add prompting cookbook

[mirnawong1]

this pr adds a prompting cookbook to provide users with effective ptrompting best practices and examples

cookbook has independent, self-contained recipes.

to does before merge:

• [ ] i still need to add links to the cookbook for better discoverability and add related docs section.
• [ ] sme review
• [ ] docs review review

notion project

---

🚀 Deployment available! Here are the direct links to the updated files:

• https://docs-getdbt-com-git-prompt-cookbook-dbt-labs.vercel.app/guides/prompt-cookbook

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Review	Updated (UTC)
docs-getdbt-com	Ready	Preview	Jan 9, 2026 5:50pm

[trevenm]

For the Semantic Model guidance, it is a little tricky.

• We have a dedicated "easy" button that will generate this for us, but the button doesn't allow us to give direction like you've illustrated here.
• In order to interact with Copilot, you need to have a SQL window open.
• If you open a new file, Copilot isn't available because it doesn't know what file type you are operating in (you'll only get the Style Guide option).
• If you open a yml file it won't give you the conversational window because yml files aren't supported yet.

You can follow the guidance in the docs here, but what it would require you to do is:

1. Open a current sql file (or make a new file and save it)
2. Put the prompt in the conversational window
3. Create a new yml file (Copilot conversation will not be displayed when the yml file is your active window)
4. Copy content from Copilot conversation (when sql file is the active window)
5. Paste content into yml file

For this reason, the "easy" button version of generating a Semantic Model is much more intuitive. In our workshop, we created a model from scratch using the SQL copilot. Then demonstrated how we added docs, tests, and a semantic model to what we just built. This was a smooth transition between conversational Copilot usage for SQL and the built-in methods for docs, testing, and semantics. Maybe a page that highlights those pre-built options?