How to use the Data Quality Frameworks agent skill

What the skill makes the agent do

Once installed, the agent knows how to set up data quality checks for your pipelines. It can suggest and write Great Expectations expectation suites, which are rules such as 'this column is never empty' or 'this ID is unique'. It can also write dbt test suites and data contracts that define what a dataset should look like between the teams that produce and consume it.

The skill also covers wiring these checks into CI/CD so that a pipeline fails when the data is bad. The source describes the testing approach as a pyramid running from schema tests, through single-column unit tests, up to cross-table integration tests.

• Great Expectations expectation suites (for example not-null and unique checks on an order_id column)
• dbt test suites
• Data contracts describing expected dataset structure
• Checks wired into CI/CD that fail the pipeline on bad data

Install the skill from the ZIP

Download the ZIP file from this page and unzip it. The skill folder contains SKILL.md and a references/details.md file. Place the folder in one of two locations: ~/.claude/skills/ for your personal use across projects, or a project's .claude/skills/ folder if you want it available only inside that project.

You can also install it through the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the ZIP there.

• Personal install: ~/.claude/skills/
• Project install: the .claude/skills/ folder inside your project
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Trigger the skill with natural requests

You do not need a special command. The skill is meant to activate when your request matches its description, which covers building data quality pipelines, implementing validation rules, or establishing data contracts. Describe the task in plain language and name the tool if you have one in mind.

Good requests say what the data is, what could go wrong and where the checks should run.

• Add Great Expectations checks to our orders table so order_id is never null and always unique.
• Write dbt tests for the customers model and explain what each one catches.
• Draft a data contract for the events feed that our analytics team consumes.
• Show me how to fail our nightly pipeline in CI when the products data is bad.

Set up a Great Expectations check

The skill's quick start uses the Great Expectations package installed with pip. It creates a project, defines a datasource, adds expectations to a suite and runs a checkpoint. The example suite named orders_suite checks that order_id is not null and is unique.

Treat the generated code as a starting point. Check the column names and the datasource against your own warehouse before running it.

• Install the package with pip install great_expectations
• Initialise a project and create a datasource
• Add expectations to a suite, then run a checkpoint

Get more worked examples

The main SKILL.md gives the core patterns. A second file, references/details.md, holds longer worked examples. For instance, it includes a Great Expectations suite builder for an orders table that checks the expected column set, with exact matching turned off so extra columns are allowed, and a not-null check on the primary key.

Ask the agent to read the details file when the short quick start does not cover your case.

Safety notes: scripts and network

This skill runs scripts, so review any code the agent writes before you run it against production data. The skill does not use the network according to its listing, but the code it generates connects to your data sources, so the access you give it matters.

Use read-only credentials or a test environment when you first try validation code.

• Runs scripts: yes. Read generated code before executing it.
• Uses network: no, according to the listing.
• Needs access to the data sources you want validated.