AGENT SKILL

clarity

Reviews and repairs text that is hard to follow: unclear wording, buried answers, dense paragraphs, and structure that is hard to scan, including for readers with ADHD or dyslexia. Use whenever someone asks to make a passage clearer, easier to scan, or answer-first without changing its claims or voice.

License 许可
MIT
最近更新
2026-10-31
Stars
★ 7
Forks
0

01 前言 Front Matter

name
clarity
license
MIT
description
Reviews and repairs text that is hard to follow: unclear wording, buried answers, dense paragraphs, and structure that is hard to scan, including for readers with ADHD or dyslexia. Use whenever someone asks to make a passage clearer, easier to scan, or answer-first without changing its claims or voice.

02 资源文件 1

03 SKILL.md 正文

Clarity

Help the intended reader find the point and understand the next action. This pass reviews wording and structure. It does not run a full style rewrite or verify factual truth. Follow the target's instructions, terminology, canonical wording, and requested output format. Installing this Skill does not authorize publication or unrelated edits.

Establish the reading task

Read the complete passage and any supplied reader preferences. Identify the reader, purpose, and requested scope. Infer them when the context is clear. Consider their starting knowledge and the explanation they need to understand or act. Show respect and empathy through clear connections, accurate meaning, and no unnecessary reading effort. If a missing choice changes the result materially, ask before dependent edits. For mixed documents, distinguish stories, instructions, explanations, and reference sections. Preserve deliberate narrative openings. Keep technical instructions predictable.

Review wording and structure

Design for readers with varied attention, working memory, and reading needs, including ADHD and dyslexia. Do not infer a diagnosis or claim that most readers have one. Respect stated reader preferences. Reader comprehension takes priority over stylistic variety. Brevity removes unnecessary words. Keep the explanations, conditions, and connections this reader needs.

Reader attention is finite. Before writing or editing, predict how much attention this piece will receive. Use the reader, purpose, destination, and supplied preferences to make that prediction. Treat it as an editorial assumption, not a measured reading time. Let it govern scope, order, and depth across the complete output. Spend the most attention on what affects the reader's decision, understanding, or next action. Keep less important material brief, linked, or optional. Do not give every point equal space. Preserve critical conditions and warnings on the main reading path. Short sentences alone do not make an output worth the attention it requires.

  • Start simple. Lead with the answer or action. Then give the common case, an example, and deeper detail. For a story, preserve a deliberate narrative opening. Apply answer-first ordering to its technical sections. Define necessary terms before using them. Keep prerequisites and consequential warnings beside the action. Put optional exceptions and background later. A deep dive still needs a clear entry point.
  • Keep blocks small. Give each sentence one main idea and each paragraph one job. Prefer familiar words, explicit actors, and direct instructions. Explain necessary jargon on first use. As editing defaults, aim for sentences under 20 words and paragraphs of one to three sentences. Review three consecutive prose paragraphs for a useful heading, list, example, table, or diagram. These numbers are review prompts, not research thresholds. Preserve meaning and deliberate literary voice.
  • Make sections findable. Use descriptive headings for distinct questions, tasks, or reference entries. Keep a logical heading hierarchy. Avoid vague labels, decorative headings, and filler between sections. Readers should find the answer by scanning headings and opening lines.
  • Choose the useful format. Use bullets for parallel points, numbers for ordered steps, and tables for comparisons. Use diagrams for relationships or branching processes. Keep connected explanations in short prose. Add an artifact only when it reduces reading effort. Avoid long lists, wide tables, and decorative visuals.
  • Keep alternatives accessible. Explain a visual's useful point in nearby text. Supply meaningful text alternatives for images. Label table columns and keep cells short. Never make color, position, or an icon the only carrier of meaning. Use whitespace and restrained emphasis. Avoid full paragraphs in bold, italics, or capitals.

Leave clear sentences and useful headings unchanged. Do not split sentences mechanically or add filler for rhythm. Preserve stable anchors and check affected links when changing headings. Keep exact code, commands, names, numbers, units, attribution, uncertainty, and required wording. Keep conditions, negation, exceptions, permissions, and step order attached to their actions. Never invent an actor, explanation, experience, or example to make a passage sound clearer. Do not use em dashes or hyphens as sentence punctuation in new prose.

Deliver the requested pass

For review only, return concrete reading problems and proposed repairs. Leave the source unchanged. For an edit, change only passages with a reading problem. Explain meaningful repairs outside the edited text. Before delivery, check that the reader can find the answer, understand the terms, and follow the required steps. Check whether the whole output fits the predicted attention and gives the important points enough depth. Compare the result with the source for changed claims, conditions, uncertainty, and voice. If rendering is available, inspect the actual output at a narrow width. Otherwise report that limit when relevant. If no repair is needed, say so. Do not invent findings or claim an accessibility audit passed.

04 同仓库更多 Skills

harlan-zw
harlan-zw/brundlefly

verify-examples

Runs documentation examples, such as commands, code snippets, curl calls, and quickstarts, from their stated setup and checks the promised output. Use whenever someone asks whether the examples in a guide, README, or tutorial still work, without rewriting the whole document.

★ 7
harlan-zw
harlan-zw/brundlefly

technical-guide

Writes, verifies, or refreshes technical guides: tutorials, how-to guides, and developer documentation, checked against current code and official sources. Use whenever someone asks for a tutorial or how-to, asks to document how developers do a task, or asks to update stale docs, even when the guide looks short.

★ 7
harlan-zw
harlan-zw/brundlefly

slop-naming-recovery

Finds generic technical names that obscure user tasks and proposes names grounded in the product domain. Use when vague terms enter code, features, UI copy, or GLOSSARY.md, including implicit names and explicitly approved terms.

★ 7
harlan-zw
harlan-zw/brundlefly

release-notes

Writes or revises release notes, changelogs, release announcements, and upgrade guidance from verified changes. Use when someone asks what a release should say, wants GitHub release notes, or asks to make an existing release article more useful and concise.

★ 7
harlan-zw
harlan-zw/brundlefly

readme

Creates or refreshes a repository README: splash, project story, supported positioning, features, verified setup, and guides or API reference. Use whenever someone asks to write a README, explain why to adopt a project, or fix an outdated README section such as install or quickstart steps.

★ 7
harlan-zw
harlan-zw/brundlefly

pull-request-summary

Drafts, revises, or checks a pull request description against the actual diff and the repository's template. Use whenever someone asks for a PR description or body, wants one updated after the scope changed, or wants it checked before publishing, including inside a delivery Skill.

★ 7