docx
Edit, revise, and redline Word documents with tracked changes, comments, and validation.
Skill by peterchang77
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/peterchang77-docx
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: documents, writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/docx
What's inside
- Skill name
docx- Description
- Edit, revise, and redline Word documents with tracked changes, comments, and validation.
- License
- MIT
- Instructions
- 70 lines in SKILL.md · view on GitHub
Files (4)
- SKILL.md
- comments.md · references/
- xml-and-validation.md · references/
- skill.json
SKILL.md (first lines)
# Word documents Use the `docx` CLI. Verify the command before editing: ```bash command -v docx docx --help ``` ## Installation If `docx` is unavailable, install the immutable `v0.1.1` release. This requires `uv`, network access, and Python 3.12 or later: ```bash uv tool install 'git+https://github.com/peterchang77/agent-docx@v0.1.1' command -v docx docx --help ``` To replace an existing installation with that release, add `--force`. Do not rely on a checkout at a machine-specific path. `docx accept-changes` additionally requires LibreOffice (`soffice`) on `PATH`. Use `Peter` as the tracked-change and comment author unless the user requests another name. Do not use em dashes in inserted or edited text. Write comments as direct statements or questions without labels such as `TODO:` or `Note:`. ## Workflow 1. Preserve the original and unpack into a new, empty working directory: ```bash docx unpack document.docx unpacked/ ``` Read `unpacked/word/document.xml` directly. It is the ground truth for content, formatting, fields, and tracked changes. Unpacking defaults to merging runs and simplifying adjacent redlines. Use `--merge-runs false --simplify-redlines false` when original run boundaries or redline boundaries matter. 2. Edit with `docx edit` whenever possible: ```bash docx edit unpacked/ --find "old text" --replace "new text" docx edit unpacked/ --find "text" --mode delete docx edit unpacked/ --find "anchor" --replace " added" --mode insert ``` Use `--paragraph N` for ambiguous matches, and `--author NAME` or `--date ISO_TIMESTAMP` when needed. Use `--allow-field-edit` only when deliberately editing a Word field. For comments, follow [the comment workflow](references/comments.md). 3. For structural changes, edit `unpacked/word/document.xml` directly. Replace complete `<w:r>...</w:r>` blocks, preserve each run's `<w:rPr>`, make matches unique, and recheck the surrounding XML after every edit. Preserve field markers (`w:fldChar`, `w:fldData`, and `w:instrText`) exactly. Use XML entities for quotes and apostrophes. See [XML and validation details](references/xml-and-validation.md). 4. Validate before packing: ```bash docx pack unpacked/ --check --diff --original document.docx docx validate unpacked/ --original document.docx docx find-artifacts unpacked/ --verbose ``` `validate` without `--original` performs schema validation but skips redlining comparison. Review artifacts before using `find-artifacts --fix`, then validate again. `pack --check` is the dry-run validation path; `pack --diff` checks untracked text and field structure. 5. Pack the result: ```bash docx pack unpacked/ output.docx --original document.docx ``` Keep `--validate` enabled unless explicitly appropriate. Accept all tracked changes only when requested and after review: ```bash docx accept-changes tracked.docx clean.docx ``` Validate the final output when practical. Run `docx <subcommand> --help` for options not shown here.
From peterchang77/agent-ai-skills, MIT.
More from peterchang77/agent-ai-skills
- 0.529score
data-science-refactor
peterchang77
Refactor research or data-science code into a reproducible, testable Python project using uv, a src layout, declarative configuration, and standalone…
SkillBFreshness: BClaude Code +9 - 0.529score
please-help
peterchang77
Turn a vague request into a concise, safe execution brief through a short plain-language conversation.
SkillBFreshness: BClaude Code +9 - 0.526score
data-science-project-bootstrap
peterchang77
Create a new, reproducible data-science repository from scratch using uv, a src layout, explicit scientific contracts, and an intentionally chosen flat or…
SkillBFreshness: BClaude Code +9 - 0.526score
single-page-web
peterchang77
Create a reviewable standalone HTML page for small dashboards, data views, analysis, annotation, or project summaries, and choose a fuller web platform only…
SkillBFreshness: BClaude Code +9 - 0.523score
skill-writing
peterchang77
Design, write, and review concise, reusable skills for AI agents.
SkillBFreshness: BClaude Code +9 - 0.521score
large-corpus-llm-extraction
peterchang77
Plan and run a controlled large-scale model-assisted extraction or review workflow over free text.
SkillBFreshness: BClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 3, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/peterchang77/agent-ai-skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from peterchang77
- 0.529score
data-science-refactor
peterchang77
Refactor research or data-science code into a reproducible, testable Python project using uv, a src layout, declarative configuration, and standalone…
SkillBFreshness: BClaude Code +9 - 0.529score
please-help
peterchang77
Turn a vague request into a concise, safe execution brief through a short plain-language conversation.
SkillBFreshness: BClaude Code +9 - 0.526score
data-science-project-bootstrap
peterchang77
Create a new, reproducible data-science repository from scratch using uv, a src layout, explicit scientific contracts, and an intentionally chosen flat or…
SkillBFreshness: BClaude Code +9 - 0.526score
single-page-web
peterchang77
Create a reviewable standalone HTML page for small dashboards, data views, analysis, annotation, or project summaries, and choose a fuller web platform only…
SkillBFreshness: BClaude Code +9 - 0.523score
skill-writing
peterchang77
Design, write, and review concise, reusable skills for AI agents.
SkillBFreshness: BClaude Code +9 - 0.521score
large-corpus-llm-extraction
peterchang77
Plan and run a controlled large-scale model-assisted extraction or review workflow over free text.
SkillBFreshness: BClaude Code +9
Cite this page
docx skill. Agent Mart. https://agentmart.dev/skills/peterchang77-docx (last updated 2026-10-24)