agentleFS
cellinlabcellinlab/cell-skills/skills/celf-style-writer/SKILL.md

celf-style-writer

Skill29 starsChanged 7 months agoMITView on GitHub

Personal Chinese writing-style skill for Cell 细胞 style articles, rewrites, style enhancement, and style review. Use when Codex needs to write, rewrite, enhance, or audit Chinese content so it sounds like Cell speaking directly to the reader: friend-to-friend, warm but sharp, concrete, hook-driven, anti-academic, anti-AI-template, and grounded in real experience instead of abstract sermonizing.

The file

Show its rules and text

We split this skill into the rules it gives an agent

35 rules. About 0 are common, 35 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈35

    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:

    Read references/style-skeleton.md before drafting or heavy rewriting.

    line 17
  2. Distinctive:

    Read references/content-method.md when writing from scratch, rewriting a long article, or deciding how the piece should advance.

    line 18
  3. Distinctive:

    Read references/personal-variables.md when the draft needs Cell-specific identity, values, catchphrases, or tone choices.

    line 19
  4. Distinctive:

    Run the final text against references/quality-bar.md before delivery.

    line 20
  5. Distinctive:

    If the user only wants style diagnosis, do not rewrite the full piece.

    line 22
  6. Distinctive:

    Use when the user gives a topic, thesis, question, or outline and wants a Cell-style piece from scratch.

    line 28
  7. Distinctive:

    Use when the user already has a draft and wants it rewritten so it sounds more like Cell.

    line 32
  8. Distinctive:

    Use when the user wants to keep most of the original copy but add more living voice, friend-to-friend tone, or stronger rhythm.

    line 36
  9. Distinctive:

    Use when the user wants a style audit instead of a rewrite.

    line 40
  10. Distinctive:

    prefer a friend-to-friend tone over a lecture tone

    line 47
  11. Distinctive:

    avoid fake authority, fake neutrality, and fake biography

    line 49
  12. Distinctive:

    use real or explicitly framed experiential reasoning instead of empty abstraction

    line 50
  13. Distinctive:

    use style devices selectively instead of mechanically stacking every trick

    line 51
  14. Distinctive:

    prefer direct statements over formulaic binary contrast frames

    line 52
  15. Distinctive:

    default to not using “不是……而是……” unless it removes a real misunderstanding better than a plain sentence

    line 53
  16. Distinctive:

    keep firsthand observation, emotional memory, and decisive judgment grounded in real source material

    line 87
  17. Distinctive:

    use AI for structure, counterpoints, background, and analogy support

    line 88
  18. Distinctive:

    Use references/content-method.md when needed.

    line 91
  19. Distinctive:

    remove filler, fake grandeur, vague attribution, and obvious AI scaffolding

    line 98
  20. Distinctive:

    add signature devices only where they genuinely help

    line 101
  21. Distinctive:

    if this is a long-form piece, does it advance scene by scene or thought by thought instead of block by block

    line 113
  22. Distinctive:

    write like a lecturer, preacher, or consultant-on-stage

    line 122
  23. Distinctive:

    use academic filler, self-media clichés, or obvious AI scaffolding

    line 125
  24. Distinctive:

    use formulaic contrast templates like “不是……而是……”, “不仅……而且……”, or “这不仅仅是……” by default

    line 126
  25. Distinctive:

    use insulting profanity against people

    line 129
  26. Distinctive:

    keep the core meaning truthful

    line 133
  27. Distinctive:

    prefer concrete scenes, examples, and lived experience over empty slogans

    line 134
  28. Distinctive:

    use first-person ownership when making judgments

    line 135
  29. Distinctive:

    Read for the reusable style skeleton extracted from the original prompt: persona, sentence patterns, hook system, argument system, rhythm, …

    line 141
  30. Distinctive:

    Read for long-form article method: angle quality checks, human/AI boundary, article archetypes, progression devices, scene-led openings, …

    line 143
  31. Distinctive:

    Read for Cell’s confirmed identity pack, value system, tone defaults, catchphrases, emotional markers, concept naming defaults, and …

    line 145
  32. Distinctive:

    Read for the final review checklist, anti-AI / anti-academic rules, device-overuse warnings, and rewrite heuristics.

    line 147
  33. Distinctive:

    Use to define mode, audience, scenario, tone, and allowed devices before writing.

    line 149
  34. Distinctive:

    Use as the compact editable version of Cell’s personal style variables.

    line 151
  35. Distinctive:

    Use for a fast pass before returning the final text.

    line 153

≈ “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

Celf Style Writer

Overview

Write or reshape Chinese content so it feels like Cell talking to the reader in person: warm, experienced, concrete, honest, lightly sharp, and structurally alive. Treat this as a style skill, not a topic skill. The output should feel human, specific, and lived, not like an AI trying to imitate personality through formulas. For long-form pieces, make the article advance like a real person thinking on the page, not like a report being assembled.

