agentleFS
jircdevjircdev/crew-plugin/skills/writing/SKILL.md

writing

Skill1 starsChanged 35 days agoMITView on GitHub

The writing craft, loadable by ANY role when authoring a piece meant to land with an audience — brief, manifesto, pitch deck, one-pager, essay, explanatory doc, speech, video script, email, announcement. Owns HOW it communicates (idea-force, narrative arc, segmentation, tone, impact principles), never the domain content. Load BEFORE drafting; triggers on 'write a brief/deck/speech/announcement', 'redactar', 'this needs to convince', or any deliverable whose success depends on an audience acting on it.

51 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use addskill, apply-firmware and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • ryanportfolioSkill

    addskill

    Create, import, update, or install repo or personal skills: runtime ownership, resources, discovery checks.

    Also in 6 projects that use this

  • ryanportfolioSkill

    apply-firmware

    Bring current folder (plain folder, existing repo, old or partial firmware copy) up to full Harness Firmware in place. Mirror external repo → adopt-repo.

    Also in 6 projects that use this

  • +10 more from ryanportfolio

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 45 edited versions we read keep these.

  • 2

    Distinctive

    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:

    Pull the WHAT from its owner: Read the project docs or consult the owning role.

    line 13
  2. Distinctive:

    Enforcement caveat: no single marker proves anything — humans use every one of them.

    line 51

“Common” is measured: kept by at least 6 in 10 of the 45 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 — the communication craft

Formerly the communications-strategist role; now a horizontal craft any role loads when it writes. The domain truth belongs to the owning role (the technical decision to system-architect, viability to commercial-strategist, the requirement to functional-analyst, the product truth to product-strategist); this skill governs how that truth lands.

Method

  1. Internalize objective and audience. What should they think, feel, or do afterward — and who exactly are they? A piece without a declared objective and audience is not ready to be written.
  2. Pull the WHAT from its owner. Read the project docs or consult the owning role; never invent or alter domain facts to make the narrative flow.
  3. Fix the idea-force: the single thing the audience must remember.
  4. Choose the impact principle that fits — golden circle (why → how → what), storytelling, framing, proof points, exponential framing — because the objective calls for it, never as decoration. State why.
  5. Design the arc: hook → development → climax → close, beat by beat; for each beat, the message, the visual support required (specify, don't design), the emotional intent.
  6. Write in the structure proper to the format. The format follows the objective, not a fixed template.
  7. Close with the call to action or the idea that stays.
  8. If spoken, add delivery notes: pacing, pauses, emphasis, where energy rises and falls.
  9. Hand off visual support as a per-beat specification to ux-architect (owner of the visual system); they design the slides or assets.

Craft rules

  • Segmentation: the same idea-force adapts to different audiences and formats without losing coherence — one project can yield a team induction and an external webinar the same week; they are different pieces.
  • Register and tone adapt to channel and audience; jargon the audience cannot decode gets glossed or cut.
  • Message before context: the why never arrives after the how. Burying the message under background is the most common failure.
  • Length sized to the decision, not to the template.
  • Senior judgment, not transcription: reorganizing the source material without adding communicational judgment is the anti-pattern this craft exists to prevent.

Voice — writing that does not read generated

Generated text betrays itself by structure more than by vocabulary: word lists age in months (a flagged term drops out of model output within a generation), while the structural scaffolds persist across model families. Attack structure first. These rules apply to every piece this skill governs, in every language.

Structures (the durable tells — banned):

  • Negative parallelism: "it's not X, it's Y", "not just X, but Y" — ES: "no se trata solo de…, sino de…", "no es X, es Y". State the claim directly; the contrast frame is a crutch.
  • "From X to Y" sweeps ("desde… hasta…") as a coverage gesture. Name what actually matters instead of gesturing at a range.
  • Symmetric hedges: "while X is true, Y also matters", "whether you're a beginner or an expert" — balanced clauses engineered to avoid committing. A piece with an objective takes a position.
  • Uniform rhythm: paragraphs of equal length, sentences of one shape, frictionless transitions, metronomic cadence. Human prose accelerates, digresses, stops short. Vary deliberately.
  • Closing and framing formulas: "in summary…", "it's important to note…" — ES: "en resumen…", "es importante destacar…".

Format:

  • No emoji in authored pieces — the ✅ especially is a statistical signature of generated text.
  • Headings in sentence case. Title Case is a tell in English and an anglicism in Spanish.
  • Prose where the content argues; bullets only where it enumerates. A wall of bullets with bolded lead-ins is the most recognizable generated layout.
  • The em dash is deliberately not banned: it is a weak marker (widely human-used, model-dependent) and this catalog's own house style. Density, not presence, is what reads generated.

Lexicon (weak signal, decays fast): trend words — "delve", "robust", "seamless", "pivotal", "leverage", "crucial", "tapestry"; ES: "robusto", "crucial", "clave" as the adjective for everything, "sin fisuras" — get replaced by the plain domain term. Do not maintain a banned-word list as if it were the defense; the structural rules above are the ones that hold.

Enforcement caveat: no single marker proves anything — humans use every one of them. When reviewing a piece, flag density and co-occurrence of tells, never one isolated marker.

Refuse to write

  • A piece without idea-force, objective, or audience — ask for them first.
  • One piece aimed at every audience at once.
  • A framework applied as decoration.
  • Facts invented or bent for narrative flow.

Boundaries

This skill owns the HOW of a written piece. It does not own: the domain content (the owning role), the visual system or screen design (ux-architect), public marketing-web strategy (commercial-strategist), or the structure and lifecycle of repository documentation (documentation-steward).

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.