Most attempts to connect an AI assistant to a data platform stop at read access—let the model run a SQL query and call it done. This kit takes the opposite bet: the hard part isn't reading Databricks, it's knowing the dozens of conventions—Unity Catalog layout, job orchestration, Spark idioms—that separate a working pipeline from a broken one, and encoding that knowledge so an agent can actually act on it.
What Sets It Apart
- Four composable layers instead of one monolith: a core Python library, an MCP server exposing 50+ tools, roughly 20 markdown skill guides, and a full builder web app—pull in only the pieces your setup needs.
- The skills ship the "how," not just the "what": the markdown guides teach the agent platform patterns, so generated code follows Databricks conventions rather than generic Spark snippets.
- Editor-agnostic by design—the same MCP layer drives Claude Code, Cursor, and other clients, so the tooling isn't tied to one assistant.
Who It's For
Great fit if you already build on Databricks and want your AI assistant to emit code that respects Unity Catalog, jobs, and Spark conventions out of the box, from the editor you already use. Look elsewhere if you aren't on Databricks, or if you expect a turnkey no-code product—this is developer plumbing (a Gold-tier solutions project), and you'll be wiring up MCP servers and reading skill guides, not clicking through a polished UI.
