Databricks AI Dev Kit

Databricks Field Engineering's agent kit: an installer for Databricks skills, an MCP server and a visual Builder App

PluginHigh risk

databricks-solutions/ai-dev-kit

Install

bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)

Installs into the current project with the DEFAULT profile. --global installs for the user.

This is third-party code. Review the repository files before installing.

What it does

The installer adds skills to a project or user profile for agent work with Databricks: Spark Declarative Pipelines, Jobs, AI/BI dashboards, Unity Catalog, Genie, MLflow, Model Serving and Databricks Apps. Skills are now shipped officially through the Databricks CLI, and the installer hands off to it while keeping the guided flow. The repository still holds a standalone MCP server with Databricks actions and the Builder App web UI. It supports Claude Code, Cursor, Codex, Gemini CLI, Copilot, Windsurf, OpenCode and other agents.

Who it is for. For data engineers and analysts building pipelines, dashboards and ML on Databricks with agents.

Good fit when

  • You want the agent to write pipelines and jobs following Databricks practices
  • You need to set up several agents for one workspace quickly
  • You need your own MCP server for Databricks actions

Not a fit when

  • You work only inside the workspace: built-in Genie Code is already there
  • You have no Databricks account

Example request

Create a Databricks pipeline with Auto Loader and SCD Type 2 for the customers table and schedule it as a job

Limitations

Requires a Databricks account and Databricks CLI 1.0 or newer. The project is run by Field Engineering, the MCP server is maintained on a best-effort basis, and the skill files in the repository itself are deprecated and kept for reference. Databricks does not operate in Russia, so it cannot be used without a VPN and a foreign account.

How to disable. Remove installed skills from the agent folders in the project, such as .claude or .cursor, and the databricks server from the MCP config. Re-running the installer removes older skills and the MCP server before installing the official AI Tools.

Security check

  • The installer runs a remote script via curl and bash
  • The agent gets access to data and resources in a cloud workspace

README in short

The README opens with an announcement that skills moved to official Databricks AI Tools and a rename table. It then compares built-in Genie Code with AI Dev Kit, lists supported use cases and prerequisites. Installation is a script for Mac and Linux with global and skill selection options; the MCP server and Builder App are covered separately.

FAQ

How does this differ from databricks aitools install?

Official skills are now installed by the Databricks CLI. AI Dev Kit calls it, guides setup and hosts Field Engineering's experimental tools.

Which skills were renamed?

For example, databricks-bundles became databricks-dabs and databricks-genie became databricks-genie-agents. The full table is in the README.

Editors’ pick

166 skills for scientific work: bioinformatics, cheminformatics, clinical data, geospatial analysis and 100+ databases

SkillMedium risk44.9K
Editors’ pick

Google's open-source MCP server for databases: ready tools for Postgres, MySQL, BigQuery, Spanner and more, plus custom tools in tools.yaml

MCP serverHigh risk16.4K
Editors’ pick

Official Hugging Face skills: Hub operations via the hf CLI, datasets, model training, Spaces, evals and deployment

SkillHigh risk11K
Editors’ pick

A token-lean MCP server for PostgreSQL, MySQL, MariaDB, SQL Server and SQLite with a read-only mode

MCP serverHigh risk3.5K