02-setup.md

2. Setup

You have two options. Codespaces is easiest — nothing to install.

Option A: GitHub Codespaces (recommended)

1. On the repo page, click Code > Codespaces > Create codespace.
2. Wait for it to build. Dependencies install automatically.
3. You get a full VS Code in your browser. Done.

Option B: Local

You need Python 3.11+.

pip install -r llm_query/requirements.txt

Build the database

dbt reads its profile from the dbt_project folder, so run commands from there:

cd dbt_project
dbt seed --profiles-dir .

dbt seed loads the four CSV files into a local DuckDB database (analytics.duckdb). DuckDB runs inside the process — no server, no password.

You should see four seeds load, the largest being 23,296 observation rows.

Next: 03-staging-models.md