End-to-end retail analytics pipeline that turns raw, messy CSV exports into business decisions, built in Python and SQL and runnable with one command.

What it does:

Cleans and validates 6 relational tables, fixes 585 data-quality issues and logs every action

Loads a SQLite warehouse with keys, constraints and window-function SQL

Customer analytics: RFM segmentation, cohort retention, CAC / LTV by channel, product Pareto, market-basket analysis

Churn prediction model validated on future data (ROC-AUC 0.71, no data leakage)

6-month revenue forecast, best of 3 models by backtest (4.2% MAPE)

Auto-generated Excel report (live formulas, native charts) and an interactive Streamlit dashboard

18 automated tests

Logging every cleaning action is the part I like most: when a number in the report looks strange, you can follow it back to the fix that changed it. With 585 issues, how did you decide which ones the script fixes alone and which ones it only reports?

The fix for an agent that keeps getting your design system wrong isn't a better prompt.

It's putting the rules where the agent reads them.

In the repo. In skills. In checks it can run itself.

Writing a course about exactly this.

What sells it for me is the refusal telling the agent exactly where the token should go. A bare "no hex" error usually just sends it looping. Does guard.mjs run only as an agent hook, or in pre-commit too so it catches humans as well?