convez.ai
SkillEscrita & ConteúdoIniciante
shaysega

Maybe Don't Post That

shaysega/maybe-dont-post-that

Revisa posts de liderança antes de publicar; separa provocativo honesto de punch-down

Ver no GitHub
InicialPluginScore · de 0 a 100

Uma skill para Claude que revisa comunicações públicas de líderes antes da publicação, identificando tom contemptivo ou passível de mal-entendimento. Mantém a força e opinião, mas elimina armadilhas retóricas que geram backlash — ideal para posts em LinkedIn/X, anúncios internos e comunicados críticos.

Instalação

$npx convez-ai shaysega/maybe-dont-post-that

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 5 meses
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Revisar posts em redes sociais antes de publicar
  • Avaliar anúncios de layoff e comunicados sensíveis
  • Corrigir tom de manifestos de cultura e cartas aos clientes
  • Decidir se deve responder ou esclarecer um post que viralizou negativamente

Veredito

Skill bem-focada e prática para líderes que escrevem em público. O conceito é sólido — separar intenção de impacto real — e a execução oferece feedback granular (red flags, reescrita, análise). Recomendado para quem quer manter edge sem derrapar em contempt.

README

A tradução para o português deste README ainda está na fila.

Maybe Don't Post That

A Claude skill that pressure-tests and rewrites public-facing leadership communication before it ships - LinkedIn/X posts, job listings, all-hands notes, layoff announcements, "culture" manifestos, customer apologies.

It catches the specific failure mode where a draft feels bold and motivating to the author but reads as contemptuous to everyone else - the kind of post that ends in a clarification thread. It is not a blandness filter: strong, opinionated, demanding writing is encouraged. The job is to separate "provocative because it says something true and hard" from "provocative because it punches down."

The one idea it runs on

Words travel without their author. Once something is published, the context stays behind: your tone, your intent, the room you imagined, what you meant. Only the words travel - and they have to survive on their own, in front of people who don't know you and won't fill in the gaps in your favor.

So the core test for any draft is: does it hold up with nothing but the words - stripped of your intent, your context, your tone? If what's left still says what you want it to say, ship it. If defending it would take "but what I meant was…", that meaning never made it onto the page - fix it now.

What you get

When you hand it a draft, it responds with five sections:

  • Verdict - ship as-is / ship with edits / don't ship yet
  • Red flags - each problem named, with the offending line quoted
  • Stronger rewrite - the full revised text, ready to paste
  • Why this works - the specific swaps, so the lesson transfers
  • Residual risk - how the rewrite could still be read badly, so the call is yours

Install

Claude apps (Claude.ai / Desktop) - easiest

Upload the packaged skill file:

  1. Download maybe-dont-post-that.skill.
  2. In Claude, go to Settings → Capabilities → Skills (location varies by plan) and upload the .skill file.

Claude Code (CLI)

Drop the skill folder into your personal skills directory:

git clone https://github.com/shaysega/maybe-dont-post-that.git
mkdir -p ~/.claude/skills/maybe-dont-post-that
cp maybe-dont-post-that/SKILL.md ~/.claude/skills/maybe-dont-post-that/SKILL.md

It loads automatically the next time you start Claude Code. Verify with /skill (or by asking Claude to review a draft post).

How it triggers

You don't have to invoke it by name. It fires when you're about to publish something public and say things like "review this post," "make this sound right," "punch this up," or even just "is this fine to post?" - and also after a post blows up, to decide whether to clarify, correct, or stay quiet.

Files in this repo

FileWhat it is
SKILL.mdThe skill itself - human-readable source. This is the thing to edit.
maybe-dont-post-that.skillThe packaged bundle (a zip of SKILL.md) for one-click upload to Claude apps. Regenerate it after editing SKILL.md.

To repackage after editing:

mkdir -p .pkg/maybe-dont-post-that && cp SKILL.md .pkg/maybe-dont-post-that/
( cd .pkg && zip -r ../maybe-dont-post-that.skill maybe-dont-post-that ) && rm -rf .pkg

License

MIT - see LICENSE.

Ver o original no GitHub

Bom para

Relacionados