Connections
Connect MinusX to your databases
MinusX connects to your existing databases to query and analyze data. Connections are configured once and shared across your workspace.
Supported databases
| Database | Type | Notes |
|---|---|---|
| DuckDB | Local file | Default for local development. Fast, embedded analytics. |
| PostgreSQL | Remote | Production databases, data warehouses |
| BigQuery | Cloud | Google Cloud data warehouse |
| Athena | Cloud | AWS Athena over S3 |
| ClickHouse | Remote | Column-oriented OLAP databases |
| SQLite | Local file | Single-file databases |
| CSV Files | File upload | Upload CSV files — automatically converted to a queryable DuckDB database |
| Excel (XLSX) | File upload | Upload spreadsheets — converted the same way as CSV |
| Google Sheets | Cloud | Import public Google Sheets as queryable tables |
Coming soon
| Database | Type |
|---|---|
| Databricks | Cloud |
| Snowflake | Cloud |
Configuring a connection
- Navigate to Connections in the sidebar
- Click New Connection
- Select your database type
- Enter connection details (host, port, database, credentials)
- Test the connection
- Save
Connection credentials are stored securely in the documents database. For production deployments, use environment variables for sensitive values.
Allowing access through a firewall
If your database only accepts connections from known addresses, allow the IPs MinusX connects from. On MinusX Cloud those addresses are shown directly in the connection form for the engines it applies to — PostgreSQL and ClickHouse — and highlighted if a connection test fails, so you can copy them straight into your firewall rule.
BigQuery and Athena are reached over their public service endpoints and authorised by credentials rather than source IP, so no allowlisting is needed for those.
Self-hosted deployments connect from your own network, so the addresses to allow are your
host's — MinusX cannot know them, and the hint is hidden. Operators who want it shown can
set MX_EGRESS_IPS to a comma-separated list of the deployment's egress addresses.
How connections are used
- Questions execute SQL against a specific connection
- Knowledge Base loads schema information from connections for table whitelisting
- AI Agent uses connection schemas to write accurate SQL
Each question is tied to a connection. When you create a new question, it uses the default connection unless you specify otherwise.