Docx

Skill yigityildiz0/universal-ai-skill-library/skills/common/docx

531 searchable AI Agent Skills for Claude Code, OpenAI Codex, and OpenCode — EN/TR catalog, platform and risk notes, direct ZIPs, and curated bundles.

Install

npx -y skills add yigityildiz0/universal-ai-skill-library --skill docx

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

3 things to look at

19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Read, create, edit, and validate Word DOCX documents while preserving structure, styles, relationships, accessibility, and existing user content. Use when DOCX is the primary input or required deliverable; prefer installed document tooling and make dependency changes only when authorized.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.8 KB, 482 tokens by cl100k_base, as published. Nobody here has run it

DOCX Workflow

Choose the path

Read/review: extract paragraphs, headings, tables, headers/footers, notes/comments, links, and media inventory without changing the source.
Create: use an installed DOCX library or the active document tool and define styles before content.
Edit: preserve the existing package, styles, numbering, sections, relationships, tracked changes, comments, fields, and embedded objects unless the task explicitly changes them.
Convert/render: use an installed office renderer in a separate output location and visually inspect the result.

Do not install libraries or office software implicitly. If the needed tooling is absent, state the exact proposed dependency and wait for authorization unless installation is already in scope.

Editing rules

Back up the original or write to a new path.
Inventory sections, page setup, styles, numbering, tables, fields, links, images, headers/footers, and accessibility metadata.
Make the narrowest semantic edit. Avoid raw ZIP/XML edits unless the library cannot preserve the feature and the OOXML relationship changes are fully understood.
Preserve macros and signatures as opaque artifacts; never execute them. Warn that editing a signed document may invalidate its signature.
Keep headings hierarchical, tables readable, links meaningful, images described, and language metadata correct.
Use real list/heading/table structures rather than visual approximations with spaces or symbols.
Do not invent citations, legal clauses, values, or document metadata.

Creation quality

Define page size/margins, body and heading styles, paragraph spacing, list numbering, table rules, captions, headers/footers, and section breaks consistently. Keep the design restrained unless the user supplies a brand system. Provide a table of contents only when headings and the target viewer support updating fields.

Validation

reopen the produced DOCX with an independent parser;
verify paragraph/table counts and required text;
inspect relationships for missing media or links;
render to PDF/images when layout matters and inspect every page for overflow, blank pages, clipping, bad page breaks, and font substitution;
compare preserved features when editing an existing file;
scan the output package for unexpected files or active content.

Report output path, source preserved, tools used, pages/sections checked, known viewer differences, and any feature that could not be preserved.