by Databricks CommunityVerified
Interact with Databricks workspaces from your AI coding assistant. Execute notebooks, run SQL queries on Unity Catalog tables, manage clusters, browse the workspace file system, and monitor job runs. The server provides comprehensive access to Databricks' lakehouse platform — query Delta tables, inspect data lineage, manage secrets, and submit batch jobs. Particularly valuable for data teams working across notebooks and SQL warehouses who want AI-assisted data engineering workflows.
npx mcp-server-databricksAdd the following to your client configuration file.
{
"mcpServers": {
"mcp-server-databricks": {
"command": "npx",
"args": [
"-y",
"@mcp/mcp-server-databricks"
]
}
}
}{
"mcpServers": {
"mcp-server-databricks": {
"command": "npx",
"args": [
"-y",
"@mcp/mcp-server-databricks"
]
}
}
}{
"mcpServers": {
"mcp-server-databricks": {
"command": "npx",
"args": [
"-y",
"@mcp/mcp-server-databricks"
]
}
}
}| Client | Status |
|---|---|
| claude-desktop | Full Support |
| cursor | Full Support |
| vscode-copilot | Full Support |
Run the install command above or copy the configuration snippet for your preferred MCP client.
Add any required API keys or authentication tokens to your MCP client configuration. Check the GitHub README for setup details.
Once configured, the server's tools become available in your AI assistant. You can access features like sql query execution on unity catalog and notebook execution and management.
Databricks MCP Server is an MCP (Model Context Protocol) server created by Databricks Community that allows AI coding assistants to interact with external services and data sources programmatically. Interact with Databricks workspaces from your AI coding assistant. Execute notebooks, run SQL queries on Unity Catalog tables, manage clusters, browse the workspace file system, and monitor job runs. The server provides comprehensive access to Databricks' lakehouse platform — query Delta tables, inspect data lineage, manage secrets, and submit batch jobs. Particularly valuable for data teams working across notebooks and SQL warehouses who want AI-assisted data engineering workflows. The server communicates via the stdio transport type and is licensed under MIT.
MCP servers like Databricks MCP Server extend the capabilities of AI assistants beyond code generation and into real-world integrations. Instead of manually switching between your AI assistant and external tools, the MCP protocol lets the assistant call structured APIs directly — reading data, triggering actions, and maintaining context across interactions. This reduces context switching and allows developers to stay in their IDE or terminal while the assistant handles the integration layer.
With 7 documented capabilities — including sql query execution on unity catalog, notebook execution and management, cluster lifecycle management — Databricks MCP Server is one of the more feature-rich MCP servers in its category. Its 2.4K GitHub stars indicate strong community adoption and interest.
Use Databricks MCP Server when you want your AI coding assistant to interact with interact with databricks workspaces from your ai coding assistant. execute noteb directly from the chat or terminal interface, without manual copy-pasting.
If you find yourself constantly switching between your AI assistant and external tools,Databricks MCP Server eliminates that friction. The assistant can read context, take actions, and report results — all within a single conversation thread.
MCP servers work well for semi-automated workflows where the AI handles repetitive integration tasks — fetching data, updating records, posting notifications — while you focus on higher-level decisions and code architecture.
Databricks MCP Server works across 3 compatible clients, including claude-desktop, cursor, vscode-copilot. This means you can use the same server configuration regardless of which AI coding tool your team prefers.
Several MCP servers offer overlapping functionality. The right choice depends on your specific integration needs, the AI clients you use, and your preference for community vs. official support. Here is how Databricks MCP Server compares:
Sentry MCP Server by Sentry is less popular by GitHub stars (1.2K vs 2.4K). Both servers are maintained by different teams — evaluate documentation quality and recent commit activity when choosing.
Elasticsearch MCP Server by Elastic Community is less popular by GitHub stars (520 vs 2.4K). Both servers are maintained by different teams — evaluate documentation quality and recent commit activity when choosing.
Datadog MCP Server by Datadog Community is less popular by GitHub stars (480 vs 2.4K). Both servers are maintained by different teams — evaluate documentation quality and recent commit activity when choosing.
PagerDuty MCP Server by PagerDuty Community is less popular by GitHub stars (320 vs 2.4K). Both servers are maintained by different teams — evaluate documentation quality and recent commit activity when choosing.
Databricks MCP Server allows AI coding assistants like Claude Code, Cursor, and VS Code Copilot tointeract with databricks workspaces from your ai coding assistant. execute notebooks, run sql queries on unity catalog tables, manage clusters, browse the workspace file system, and monitor job runs. the server provides comprehensive access to databricks' lakehouse platform — query delta tables, inspect data lineage, manage secrets, and submit batch jobs. particularly valuable for data teams working across notebooks and sql warehouses who want ai-assisted data engineering workflows through the Model Context Protocol. Instead of manually performing these actions yourself, the AI assistant invokes server tools directly during your conversation.
Run `npx mcp-server-databricks` in your terminal, or copy one of the JSON configuration snippets above into your MCP client settings file. Most clients (Claude Desktop, VS Code, Cursor) use a JSON config file — see the Configuration section for client-specific snippets.
Yes, Databricks MCP Server is free and open source under the MIT license. You can use it, modify it, and contribute to it via the GitHub repository. Note that while the server itself is free, the external services it connects to (such as APIs or cloud platforms) may have their own usage costs.
Databricks MCP Server has been tested with claude-desktop, cursor, vscode-copilot (full support). Any MCP-compatible client should work, but verify specific tool support in the client's documentation. The compatibility table above shows the latest status.
Share your experience with Databricks MCP Server and help other developers.