Agent skills for AI coding assistants
Agent skills are task-specific instruction files that AI coding assistants like Claude and GitHub Copilot can load to perform Databricks development tasks. Skills package domain-specific knowledge, best practices, and workflows into a format optimized for AI consumption. To learn how to extend Genie Code in the Databricks workspace, see Extend Genie Code with agent skills.
Skills follow the open Agent Skills standard. Each skill is a Markdown file with front-matter metadata that describes when and how the skill should be used. AI coding assistants automatically discover and load relevant skills based on the task at hand.
Install AI tools
Install Databricks AI tools using the Databricks CLI aitools command group. The CLI detects supported coding agents and installs Databricks skills and plugins so your agent can work effectively with Databricks resources such as bundles, jobs, and SQL.
# Install skills for all detected agents
databricks aitools install
# Install skills for a specific agent
databricks aitools install --agents claude-code
# Install skills into the current project instead of globally
databricks aitools install --scope project
# Install specific skills
databricks aitools install --skills bundles,sql
Use databricks aitools list, databricks aitools update, and databricks aitools uninstall to manage installed AI tools. For all options, see aitools command group.
To install skills from arbitrary GitHub repositories that are not distributed through the Databricks CLI, use the Skills CLI, an open-source package manager for agent skills. The Skills CLI scans a GitHub repository for skill files and installs them into your project so that your AI coding assistant can discover and use them automatically.
AI tools and managed MCP servers
AI tools and managed MCP servers solve different halves of the same problem, and most setups use both:
| AI tools | Managed MCP servers |
|---|---|---|
What it provides | Knowledge: Databricks patterns, APIs, and workflows the agent applies as it writes code. | Tools: callable endpoints an agent uses to query Unity Catalog data, run SQL, search indexes, and call Genie Agents. |
Where it runs | Locally, in your coding agent's skills or plugin directory. | Hosted by Databricks, governed by Unity Catalog. |
How to set it up |
| Add the server URL to your client's MCP configuration and authenticate. See Connect MCPs to AI assistants and coding agents. |
Skills teach the agent how to work; MCP servers let it do the work. A skill can direct an agent to a managed MCP server for a task, but the skill is not itself an MCP server, and installing AI tools does not configure MCP servers for you. If you want your agent to both write correct Databricks code and read live workspace data, install AI tools and connect the managed MCP servers you need.
Skill repositories
GitHub repository | Description | Skills |
|---|---|---|
Officially maintained core skills and plugins for Databricks development across compute, orchestration, storage, and apps. These skills are installed using | Among others, includes skills for Agent Bricks, AI Functions, AI/BI Dashboards, Databricks Apps, Bundles, Databricks CLI, Databricks Lakehouse, Genie, Iceberg, Lakebase, Lakeflow Jobs, metric views, MLflow evaluation, Model Serving, Python SDK, Lakeflow pipelines, serverless migration, Structured Streaming, synthetic data, Unity Catalog, AI Search, and Zerobus ingest | |
Curated community skills covering Databricks development patterns. | Deprecated. These skills are available with Databricks agent skills. To avoid conflicts, uninstall AI Dev Kit skills before installing Databricks agent skills. See Uninstall AI Dev Kit. | |
Task-specific skills embedded inside Databricks app templates for agents (LangGraph, LangChain, OpenAI Agents SDK), App Kits (Lakebase, Genie, Analytics), and chatbot/data app frameworks (Streamlit, Dash, Gradio, Shiny, Flask, Node.js). | Quickstart, deploy, modify-agent, add-tools, create-tools, discover-tools, migrate-from-model-serving, run-locally, load-testing, supervisor APIs | |
Skills for instrumenting, debugging, and evaluating LLM agents with MLflow. | MLflow onboarding, MLflow agent, instrumenting tracing, retrieving and analyzing traces, querying metrics, agent evaluation, chat session analysis, MLflow doc search |
Additional resources
aitoolscommand group: Fullaitoolscommand reference.- Databricks managed MCP servers: Give agents governed access to Unity Catalog data and tools.
- Connect MCPs to AI assistants and coding agents: Connect Claude, Cursor, and other clients to Databricks MCPs.
- Extend Genie Code with agent skills: Create skills for the built-in Genie Code in your workspace.
- Author an agent and deploy it on Databricks Apps: Build and deploy an agent on Databricks Apps.