Data sources
Connect the databases and warehouses your Projects query. The Data Sources page (/data) has up to three tabs: Catalog (the providers you can connect), Connected (the data sources your workspace has set up), and Profiles (Snowflake profiles, workspace admins only).
Audience: IT (the enabler), workspace admins, and team admins
Who can manage data sources
Workspace admins and the current active team's admin can connect data sources. Non-admins see a hard-block surface naming the admins who can help.
Where new connections land. New data sources land in the active team. Snowflake profiles remain workspace-wide. The page subtitle calls this out for the people who can see it.
Supported providers
The Catalog ships ten provider connectors. Each has its own connection form with provider-specific fields.
| Provider | Category | Sign-in methods |
|---|---|---|
| Snowflake | Cloud warehouse | OAuth (directly or through a workspace Snowflake profile), key pair, access token |
| Postgres | Relational database | Password |
| TimescaleDB | Time-series database | Password |
| BigQuery | Serverless warehouse | Service account |
| Databricks | Lakehouse SQL | Token, OAuth machine-to-machine |
| Redshift | Cloud warehouse | Password, IAM |
| MySQL | Relational database | Password |
| ClickHouse | Columnar analytics | Password |
| Microsoft SQL Server | Enterprise database | Password, Entra service principal, OAuth |
| Oracle | Enterprise database | Password |
Snowflake has a dedicated dialog and the additional Profiles tab. The other nine share a composer pattern: pick the provider on the Catalog tab, fill the provider-specific form, save.
Check a connected source
The Connected tab counts your connections, how many are still being read, and how many are ready to query. Select a source to open its details.
After you connect a source, the platform reads its structure first, then profiles each table in the background. Overview shows that progress, the last sync and test, and Readiness: how many schemas and tables were discovered and how many are profiled. Test connection checks only that the source can be reached. Refresh tables reads the source's schemas, tables, and columns again, for example after tables were added or changed.
-
Schema lists every table with its column and row counts and its profiling status.
The Schema tab: each table with its column and row counts, marked Profiled -
History and Recent activity show what happened to the source, with a link to the full trail in Audit.
-
Delete data source removes it from the workspace.
Query log
A source can keep a log of the queries run against it, with a short preview of each result. New sources start with the log off. Sources connected before version 3.3.3 keep their setting.
To change it, open the source and go to Settings > Query log:
- Record queries turns the log on or off. Turning it off stops new entries, and existing entries still expire.
- Keep queries for (days) is 1 to 90 days, 30 by default.
- Maximum retained queries is 1 to 30,000, 10,000 by default.
The oldest entries are removed when either limit is reached, and lowering a limit also removes older entries. Workspace and team administrators of the source can change these settings.
What Projects can do with a source
Connecting a source does not give any Project access to it. A Project owner or administrator attaches the source to a Project and chooses which tables it sees, and which of those it may change: add, update, or delete rows, either through approved verified queries or directly. New attachments start read only.
The credentials you connect with are always the upper limit: no Project can do more than the connection's own database role allows. Connect with a least-privilege role, and grant write privileges only when a Project's work genuinely needs to change data. For actions outside a database, such as sending email, use an integration.
See How the agent uses your data for what the agent does with an attached source.