Your AI Editor's Field Guide to Databricks

What it does

The AI Dev Kit is essentially institutional knowledge exported as code. Curated by Databricks field engineers, it bundles a Python core library, an MCP server, markdown skills, and a visual builder web app. The goal is to let external AI coding assistants—Cursor, Claude Code, Gemini, Copilot, and others—actually interact with Databricks resources like Unity Catalog, jobs, and SQL warehouses instead of guessing syntax.

The interesting bit

The kit is deliberately modular: you can ingest just the skills, just the MCP tools, the core library, or the whole stack. A neat twist is the Visual Builder App, which deploys as a chat UI but can also double as an MCP gateway at /mcp, exposing the tool suite to Genie Code and other MCP clients from a single deployment.

Key highlights

• Curated by Databricks field engineering, not auto-generated documentation.
• Ships as four composable pieces: a Python core, an MCP server, markdown skills, and a full-stack builder app.
• The MCP server exposes over 50 executable tools, letting agents manipulate Databricks resources directly.
• Supports a wide roster of AI editors: Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, Windsurf, and others.
• The core library plugs directly into Python frameworks like LangChain and the OpenAI Agents SDK.

Caveats

• Cursor and Copilot still require manual settings tweaks after installation; the scripts don’t fully automate everything.
• By default the installer scopes itself to the current project directory, so your AI client must be launched from that exact path.

Verdict

Grab this if your team lives in Cursor or Claude Code but needs to ship Databricks pipelines, jobs, or apps. If you never leave the Databricks workspace, the built-in Genie Code is probably the simpler (and free) path.