Datanika MCP server
{
"mcpServers": {
"datanika": {
"command": "uvx",
"args": [
"datanika-mcp"
],
"env": {
"YOUR_KEY": "<your_key>"
}
}
}
}{
"mcpServers": {
"datanika": {
"command": "uvx",
"args": [
"datanika-mcp"
],
"env": {
"YOUR_KEY": "<your_key>"
}
}
}
}{
"mcpServers": {
"datanika": {
"command": "uvx",
"args": [
"datanika-mcp"
],
"env": {
"YOUR_KEY": "<your_key>"
}
}
}
}{
"servers": {
"datanika": {
"command": "uvx",
"args": [
"datanika-mcp"
],
"env": {
"YOUR_KEY": "${input:your_key}"
}
}
},
"inputs": [
{
"id": "your_key",
"type": "promptString",
"description": "Your key",
"password": true
}
]
}{
"mcpServers": {
"datanika": {
"command": "uvx",
"args": [
"datanika-mcp"
],
"env": {
"YOUR_KEY": "<your_key>"
}
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsoninstall uvx datanika-mcpversion 0.2.0
What it does
MCP server for Datanika, an open-source data pipeline platform. Exposes tools to browse data sources, run dbt transformations, and manage ELT pipelines. For developers integrating Datanika operations into AI agents and LLM workflows.
- Query and inspect data from 35 supported connectors (PostgreSQL, MongoDB, BigQuery, Stripe, etc.) without executing transforms.
- Trigger and monitor dbt transformations, including model runs, tests, and snapshots with run history and logs.
- Create, schedule, and manage visual data pipelines with DAG dependencies and cron-based scheduling.
Tools this server exposes
| Tool | Does |
|---|---|
| browse_connections | List and inspect available data source connections. |
| query_data | Execute queries against connected data sources. |
| list_dbt_models | List available dbt models and their metadata. |
| run_dbt_model | Execute a dbt model transformation. |
| list_pipelines | List all configured data pipelines. |
| get_pipeline_status | Retrieve the current status and run history of a pipeline. |
| trigger_pipeline | Start a pipeline execution. |
| get_run_logs | Fetch logs from a pipeline or transformation run. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.