Loading...
Loading...
Found 4 Skills
Access PUDL table data plus table/column/source metadata in Jupyter or Marimo notebooks for debugging and visualization. Use when users ask what a table contains, how to read it, or how columns are defined.
Work in an existing vignette catalog: set it up, run or inspect its marimo notebooks, and answer data questions by reusing their @app.function helpers in a live kernel. Use whenever a repository has catalog.toml and the user asks to get started, run notebooks, explore the data, make an analysis or figure, or compose a new notebook. Do not use for generic notebooks or for creating a new catalog.
Manage Jupyter notebooks — create, execute cells, manage kernels via the container's Jupyter Server REST API.
Disposable builder skill (delete after the harness is built): scaffolds a Colab-centric project — a repository whose deliverable is Google Colab notebooks, mirrored locally and validated in an official Colab-runtime devcontainer before running on real Colab. Use when creating or hardening a repository that produces notebooks people open and run in Colab — tutorials, course materials, runnable demos, shareable analyses — or when a project uses Colab as its runtime, including driving Colab VMs as disposable compute for local scripts. Not for generic Jupyter-notebook work, projects where notebooks are incidental exploration beside a library or pipeline, or mature migrations; preserves working choices.