agentleFS
fockusfockus/writing-clearly/SKILL.md

writing-clearly

Skill8 starsChanged 4 months agoMITView on GitHub

Use when writing or editing ANY prose for humans, in Russian or English — documentation, README, commit and PR descriptions, reports, business letters, posts, announcements, release notes, error messages, UI text, explanations, summaries, chat answers. Триггеры — «напиши текст/пост/письмо/статью/README/доку», «отредактируй», «упрости», «сократи», «объясни просто», «сделай понятнее», «инфостиль», «по Ильяхову»; and English "write", "rewrite", "simplify", "explain", "edit this text", "make it clearer". Apply even when the user asks for prose without naming a style — clear writing is the default. Based on Maxim Ilyakhov's «Пиши, сокращай» and «Ясно, понятно» — clarity first, then brevity. NOT for source code, identifiers, commit subjects under 50 chars, or literary fiction.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. theyoungastronauts/polaris★ 4same name
    2. Sach1ng/PM-operating-OS★ 3same name
    3. ianklassen/ai-pm-system★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When NOT to use: source code and identifiers, literary fiction, legal text where exact boilerplate is required, texts where the user …

    line 29
  2. Distinctive:

    Test: if the header fits any other text on the topic, it's bad.

    line 54
  3. Distinctive:

    Diagonal reading test: Read only the headers and first sentences of paragraphs: the skeleton of the argument must be visible.

    line 57
  4. Very specific:

    Write, then pass through the text with these operations.

    line 64
  5. Distinctive:

    If the text was mostly good, minimal edits are the right answer.

    line 102
  6. Distinctive:

    Sameness: all sentences the same length reads robotic.

    line 148

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Writing Clearly — простой и понятный текст

Core principle

Care for the reader. A text works when the reader understands it on first pass and gets something useful from it. Two rules, in this order:

  1. Ясно — the reader must understand. Clarity comes first.
  2. Коротко — every word must earn its place. Brevity serves clarity, never replaces it.

Cutting words is not the goal. «Коротко — не значит ясно»: throw out the trash, but ADD tools of clarity — examples, concrete details, structure. A long clear text beats a short murky one.

When to use

Any prose a human will read: docs, README, PR descriptions, reports, letters, posts, release notes, error messages, UI text, explanations, chat answers — Russian or English.

When NOT to use: source code and identifiers, literary fiction, legal text where exact boilerplate is required, texts where the user explicitly asks for a different style (academic, SEO, poetic) — the user's explicit request always wins.

Workflow

Step 0 — Reader and context (before writing a single word)

From «Ясно, понятно»: a text fails most often not because of bad sentences, but because the author ignored the reader. Answer briefly (to yourself, don't output):

  • Who reads this? What do they already know? What terms are familiar to them?
  • Where do they meet the text? Chat, README on GitHub, email, commit log, error dialog.
  • What should they get? An action to take, a decision to make, an understanding to gain.
  • What's their likely objection or prejudice? Address it, don't ignore it.

Then pick the register: professional for docs, neutral for posts, conversational for chat. Formality never excuses murkiness — even a formal text must be clear on first read. Details and register scale: references/yasno-ponyatno.md.

Step 1 — Structure (подача)

  • Main idea first. In the text, in every section, in every paragraph. A reader who stops at any paragraph still leaves with the point.
  • Informative headers. A header states the topic and the benefit, not a teaser. Test: if the header fits any other text on the topic, it's bad. «Преимущества» → «Экономия 40% на отоплении». "Overview" → "What the tool does and when to use it".
  • One thought — one paragraph (3–5 sentences). Longer means two thoughts got mixed; split.
  • Diagonal reading test. Read only the headers and first sentences of paragraphs: the skeleton of the argument must be visible.
  • Every abstraction gets an example. Ideally an example AND a counterexample. If you state a rule, show it applied.

Step 2 — Editing (инфостиль)

Write, then pass through the text with these operations. Russian specifics and full tables: references/infostyle-ru.md. English: references/plain-english.md.

