Guide · Agent skill
How to use the DOCX Word Document Toolkit agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The docx-official skill gives an AI agent a structured method for working with Microsoft Word (.docx) files: creating documents, filling templates, reading structure, resolving tracked changes, adding comments and checking the output. It is aimed at professionals drafting reports or contracts, teams filling the same template repeatedly, and developers building extraction or conversion pipelines. After this guide you will know what the skill does, how to install it, how to phrase requests so it triggers, and what it needs on your machine.
DOCX Word Document ToolkitXiaomiMiMo/MiMo-Code · MIT Download the skill Quick start
- 1Download the ZIP. Get the docx-official skill ZIP from this page.
- 2Unzip the skill. Unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ in a project. You can also upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Install the requirements. Install Python 3 and uv, or install python-docx and lxml with pip. Add LibreOffice if you want PDF previews.
- 4Make a natural request. Describe the Word task and name the file, for example asking to fill a template with data from a CSV.
- 5Review the output. Open the generated file, check the PDF preview and confirm no placeholder text remains.
What the skill makes the agent do
An agent skill is a folder of instructions and scripts that an agent loads when a task matches the skill's description. This one covers any task where a Word (.docx) file is produced, opened, transformed or read.
Once installed, the agent follows a defined path depending on the job: build a new document from a prompt or data, fill a template with values, make deep XML-level edits, extract text and structure, accept tracked changes, add review comments, and render a PDF preview to check the result.
- Create a new document from scratch with python-docx
- Fill or lightly modify an existing template while keeping its styles
- Explode a file into XML, edit it, and reassemble it for complex changes
- Extract plain text, structure, styles, headings, tables and comments
- Accept tracked changes and add review comments
- Render a PDF for visual checking
Install the docx-official skill
Download the ZIP of the skill from this page. Then choose one of the following locations.
For a personal install across projects, unzip the folder into ~/.claude/skills/. For a single project, unzip it into the .claude/skills/ folder inside that project. In the Claude apps you can instead upload the skill under Settings → Capabilities → Skills.
- Personal: unzip into ~/.claude/skills/
- Single project: unzip into .claude/skills/ in the project root
- Claude apps: Settings → Capabilities → Skills, then upload
Install the requirements
The scripts need Python 3. The recommended route is uv, which reads the inline metadata in each script and installs dependencies automatically. Without uv, install python-docx and lxml with pip.
LibreOffice (soffice) is optional but needed for PDF rendering and visual QA. Poppler is optional for converting PDF pages to images. pandoc is optional for some conversions. If the environment variable MIMO_PYTHON is set, the skill uses that preinstalled interpreter instead of uv or pip.
- Python 3 with uv (recommended), or python-docx and lxml installed via pip
- Optional: LibreOffice (soffice) for PDF rendering and QA
- Optional: Poppler for PDF to image conversion
- Optional: pandoc for conversions
python3 -m pip install --upgrade python-docx lxmlTrigger the skill with natural requests
You do not need a special command. The agent loads the skill when your request mentions a Word doc, DOCX, an Office document, a filename ending in .docx, or a task such as turning content into a Word report. Say what you want the file to contain and where it is.
The skill's description lists the situations it covers, so requests that name one of those tasks are the most reliable triggers.
- "Turn this into a Word report"
- "Fill contract_template.docx with the client details in clients.csv"
- "Extract the headings and tables from proposal.docx"
- "Accept all tracked changes in review.docx and give me a clean copy"
- "Convert board-notes.docx to plain text"
Run the core commands
These are the commands the skill uses. You can also run them yourself to test the install. Each script is a small self-contained Python file, and its top section lists the full CLI options.
uv run scripts/extract_text.py input.docx > input.txt
uv run scripts/explode.py input.docx exploded/
uv run scripts/assemble.py exploded/ output.docx
uv run scripts/render_pdf.py output.docx
uv run scripts/audit.py output.docx
uv run scripts/resolve_revisions.py reviewed.docx clean.docx
uv run scripts/annotate.py exploded/ "Please check" --author "Reviewer" --anchor "text"Check the result before you trust it
The skill's QA checklist asks the agent to assume something may be wrong. It checks that the file opens cleanly, that no placeholder text such as TODO, TBD or {{ remains, and that the rendered PDF has no widowed headings, awkwardly split tables or missing page numbers.
Keep the original file until the new one has passed these checks. The PDF preview comes from LibreOffice, so it can differ from how Word itself renders the page.
- Confirm the file loads without a repair prompt
- Search the extracted text for leftover placeholders
- Render a PDF and review the first and last pages
- Keep the original until the new file passes QA
Safety notes: scripts and network
The skill runs scripts. They execute locally with the same file access the agent has, so review them before processing sensitive documents. The skill does not use the network.
The skill's own guidance is to rely on named styles such as Heading 1 and Normal, and to include alt text on every image. Those rules make output easier to navigate and more accessible.
- Scripts run locally with the agent's file access
- No network use
- Review scripts before using them on confidential files
What people use it for
- Drafting a report, letter or contract as a real Word file from a prompt and data
- Filling the same Word template repeatedly with values from a spreadsheet
- Extracting the text or heading structure of a long proposal for summarising
- Accepting tracked changes to produce a clean copy of a reviewed document
- Adding review comments to a document's exploded XML for a reviewer to see
Frequently asked questions
How do I install the docx-official skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ inside a project. In the Claude apps you can upload it under Settings → Capabilities → Skills.
What do I need installed to use it?
You need Python 3 with uv recommended, or python-docx and lxml installed via pip. LibreOffice is optional for PDF rendering and QA, Poppler is optional for PDF to image conversion, and pandoc is optional for some conversions.
Is the docx-official skill free?
The skill is published under the MIT licence, as stated on this page. Check the licence file in the source repository for the full terms before redistributing it.
Does the skill need an internet connection?
No. The skill is listed as not using the network, but its scripts run locally with the file access the agent has, so review them before processing sensitive documents.
Why does the PDF preview look different from Word?
The PDF preview is rendered by LibreOffice, which can differ from Word's exact layout. If a file must match Word's rendering precisely, check it in Word itself.
What is an alternative if I only need to read a Word file?
For plain text only, the skill's extract_text.py script is the lightweight route. If you do not want an agent skill at all, python-docx can read the same file directly in a short Python script.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you regularly produce, fill or check Word files with an AI agent and can run Python and optionally LibreOffice on your machine. Skip it for one-off text answers where a plain document is unnecessary, or for documents that must match Word's exact rendering or that you cannot send through unreviewed scripts.
Download the skill