Pudl

If you're working with PUDL energy data outputs and need to understand what's actually in those Parquet tables, this gets you oriented fast. It surfaces table schemas, column definitions, and source metadata directly in Jupyter or Marimo notebooks without making you grep through documentation or ETL code. The skill treats its metadata references as canonical, so you get consistent answers about field semantics and methodology whether you're analyzing data or debugging pipelines. It's scoped specifically for data users and analysts, not for people modifying the ETL itself. Honest take: this is the kind of contextual lookup layer that saves you from constantly asking "wait, what does this column actually measure?"

Install to Claude Code

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