Update Data Dictionary

Update Data Dictionary

Keep an existing data dictionary in sync with the current dataset.

When to invoke

• A schema-changing operation just ran and the dataset's folder contains a data_dictionary.{md,yaml,json,csv}.
• User says "update the data dictionary", "re-document", "re-profile".
• Another wrangler skill needs to log a transformation.

Procedure

1. Find the dictionary — look for data_dictionary.* in the dataset's folder. If none exists, suggest add-data-dictionary instead.
2. Load the current dataset and profile columns.
3. Diff against the existing dictionary:

• New columns — present in data, missing from dictionary → add row with stub description, ask user to fill in.
• Removed columns — present in dictionary, missing from data → mark as "REMOVED " or offer to delete the row (ask user).
• Type changes — update the type column; log in provenance.
• Renamed columns — heuristic only; ask the user to confirm rename mappings rather than guessing.

1. Append to the provenance / transformations section with today's date and a one-line summary of what changed.
2. Preserve user-written descriptions — never overwrite a description the user has filled in. Only touch auto-generated fields (type, examples, nullability).
3. Write the updated dictionary back to the same file, preserving its format (md/yaml/json/csv).
4. Report a summary diff: N columns added, M removed, K types changed.

Dependencies

pip install pandas pyarrow openpyxl pyyaml

Edge cases

• Markdown round-trip — parse the existing columns table carefully; preserve any additional text sections (notes, known issues) that the user has added by hand.
• Multiple dictionary files — if both data_dictionary.md and data_dictionary.yaml exist, ask the user which is canonical; optionally update both.
• Dictionary out of date with multiple intermediate changes — the provenance log may have gaps. Log today's transformation; flag the gap but don't fabricate history.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: danielrosehill
• Source: danielrosehill/Claude-Data-Wrangler-plugin
• License: MIT

Install and usage instructions live in the source repository linked above.