  1. Stop words out — вводный мусор («стоит отметить», «как известно»), паразиты времени («на сегодняшний день»), hedging, filler. Delete; meaning survives.
  2. Evaluations → facts — «качественный», «эффективный», "robust", "powerful": replace with the fact behind them («экономит 20% времени», "handles 10K rps") or delete.
  3. Nominalizations → verbs — «осуществление поддержки» → «поддерживаем»; "perform validation" → "validate".
  4. Passive → active, strong actor — «было принято решение» → «мы решили»; "the file is processed by the script" → "the script processes the file".
  5. Bureaucratese and pretentious words → plain words — «данный» → «этот», «является» → тире, «имплементация» → «внедрение»; "utilize" → "use", "leverage" → "use".
  6. Euphemisms → direct statements — «определённые сложности» → «серьёзные проблемы»; "suboptimal results" → "it failed". The reader must see there IS a problem.
  7. Unfounded claims out — «всё больше людей», «стремительно набирает популярность», "many developers believe": cite data or delete. Substitution test: if any subject fits the sentence, the sentence says nothing.
  8. Close synonyms → the strongest one — «долгого, нудного и утомительного» → «нудного», or better a fact: «два месяца». Beware the rule-of-three reflex.
  9. One thought per sentence; simplify syntax — break split constructions («не только… но и…» with long parts), participle chains, «который» nesting.
  10. Typography (Russian text) — «ёлочки», тире —/–, буква «ё», list punctuation.

Step 3 — Self-check (before returning the text)

Reread as a reader who never saw the draft:

  • Main idea in the first paragraph? In each section's first sentence?
  • Headers informative — would survive being read alone?
  • Every abstraction has a concrete example?
  • Any evaluation without proof left? Any unfounded claim?
  • Any «данный/является/осуществляется» (RU) or "utilize/leverage/delve" (EN)?
  • Passive where active would work?
  • Read aloud (inner voice): stumble → rewrite; out of breath → sentence too long.
  • Would you say this sentence to a colleague over coffee? If not — simplify.
  • Meaning preserved? Nothing added that wasn't there? Not over-edited?

If the text was mostly good, minimal edits are the right answer. Quality, not edit count.

Quick reference — top offenders

Плохо (RU)ХорошоBad (EN)Good
данныйэтотutilizeuse
является— (тире) / этоleverageuse
осуществляет продажупродаётperform an analysisanalyze
в настоящее времясейчасat this point in timenow
в целяхчтобыin order toto
было принято решениемы решилиit was decidedwe decided
определённые сложностисерьёзные проблемыcertain challengesserious problems
стоит отметить, что(удалить)it is worth noting that(delete)
широкий спектр(перечислить)a wide range of(list them)
в рамках проектав проектеwithin the framework ofin

Text-type adjustments

  • Technical docs / README — keep established terms (API, деплой, backend) even if «заумные»; kill bureaucratese around them. Precision over casualness.
  • PR / commit descriptions — what changed and why, main change first; facts, no «улучшена стабильность» without saying what exactly.
  • Business letters / internal comms — euphemisms die first: the reader must understand the real situation. Ask or state directly, politely.
  • Posts / announcements — personality allowed, first person welcome; rhythm varies; no corporate stamps («команда профессионалов»).
  • UI text / errors — shortest register: what happened + what to do next.
  • Chat answers — lead with the answer, then the reasoning.

Reference files

FileWhen to load
references/infostyle-ru.mdEditing or writing Russian text — full stop-word, канцелярит, заумь, эвфемизм tables + typography + AI-marker checklist
references/yasno-ponyatno.mdWriting from scratch, structuring a long text, choosing register, reader analysis
references/plain-english.mdWriting or editing English text
references/examples.mdBefore/after pairs by text type — load when unsure how far to edit

Common mistakes

  • Over-editing: stripping the author's voice along with the trash. Preserve meaning and intent; do not add facts that weren't there.
  • Brevity worship: cutting the example that made the text clear. Clarity first.
  • Simplifying terms for experts: «оптимизация» is fine in an SEO text; the reader scale decides, not the word list.
  • Sameness: all sentences the same length reads robotic. Vary short and long.
  • Announcing instead of saying: «Рассмотрим следующий вопрос…» — just say the thing.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.