Course overview/Governance as agent context1 of 5

Descriptions and tags

Descriptions and tags

Write context that records a data caveat an agent cannot infer from a column name.

A description is agent input

customer_id: the customer id restates the name. A useful description records a unit, population, caveat, canonical alternative, or source-system constraint. In this project, net_price is the line-revenue input while unit_price is the list price; order_total is an order-header figure that does not reconcile to line revenue.

Tags select work and mark layers. domains and owner set accountability. Column metadata carries structured facts. bruin ai enhance can draft descriptions, but each claim needs a query-based check before it becomes context another agent trusts.

Your task

For every column in weekly_category_revenue.sql, run a query that supports the description you plan to write. Add a non-redundant description for each column and preserve layer:mart in the asset tags. Validate the asset.

Check your understanding

  • What makes customer_id: the customer id a useless description, precisely?
  • Why is orders.order_total a case where the description has to carry a warning rather than a definition?
  • A description says order_status needs UPPER(TRIM(...)) before filtering. Why is that description still wrong?
  • Why verify bruin ai enhance output against the data instead of publishing it directly?

Do it with your agent

Say next lesson, write only descriptions you can support, then say review my work. The review should reject restated names and unverified claims.

Sign up to our newsletter

Practical updates on open-source data pipelines, AI analysts, governance, and what we are shipping at Bruin.

The signup form is hosted by Brevo. Allow marketing cookies to load it.