messydata

Run `npx skills add sodadata/messydata --all -g -y` in your terminal to install this skill globally.

Define your columns and distributions in a YAML config, then run the generate command to get a messy dataset with missing values, duplicates, and outliers injected automatically.

It supports missing values, duplicate rows, invalid categories, invalid dates, and statistical outliers, each with configurable probability and rate.

Yes. The included SKILL.md teaches your agent to write and validate configs from plain-English requests, then run the CLI to produce the data file.

Yes. Mark one field as temporal and use single-date or date-range modes to generate realistic day-by-day data, ideal for simulating daily pipeline runs.