agentleFS
trac3r00trac3r00/agent-skills/skills/nbj-write-clearly/SKILL.md

nbj-write-clearly

Skill0 starsChanged 3 months agoMITView on GitHub

Drafts, revises, and audits reader-first technical and product documentation. Use when working on developer docs, procedures, release notes, technical explanations, help-center content, or UI copy where clarity, source fidelity, accessibility, or global readability matters. Do not auto-apply it to marketing, legal, academic, fictional, or personal writing unless the user explicitly requests this style.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. About 0 are common, 16 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈16

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this order of precedence:

    line 19
  2. Distinctive:

    Follow the user's explicit request and the destination's requirements.

    line 21
  3. Distinctive:

    Put the result or purpose first.

    line 33
  4. Distinctive:

    Put a condition or circumstance before the instruction it controls.

    line 35
  5. Distinctive:

    Prefer familiar, precise words.

    line 36
  6. Distinctive:

    Use short sentences and paragraphs, but don't flatten every sentence into the same rhythm.

    line 37
  7. Distinctive:

    Remove throat-clearing, repeated conclusions, fake quotations, excessive claims, pre-announcements, clichés, idioms, and decorative …

    line 38
  8. Distinctive:

    Read references/guide.md when the artifact includes procedures, code, commands, UI labels, tables, images, accessibility requirements, or a …

    line 40
  9. Distinctive:

    Do not add facts, certainty, praise, urgency, or product claims.

    line 45
  10. Distinctive:

    Do not replace exact code, commands, filenames, API names, UI labels, or quotations with stylistic alternatives.

    line 47
  11. Distinctive:

    Do not rewrite when the user asked only for an audit.

    line 48
  12. Distinctive:

    Do not force documentation conventions onto dialogue, fiction, legal language, quotations, or a deliberately personal voice.

    line 49
  13. Distinctive:

    Do not make prose childish or robotic in the name of simplicity.

    line 50
  14. Distinctive:

    Before returning the result, check that:

    line 54
  15. Distinctive:

    every instruction names or clearly implies the actor.

    line 57
  16. Distinctive:

    Stop when the content is clear, accurate, consistent, and fit for its destination.

    line 66

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

NBJ Write Clearly

Outcome

Produce prose that lets the reader understand the point, identify the actor and action, and complete the task without rereading. Preserve the author's facts, intent, uncertainty, and useful voice.

Apply the right authority

Use this order of precedence:

  1. Follow the user's explicit request and the destination's requirements.
  2. Preserve source facts, quotations, code, UI labels, product names, and intentional terminology.
  3. Follow project-specific style.
  4. Apply this skill's Google-derived guidance.
  5. Consult other references only when the preceding sources are silent.

Depart from a guideline when doing so makes the content clearer for its actual readers. Stay consistent after making that choice.

Write or revise

  1. Identify the reader, their goal, the artifact type, and whether the task is to draft, revise, or audit.
  2. Mark content that must not drift: facts, claims, qualifications, quoted language, technical tokens, links, and required structure.
  3. Put the result or purpose first. Give each paragraph one idea and put critical information early.
  4. Name the actor. Prefer active voice, present tense, and second person when addressing the reader. Use imperatives for steps.
  5. Put a condition or circumstance before the instruction it controls.
  6. Prefer familiar, precise words. Define necessary jargon or abbreviations on first use. Keep one term for one concept.
  7. Use short sentences and paragraphs, but don't flatten every sentence into the same rhythm. Use common contractions when they sound natural.
  8. Remove throat-clearing, repeated conclusions, fake quotations, excessive claims, pre-announcements, clichés, idioms, and decorative metaphors.
  9. Structure for scanning: sentence-case headings, numbered lists for sequences, bullets for parallel items, and descriptive links.
  10. Read references/guide.md when the artifact includes procedures, code, commands, UI labels, tables, images, accessibility requirements, or a line-level style audit.
  11. Read references/official-index.md only when the user requests Google Style Guide compliance or the task turns on a specialized rule such as word choice, product naming, punctuation, grammar, dates, units, mathematical notation, HTML, Markdown, filenames, or trademarks. Consult only the relevant official page when browsing is available.

Protect meaning and voice

  • Do not add facts, certainty, praise, urgency, or product claims.
  • Preserve source modality such as can, might, should, and will. Do not change it only to satisfy a tense preference.
  • Do not replace exact code, commands, filenames, API names, UI labels, or quotations with stylistic alternatives.
  • Do not rewrite when the user asked only for an audit. Report findings in priority order and give bounded examples.
  • Do not force documentation conventions onto dialogue, fiction, legal language, quotations, or a deliberately personal voice.
  • Do not make prose childish or robotic in the name of simplicity. Prefer the clearest accurate term, even when it is technical.

Validate

Before returning the result, check that:

  • the opening answers the reader's main question;
  • every instruction names or clearly implies the actor;
  • conditions appear before the actions they govern;
  • each pronoun has an unambiguous referent;
  • terminology, capitalization, and formatting are consistent;
  • claims remain factual, scoped, and supported by the source;
  • a global reader can understand the prose without decoding slang or culture-specific references;
  • the revision preserves all required facts, caveats, and technical tokens;
  • the output still sounds like the author when voice matters.

Stop when the content is clear, accurate, consistent, and fit for its destination. Do not keep polishing lines that already do their job.

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.