菜单
Evaluates grammar, spelling, and formatting in design systems' documentation. Use when checking mechanics like spelling, punctuation, heading hierarchy, and list formatting.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/canonical/ds-docs-meta-prompting --skill copy-editing-evaluation命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | copy-editing-evaluation |
| description | Evaluates grammar, spelling, and formatting in design systems' documentation. Use when checking mechanics like spelling, punctuation, heading hierarchy, and list formatting. |
Hard technique for evaluating grammar, spelling, and formatting in design systems documentation.
copy-editing script to get objective findingsThe deterministic pass catches spelling errors, flagged words, heading issues, and formatting problems. The qualitative pass covers everything else: list structure, link quality, terminology consistency, and punctuation style.
Run copy-editing.js from the same directory as this skill file. Pass the path to a markdown file as the only argument.
node <skill-directory>/copy-editing.js <file.md>
The script:
Multi-word phrases in @.claude/dockit/references/flagged-words.txt (e.g. "in order to", "end user") are not detected by the script. Flag those during the qualitative pass.
The script outputs a JSON object with five keys:
{
"spelling": [
spelling — misspelled words detected by cspell. Each entry has the word, the line number (1-indexed), and the character offset from the start of the file.
forbiddenWords — words from @.claude/dockit/references/flagged-words.txt found in the document. Entries with a suggestion field are words that have a preferred replacement. Entries without suggestion are strictly forbidden and should be removed or the sentence restructured.
headings — heading hierarchy and capitalisation issues. The type field is one of skipped-level, multiple-h1, or not-sentence-case. The message field provides a human-readable description.
formatting — punctuation issues. Currently detects em-dashes (—).
summary — counts of issues by category and a total across all categories.
The copy editor checks mechanics: grammar, spelling, punctuation, formatting, and structural consistency. It does not evaluate:
If a sentence is grammatically correct but does not match the style guide's tone, do not flag it. If a section is thin, do not flag it. Stay in your lane.
The following checks are performed by the LLM during the qualitative pass. The script does not cover these.
The script only flags single words. During the qualitative pass, also check for multi-word phrases from @.claude/dockit/references/flagged-words.txt such as "in order to", "for the purpose of", "end user", and "in conjunction with".
Common terms and their preferred spellings. Flag deviations.
| Preferred | Common variants to flag |
|---|---|
| dropdown | drop-down, drop down, DropDown |
| checkbox | check box, check-box, CheckBox |
| tooltip | tool tip, tool-tip, ToolTip |
| popover | pop over, pop-over, PopOver |
| dialog | dialogue (in UI context), Dialog |
| modal | Modal (when not a proper noun) |
| open source | open-source |
| sidebar | side bar, side-bar |
| navbar | nav bar, nav-bar, navigation bar |
| login (noun/adj) | log in (noun/adj) |
| log in (verb) | login (verb) |
| setup (noun/adj) | set up (noun/adj) |
| set up (verb) | setup (verb) |
| frontend | front-end, front end |
| backend | back-end, back end |
| webpage | web page |
The style guide at @.claude/dockit/references/style-guide.md has its own terminology section. If it specifies a different preference, the style guide wins.
All items in a list should follow the same grammatical structure.
Good (all imperative):
Bad (mixed structures):
Use the explore-existing-design-systems-documentation skill to check:
... not . . .<output_format>
Each finding:
[severity] Section: {section name} (line ~{number})
Issue: {description}
Current: "{the current text}"
Suggested: "{the corrected text}"
Rule: {which rule this falls under}
<severity_levels>
[critical] — Spelling error in a heading, broken code example, skipped heading level[suggestion] — Inconsistent formatting, parallel structure issue[nitpick] — Minor punctuation, optional style preference</severity_levels>
Group findings by section, then by severity within each section.
End with a summary:
Copy Editing Summary:
Spelling: {count} findings
Headings: {count} findings
Lists: {count} findings
Links: {count} findings
Terminology: {count} findings
Punctuation: {count} findings
Formatting: {count} findings
Total: {count} findings ({critical} critical, {suggestions} suggestions, {nitpicks} nitpicks)
</output_format>