A Model Context Protocol server that lets LLMs query Databricks Genie spaces, run SQL, and converse with Genie agents.
https://github.com/yashshingvi/databricks-genie-MCPStop wrestling with SQL syntax when you need quick insights from your Databricks lakehouse. This MCP server connects your favorite LLM directly to Databricks Genie, turning natural language questions into instant data answers.
You know the drill: you need to check customer metrics, but first you have to remember table schemas, join patterns, and figure out which Genie space has the right data. Then you craft the perfect SQL query, wait for results, and realize you need to ask three follow-up questions. Each iteration means more context switching and lost momentum.
Direct LLM-to-Databricks Pipeline: Your LLM can now start Genie conversations, ask follow-ups, and retrieve structured results without you ever leaving your chat interface. No more tab-switching between Claude and Databricks workspaces.
Conversational Data Exploration: Ask "What were our top-performing campaigns last quarter?" and immediately follow up with "Break that down by region" or "Show me the trend over time." The MCP server maintains conversation context across your entire data exploration session.
Structured Results in Your Workflow: Get SQL queries and result tables back in clean, structured formats that your LLM can reason about, summarize, or transform for presentations.
Executive Briefings: Ask your LLM to "pull our monthly active users for the executive summary" and get instant charts and insights without disturbing your data team.
Ad-hoc Analysis: During product reviews, quickly validate hypotheses like "did the checkout redesign improve conversion rates?" with immediate data validation.
Data Discovery: New team members can explore your data landscape by asking "what customer segments do we track?" and get immediate context about available datasets.
Automated Reporting: Build LLM-powered workflows that pull specific metrics on schedule and format them for stakeholder consumption.
# After setup, your LLM can immediately:
# 1. List available Genie spaces
# 2. Start conversations with natural language
# 3. Follow up on results
# 4. Get structured SQL and data back
The server handles all the Databricks API complexity, authentication, and conversation management. Your LLM gets clean, structured responses it can work with directly.
Data teams are bottlenecked by SQL-writing overhead, and business users can't self-serve insights without technical help. This MCP server removes that friction entirely - your LLM becomes a universal data interface that speaks both natural language and SQL.
The manual space ID configuration is a minor setup step that pays dividends immediately. Once configured, you have a permanent bridge between conversational AI and your enterprise data that works across any MCP-compatible LLM.
Perfect for data teams who want to democratize data access, product managers who need instant metric validation, and executives who want data-driven insights without technical dependencies.