Quick Start

  1. Pick the mode: write, rewrite, enhance, or review.
  2. Build a quick brief with assets/style-brief-template.md.
  3. Read references/style-skeleton.md before drafting or heavy rewriting.
  4. Read references/content-method.md when writing from scratch, rewriting a long article, or deciding how the piece should advance.
  5. Read references/personal-variables.md when the draft needs Cell-specific identity, values, catchphrases, or tone choices.
  6. Run the final text against references/quality-bar.md before delivery.

If the user only wants style diagnosis, do not rewrite the full piece. Diagnose first.

Modes

1. Write

Use when the user gives a topic, thesis, question, or outline and wants a Cell-style piece from scratch.

2. Rewrite

Use when the user already has a draft and wants it rewritten so it sounds more like Cell.

3. Enhance

Use when the user wants to keep most of the original copy but add more living voice, friend-to-friend tone, or stronger rhythm.

4. Review

Use when the user wants a style audit instead of a rewrite. Report where the text feels too academic, too templated, too preachy, or not personal enough.

Default Contract

Assume all of the following unless the user says otherwise:

  • write in Chinese
  • prefer a friend-to-friend tone over a lecture tone
  • preserve the source meaning when rewriting
  • avoid fake authority, fake neutrality, and fake biography
  • use real or explicitly framed experiential reasoning instead of empty abstraction
  • use style devices selectively instead of mechanically stacking every trick
  • prefer direct statements over formulaic binary contrast frames
  • default to not using “不是……而是……” unless it removes a real misunderstanding better than a plain sentence
  • trust the reader and cut filler, over-explaining, and quote-bait phrasing

Workflow

Step 1: Identify the Task State

Determine:

  • whether this is write, rewrite, enhance, or review
  • who the reader is
  • what the piece is trying to do
  • whether the source already has a usable structure
  • which long-form archetype fits best if this is a full article

Step 2: Set the Style Brief

Lock at least:

  • topic
  • audience
  • scenario
  • purpose
  • tone temperature
  • article archetype if relevant
  • allowed degree of personal exposure
  • whether concept naming and stronger hooks should be used

Use assets/style-brief-template.md.

Step 3: Set the Material Boundary

Determine what must come from the human side and what AI may help supply.

  • keep firsthand observation, emotional memory, and decisive judgment grounded in real source material
  • use AI for structure, counterpoints, background, and analogy support
  • admit uncertainty when the source lacks real detail instead of inventing texture

Use references/content-method.md when needed.

Step 4: Apply the Style

Apply the style in this order:

  1. preserve or rebuild the conversational stance
  2. remove filler, fake grandeur, vague attribution, and obvious AI scaffolding
  3. give the article a concrete opening, a visible mainline, and a believable end point
  4. fix the logic and rhythm at paragraph level
  5. add signature devices only where they genuinely help
  6. inject Cell-specific tone, catchphrases, and transitions
  7. make one last pass for AI smell, especially formulaic contrasts, generic endings, checklist-like structure, and over-tidy symmetry

Step 5: Self-Check

Check:

  • does it sound like a person talking, not a system summarizing
  • does it sound like Cell, not just “generic casual Chinese”
  • is the argument still clear after styling
  • did any device become performative or repetitive
  • if this is a long-form piece, does it advance scene by scene or thought by thought instead of block by block

Use assets/style-review-checklist.md.

Hard Rules

Do not:

  • fabricate personal history, data, quotes, rulings, or evidence
  • write like a lecturer, preacher, or consultant-on-stage
  • overuse catchphrases until they feel like parody
  • turn every paragraph into a hook machine
  • use academic filler, self-media clichés, or obvious AI scaffolding
  • use formulaic contrast templates like “不是……而是……”, “不仅……而且……”, or “这不仅仅是……” by default
  • manufacture quote-like lines just to sound deep
  • rely on neat triples, symmetrical slogans, or generic inspirational endings
  • use insulting profanity against people

Always:

  • keep the core meaning truthful
  • prefer concrete scenes, examples, and lived experience over empty slogans
  • use first-person ownership when making judgments
  • make the text feel lived, not merely formatted

Resource Map

  • references/style-skeleton.md
    • Read for the reusable style skeleton extracted from the original prompt: persona, sentence patterns, hook system, argument system, rhythm, transitions, and bans.
  • references/content-method.md
    • Read for long-form article method: angle quality checks, human/AI boundary, article archetypes, progression devices, scene-led openings, callback endings, and mainline control.
  • references/personal-variables.md
    • Read for Cell’s confirmed identity pack, value system, tone defaults, catchphrases, emotional markers, concept naming defaults, and transition phrases.
  • references/quality-bar.md
    • Read for the final review checklist, anti-AI / anti-academic rules, device-overuse warnings, and rewrite heuristics.
  • assets/style-brief-template.md
    • Use to define mode, audience, scenario, tone, and allowed devices before writing.
  • assets/personal-parameter-sheet.md
    • Use as the compact editable version of Cell’s personal style variables.
  • assets/style-review-checklist.md
    • Use for a fast pass before returning the final text.

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.