Use Lakehouse Plumber with your agent

Lakehouse Plumber (LHP) is an open-source, YAML-driven ETL framework for Databricks Lakeflow Spark Declarative Pipelines (SDP). Describe load, transform, write, and test actions in YAML; LHP generates readable Python for your pipelines.

Use the guides below with Databricks Genie Code, Claude, Codex, Gemini, Copilot, or another agent that can read public web pages. No account or API key is needed to read them.

Set up a project

Give your agent the setup guide. It covers installing LHP in an isolated Python environment, choosing a sample or empty project, and installing the LHP authoring skill with lhp skill install.

Read https://lakehouse-plumber.dev/agent.md and help me set up Lakehouse Plumber. Ask whether I want a sample or an empty project, and confirm the project name and folder.

Prefer a terminal? Install the Python package with pip install lakehouse-plumber in a virtual environment, then follow the Get Started course. LHP requires Python 3.11 or later.

Learn with guided examples

The learning guide uses the current published documentation. Your agent will ask about your SQL, Python, and Databricks experience, then explain a small example one step at a time. A dedicated teaching skill is not available yet.

Read https://lakehouse-plumber.dev/learn.md and help me learn Lakehouse Plumber. Ask about my experience and what I want to build. Explain the YAML and generated Python using the current documentation.

Read the source of truth

• The action model: how actions connect through named views.
• Load actions: sources and ingestion.
• Transform actions: SQL, Python, and data quality.
• Write actions: pipeline destinations.
• Data tests: tests alongside the flow.
• Guides: reusable patterns and workflows.
• Source code and releases: implementation, examples, and changes.

How LHP fits into Lakeflow

LHP provides a metadata-driven authoring layer: reusable YAML configuration becomes native pipeline code that you can read and review. Databricks runs the generated pipelines. Lakeflow Spark Declarative Pipelines was previously called Delta Live Tables (DLT); current LHP documentation uses the newer terminology.

Agents should read the relevant reference before proposing configuration, verify commands in the user's environment, and confirm deployment intent before deploying. The setup guide includes extra checks for the execution tools available in a Databricks workspace.

Machine-readable entry points

• llms.txt: a compact index of verified website and documentation links.
• Homepage as Markdown: the product overview in plain text.
• agent.md and learn.md: task-specific instructions.