readability-check
Runs a readability audit on a blog post draft or other multi-paragraph prose, calibrated for readers who read English as a second language. Checks ten categories — overall structure and topic order, paragraph structure, opening paragraph strength, tiered sentence length, passive voice, difficult words, filler and hedging, transitions, variation, and heading hierarchy — and reports a Flesch Reading Ease score with a per-category status. Use when the user asks to check readability, run a readability pass, or asks "is this readable", or proactively as a second pass after a substantial draft is complete. Also invoked by the github-repo, github-profile, and wp-readme-optimizer skills on their generated prose. For short strings (titles, meta descriptions, taglines, bios), use the `metadata-check` skill instead — Flesch and paragraph-level checks don't apply to them.
jdevalk · 已有 AI 审核记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add jdevalk/skills --skill readability-check技能说明
name: readability-check
description: >
Runs a readability audit on a blog post draft or other multi-paragraph
prose, calibrated for readers who read English as a second language.
Checks ten categories — overall structure and topic order, paragraph
structure, opening paragraph strength, tiered sentence length, passive
voice, difficult words, filler and hedging, transitions, variation,
and heading hierarchy — and reports a Flesch Reading Ease score with
a per-category status.
Use when the user asks to check readability, run a readability pass,
or asks "is this readable", or proactively as a second pass after a
substantial draft is complete. Also invoked by the github-repo,
github-profile, and wp-readme-optimizer skills on their generated
prose. For short strings (titles, meta descriptions, taglines, bios),
use the metadata-check skill instead — Flesch and paragraph-level
checks don't apply to them.
Readability check
Run a readability audit on a blog post draft or other multi-paragraph prose. Use when the user asks to check readability ("check readability", "readability pass", "is this readable"), or proactively after a substantial draft is complete — as a second pass after the blog-drafting skill's critical read, not during active drafting.
For short strings — page titles, meta descriptions, schema description fields, FAQ answers, profile bios, repo taglines — use the metadata-check skill. Flesch scoring and the ten-category rubric below don't fit a 5–30 word string and will mislead.
For whether a post earns a ranking — search intent fit, keyphrase placement, E-E-A-T, internal linking — chain into the content-seo skill after this audit. Readability is a prerequisite for ranking, not a substitute.
Audience calibration
Always assume the reader reads English as a second language. That's the default, not a fallback.
In technical posts, the technical sections can use domain terms the audience expects — but any non-technical paragraph (introduction, context, conclusion, transitions, examples, analogies) must be readable by a non-technical L2 reader. Setup and motivation paragraphs carry the post for readers who don't know the domain yet; they're where you lose people.
Conversational beats formal. Posts that address the reader directly ("you", "your") and occasionally ask them a question hold L2 readers far better than impersonal prose. Flag long stretches of detached, third-person register in non-technical sections.
How readers actually read
Readers scan before they commit. They look at the headings, the first paragraph, and the first sentence of each paragraph, and decide from those alone whether to read on. Search engines and AI systems weight the same elements when working out what a text is about. That's why the audit leans hard on those three places: a post whose headings and first sentences carry the argument works for scanners, full readers, and machines at once.
What to check
Read the full post, then report on each criterion below. For every issue, quote the specific text, reference its location (section heading or "intro" / "conclusion"), explain the problem, and suggest a concrete fix.
1. Overall structure and topic order
The post should read as if it was planned before it was written — topics grouped, ordered, and finished one at a time.
- The post should follow one recognizable ordering principle: thematic (aspect by aspect), chronological (old to new), didactic (easy to hard — best for explaining complicated subjects), problem–solution (state the problem, then the options), or inverted pyramid (most important first, details and background after — best for news and announcements).
- Related topics must sit together. Flag topic ping-pong: a subject discussed, dropped, and picked up again two sections later.
- The skim test: reading only the headings and the first sentence of each paragraph should yield the post's full argument. If a skimmer would miss a key point, it's buried mid-paragraph — surface it.
- The conclusion should restate the core message, using concluding signal words (so, in short, the takeaway is). If the intro opened with a story, question, or statistic, the conclusion should circle back to it — that closes the loop and makes the post feel complete.
2. Paragraph structure
A paragraph is a thematic unit, not a visual one. Whitespace placed for looks, with no shift in topic, breaks the reader's map of the text.
- Every paragraph must lead with its most important sentence — the core sentence. The opening sentence should make sense standalone: it's what scanners read and the unit AI systems extract for answers.
- The rest of the paragraph elaborates on that core sentence: supporting sentences that explain, evidence, or add nuance, optionally ending with a concluding or transitional sentence. Longer paragraphs benefit from a final summarizing sentence.
- One idea per paragraph. Break paragraphs that do two things.
- Each paragraph should be complete: say everything about its topic in one place. Flag the inverse of ping-pong — a paragraph break where the next paragraph just continues the same point (the break is aesthetic, not structural).
- Visual density matters more than a strict sentence count, but flag paragraphs over ~8 sentences or ~120 words. An overlong paragraph almost always hides two topics — split it by topic, not at an arbitrary midpoint.
- Mixing lengths is good: a run of identical-length paragraphs reads as monotone. Don't flag a short punchy paragraph between longer ones.
3. Opening paragraph
The first paragraph carries disproportionate weight — it's what AI systems quote and what readers use to decide whether to keep reading. There's no room for a print-style teaser that warms up to the point; web readers give you seconds. A good intro does three jobs: states the message, hooks the reader, and sets expectations. Check specifically:
- Does the f