The Atlan MCP Cookbook

Know what your domain owns, and whether it's live

Ask what data products exist under a business domain and what actually backs one of them, instead of hunting table by table.

All recipes

On the menu

Use case
Find & understand data
Action
Look up
Persona
Data/Business User, Data/BI Analyst, Data Governance Lead
Effort
Moderate
Runs in
any AI tool

The task

Scenario. In Monday standup, the Finance domain lead gets asked how many data products exist under the Opportunity Pipeline product line, and which tables actually feed it. The honest answer today involves opening three different schema browsers and hoping nothing's missing. There's no single place to ask "what do we own, and what's actually behind it."

The ask

  • The data products registered under the domain
  • The tables and columns that actually back a given product
  • Whether those backing assets are still being fed, or have quietly gone stale

Hero prompt

Hero prompt

List the data products registered under the Finance domain, then show me exactly which tables and columns back the Opportunity Pipeline product, and whether that product is actively fed by anything upstream.

Step by step

  1. Find the domain.

    Step 1

    Confirm the <domain — e.g. Finance> domain exists and show its details.

  2. List its data products.

    Step 2

    List every data product registered under the <domain> domain, so I have the full inventory in front of me.

  3. Pull what backs one product.

    Step 3

    For the <product — e.g. Opportunity Pipeline> product, show the underlying tables and columns that back it.

  4. Check it's actively fed.

    Step 4

    Show whether <product>'s assets have active upstream lineage, so I know if it's genuinely fed or quietly stale.

What you'll give it

Required context

  • The domain name you're asking about
  • The specific product name, if you're going deeper than the inventory list

Optional context

  • An ownership filter, if you only want products a specific team owns
  • A freshness threshold, if staleness is the actual concern

What Atlan creates

Outputs

  • A list of data products registered under the domain
  • The concrete tables and columns backing the one product you drilled into
  • A lineage read confirming whether that product is actively fed or stale

Summary. You walk into standup with the actual inventory and a real answer on freshness, not "I think we have a few things under Finance."

Read-only

Read-only. Nothing is written; this only surfaces what's already registered.

Follow-up prompts

Keep going once the first answer lands.

Roll it up

Roll it up

Roll this up across every domain for a full data mesh coverage report.

Find the gaps

Find the gaps

Which products in this domain have no owner assigned?

Make it recurring

Make it recurring

Re-run this monthly as a standing input to the domain review.

Tips & troubleshooting

Cook's note

The domain list answers "what do we own." The lineage check answers "is it actually alive." Don't stop at the first one.

  • Match the client to the work.** This reads well in a chat client threaded into a recurring standup channel, since the answer doubles as a status update for the room.
  • Automate or chain it.** Schedule this as a recurring pull ahead of your domain review cadence, instead of someone manually compiling it the morning of.
  • Recheck before you report it.** Data products get added and deprecated between reviews. Re-pull the inventory right before you present it, not from last month's notes.
THE ATLAN MCP COOKBOOK

Point your AI at the Atlan MCP.
Run your first recipe.

[Website env: production]