choose Environments… in the connection dialog or in Manage connections to add your own (for example QA), rename them, change their color, reorder them or remove them. Restore defaults brings back the original three. The list is stored on this computer and is not included in connection exports. If you remove an environment that connections use, they show no environment until you pick another.
MCP access
Section titled “MCP access”Each connection controls what AI assistants connected through MCP can do with it: Off, Read-only (default) or Read and write. See MCP integration.
Export and import
Section titled “Export and import”In Manage connections, select connections and choose Export to save them to an omni-sql JSON file (Import… is next to it). The export keeps label, type, endpoint, user, folder, environment, MCP access, SSL and selected schemas. Passwords are never exported.
Import accepts an omni-sql export or DBeaver’s data-sources.json, with a
preview to choose which connections to bring in:
- Existing connections are flagged; folders are reused or created.
- DuckLake catalogs are re-linked to PostgreSQL connections imported in the same batch.
- DBeaver PostgreSQL, MySQL, MariaDB, SQL Server and Oracle connections are supported; others are listed with the reason they were skipped.
- DBeaver’s
credentials-config.jsoncan also be read to fill in users and passwords. It is decrypted in memory and passwords go straight to the keyring.
Metadata
Section titled “Metadata”omni-sql caches schemas, tables, columns and other objects per connection to power the sidebar and completion. Refresh a connection’s metadata from the sidebar after schema changes. With the sidebar open, Ctrl+P searches database objects.