Prompt Library

A curated collection of 130+ ready-to-use prompts for working with Databricks through Claude Code and the AI Dev Kit. Each prompt is tested, copy-paste ready, and designed to leverage the right combination of skills and MCP tools automatically.

How to Use

1. Browse a category below or use the sidebar to navigate
2. Find a prompt that matches your task
3. Copy it into Claude Code (or any agent with AI Dev Kit installed)
4. Claude activates the appropriate skills and MCP tools to complete the task

Tip

You don’t need to specify which skill or tool to use. Claude reads the prompt, matches it to the relevant skill, and calls the right MCP tools automatically.

Categories

Data Engineering SDP pipelines, Spark Streaming, Zerobus Ingest, and custom data sources. 24 prompts.

SQL & Analytics SQL execution, AI Functions, dashboards, Genie Spaces, and Metric Views. 26 prompts.

AI & ML Model Serving, Vector Search, MLflow Evaluation, Agent Bricks, and synthetic data. 25 prompts.

Apps & Databases Python apps, AppKit, Lakebase Autoscale, and Lakebase Provisioned. 21 prompts.

Governance & Catalog Unity Catalog, permissions, tags, volumes, Iceberg, Delta Sharing, and monitoring. 21 prompts.

DevOps & Config Workspace auth, Asset Bundles, Jobs, Python SDK, and compute management. 23 prompts.

Cross-Cutting Scenarios

These prompts span multiple skills and tools for end-to-end workflows. Each one triggers a multi-step build across different parts of the Databricks platform.

End-to-End Data Platform

Build a complete data platform:
1. Create a medallion pipeline (bronze/silver/gold) using Spark Declarative Pipelines
2. Schedule it with a Databricks Job that runs every hour
3. Create an AI/BI dashboard that visualizes the gold layer
4. Package everything as a Databricks Asset Bundle for deployment

Skills involved: databricks-spark-declarative-pipelines, databricks-jobs, databricks-aibi-dashboards, databricks-bundles

RAG Application

Build an end-to-end RAG application:
1. Parse PDF documents from a volume using ai_parse_document
2. Chunk the text and create embeddings with a Vector Search index
3. Build a Knowledge Assistant that answers questions using the index
4. Deploy it as a Streamlit app on Databricks Apps

Skills involved: databricks-ai-functions, databricks-vector-search, databricks-agent-bricks, databricks-app-python

Real-Time Analytics

Build a real-time analytics system:
1. Ingest events using Zerobus Ingest
2. Process them with a Spark Structured Streaming pipeline
3. Write aggregated results to a Lakebase database for low-latency serving
4. Create a Genie Space for business users to ask questions about the data

Skills involved: databricks-zerobus-ingest, databricks-spark-structured-streaming, databricks-lakebase-autoscale, databricks-genie

ML Ops Pipeline

Set up an ML pipeline:
1. Generate synthetic training data
2. Train a model and log it with MLflow
3. Evaluate the model using MLflow GenAI evaluation with custom scorers
4. Deploy the best model to a serving endpoint
5. Create a monitoring dashboard that tracks prediction quality

Skills involved: databricks-synthetic-data-gen, databricks-mlflow-evaluation, databricks-model-serving, databricks-aibi-dashboards

Multi-Agent System

Build a multi-agent customer service system:
1. Create a Genie Space for SQL-based product and order queries
2. Create a Knowledge Assistant for policy and FAQ documents
3. Build a Supervisor Agent that routes customer questions to the right agent
4. Deploy the supervisor as a serving endpoint
5. Build a Gradio chat app as the customer-facing interface

Skills involved: databricks-genie, databricks-agent-bricks, databricks-model-serving, databricks-app-python