read-docx-review

Extract comments and tracked changes from reviewed docx files into reconciliation tables.

1.4k|216|Updated Oct 22, 2025
One-click install
npx skills add https://github.com/daymade/claude-code-skills --skill read-docx-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: read-docx-review
Source: https://github.com/daymade/claude-code-skills/tree/main/daymade-docs/read-docx-review
Command: npx skills add https://github.com/daymade/claude-code-skills --skill read-docx-review

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires dotnet-script, DocumentFormat.OpenXml, and includes scripts (resource) and references (resource) components.

What problem does it solve?

When a contract, manuscript, or report comes back from a reviewer with Word/WPS comments and tracked changes, reading it with naive tools like python-docx silently drops revision-inserted text (w:ins) and misreads paragraph-level deletions (w:del). This Skill reads the reviewed docx through the official OpenXML SDK and produces a per-item reconciliation table so nothing the reviewer wrote is missed.

Core Features & Use Cases

  • Comment Extraction: Lists every comment with author, timestamp, anchored paragraph, reply-thread structure, and resolved/unresolved status, rendered as a Markdown table with an empty decision column for item-by-item adjudication.
  • Revision-Aware Track Changes Parsing: Reports per-paragraph insertions and deletions, including whole-paragraph insert/delete marks, using dotnet-script csx scripts built on DocumentFormat.OpenXml.
  • Machine-Readable Output: A --json flag emits the full untruncated dataset, and --include-resolved surfaces already-handled comments.
  • Use Case: A publishing editor receives a WPS-exported reviewed manuscript with 96 comments; one command produces a reconciliation table showing who commented what and where, ready for line-by-line decisions.

Quick Start

Ask the AI to extract all comments and tracked changes from your reviewed docx file into a reconciliation table by running scripts/extract_review.py against the file.

Frequently Asked Questions about read-docx-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract comments from a docx file?

Run scripts/extract_review.py with the docx path to get a Markdown table of all comments with author, date, anchored paragraph, and resolved status. Use the --json flag for full machine-readable output including reply threads.

How to read tracked changes from Word documents in Python?

This Skill routes parsing through dotnet-script csx scripts built on the OpenXML SDK, which treats insertions and deletions as first-class elements. The Python bridge module exposes list_revisions and extract_views functions that return per-paragraph revision details.

Why does python-docx miss tracked changes content?

python-docx reads paragraph text via runs that only take direct child w:r elements and do not recurse into w:ins wrappers, so all revision-inserted text is silently dropped. Bare lxml XPath reads insertions but cannot recognize whole paragraphs wrapped in w:del deletion marks.

Does it work with WPS cloud document exports?

Yes, WPS cloud documents downloaded as docx retain comments and resolved status, and the extractor handles WPS files where some comments only have reference points without ranges by falling back to single-paragraph anchoring.

What dependencies are required to run the docx extraction?

You need any .NET SDK plus the dotnet-script global tool, installed via dotnet tool install -g dotnet-script. The Python side uses only the standard library, and the DocumentFormat.OpenXml NuGet package restores automatically on first run.

Can this tool accept or reject revisions in the docx?

No, the Skill is strictly read-only and does not accept or reject revisions, reply to comments, or modify the reviewed file. Apply changes in your own source document and regenerate outputs through your drafting workflow.