Datapackage

If you're working with datasets that ship a datapackage.json descriptor, this skill teaches Claude how to parse it without blowing up context. It uses jq and DuckDB to query metadata selectively, surfaces usage warnings buried in resource descriptions, and loads data from Parquet, SQLite, or CSV files once you know what you need. Built by Catalyst Cooperative for their PUDL energy data product but works with any Frictionless Data Package. The core insight is treating the descriptor itself as queryable data rather than documentation, which matters when you're dealing with descriptors that catalog hundreds of tables and run into the megabytes.

skills.sh installs: 2.1K
GitHub stars: 3
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add catalyst-cooperative/agent-skills --skill datapackage --agent claude-code

Installs into .claude/skills of the current project.

Files

Frictionless Data Package Guide

This skill covers any dataset described by a Frictionless Data Package descriptor file (datapackage.json). It is intentionally generic — it works for any conforming datapackage, regardless of who published it or what the data contains.

For PUDL-specific knowledge (S3 bucket paths, table tier conventions, data source context, usage warnings), also use the pudl skill on top of this one.