LinkedIn Post Skill
ChanMeng666/linkedin-post-skill
Redige posts LinkedIn de marca com visuais gerados e drafta no compositor
Ver no GitHubSkill para Claude Code que redige posts em inglês para LinkedIn (ideias ou recaps de eventos), gera visuais de marca (Mermaid, HTML-to-PNG via Playwright, fallback OpenAI) e os drafta diretamente no compositor do LinkedIn. Nunca posta automaticamente — sempre aguarda revisão.
Instalação
npx convez-ai ChanMeng666/linkedin-post-skill1 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
Casos de uso
- Draftar posts de autoridade com frameworks e opiniões para LinkedIn
- Criar recaps de eventos com crédito e tags automáticas de participantes
- Gerar visuais de marca consistentes (Mermaid, HTML, ou OpenAI) para posts
Veredito

Ferramenta sólida e bem-arquitetada para automação de conteúdo em LinkedIn. O fluxo de pesquisa/verificação de nomes antes de taggers é um diferencial; o sistema de visuais em três camadas (Mermaid → HTML → OpenAI) é pragmático. Código limpo, mas exige setup local (Playwright, chaves OpenAI opcionais) e integração manual com Claude Code.
README
A tradução para o português deste README ainda está na fila.
linkedin-post — a Claude Code skill for on-brand LinkedIn posts
A Claude Code skill that writes English LinkedIn posts for personal-brand marketing, generates an on-brand visual, and drafts the post straight into the LinkedIn composer for review. It never auto-posts — it stages the draft and stops so you review and publish.
Two modes:
- idea — share a framework or opinion to build authority with your network.
- recap — reflect on an offline tech event, crediting and tagging speakers, organizers, and companies to grow your network. It researches and verifies names/titles/handles before tagging.
How it works
/linkedin-post idea # or: /linkedin-post recap (or just describe what you want)
SKILL.md orchestrates the flow:
- Gather inputs — the idea/angle, or the event details and people to credit.
- Research & verify (recap) — confirm exact names, titles, employers, and LinkedIn URLs.
- Draft the copy — strong hook → emoji-led framework → mental shift → CTA → hashtags, in a consistent voice (architecture-grade, evidence-first, no marketing fluff).
- Generate one visual (priority order): Mermaid diagram → HTML-rendered image → OpenAI image (fallback).
- Draft into the composer via Claude-in-Chrome, then stop for your review.
Visuals
All visuals follow a single design system ("Caldera": a basalt canvas, one Digital-Orange accent, Anton + DM Sans type, flat color blocks). Three ways to make one:
| Tool | Script | Best for |
|---|---|---|
Mermaid (mmdc) | scripts/render-mermaid.mjs | frameworks, before/after, concept maps |
| HTML → PNG (Playwright, 2× DPI) | scripts/render-card.mjs (quick card) · scripts/render-html.mjs (bespoke layouts) | polished quote/recap cards and custom designed visuals |
OpenAI image (gpt-image-1) | scripts/gen-image.mjs | abstract hero images, used only as a fallback |
Author a bespoke visual by writing an HTML file that links templates/brand.css (brand tokens +
helpers), then screenshot it with render-html.mjs.
Repository layout
SKILL.md The orchestrator (invoked as /linkedin-post)
references/ Brand tokens, post voice/structure, hashtag bank, recap playbook
templates/ Mermaid brand theme · brand.css · post-card.html · logo asset
scripts/ render-mermaid · render-card · render-html · gen-image · lib/load-env
examples/ A real reference post + its diagram
output/ Generated drafts (gitignored)
Setup
npm install
npx playwright install chromium
# Optional — only needed for the OpenAI image fallback:
cp .env.example .env # then set OPENAI_API_KEY (the loader also reads .env.local)
Without an OpenAI key, the image step falls back to a Mermaid diagram or HTML visual — everything
else works. To make the skill invocable as /linkedin-post, expose this folder to Claude Code's
skills directory (e.g. a junction/symlink at ~/.claude/skills/linkedin-post).
Security
API keys are read from the environment or a local, git-ignored .env / .env.local file — they are
never committed. .gitignore excludes .env, .env.local, node_modules/, and output/. Use
.env.example as the template.
Notes
This is a personal-brand tool tuned to one author's identity and voice; the brand system, hashtag
bank, and reference copy under references/ are meant to be edited to fit your own. Built as a
demonstration of composing Claude Code skills with browser automation, Mermaid, Playwright, and
image generation.