convez.ai
SkillAutomaçãoIntermediário
Enriquemarq1

Zernio Library Skills

Enriquemarq1/zernio-library-skills

Distribua conteúdo em 13 plataformas sem sair do Claude

Ver no GitHub
BomPluginScore · de 0 a 100

Conjunto de skills Python para Claude que automatiza distribuição de conteúdo em 13 plataformas sociais, automação de comentários-para-DM e lançamento de campanhas Meta Ads. Funciona por manifesto declarativo, sem configuração local e sem publicação automática — você revisa tudo antes de enviar.

Instalação

$npx convez-ai Enriquemarq1/zernio-library-skills

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
54
Forks
13
Acompanhando
0
Saúde
42%
Último commit
há 3 semanas
Versão
v0.1.15
Issues abertas
0
Contribuidores
2

Casos de uso

  • Publicar vídeos, imagens e carrosséis em YouTube, TikTok, Instagram, LinkedIn e outras plataformas simultaneamente
  • Automatizar respostas a comentários com mensagens diretas em Instagram e Facebook
  • Lançar campanhas de anúncios Meta em lote com múltiplas variações criativas

Veredito

Skill robusta e prática para quem precisa gerenciar múltiplos canais sociais. O fluxo aprovação-antes-de-publicar é um acerto de segurança; a ausência de auto-publish evita acidentes. Requer chave Zernio e é voltada a equipes de marketing/conteúdo com API keys.

README

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

Zernio Library Skills

A library of Claude Code skills for Zernio. It ships:

  • zernio-publish — end-to-end social publishing across 13 platforms.
  • zernio-comment-to-dm — build Instagram/Facebook comment-to-DM automations + DM sequences on command. Tell Claude a post, a keyword, and the message — it creates the Zernio automation and orchestrates the follow-up DMs, so you never click through a dashboard or clone workflows. Reuse it for any new post/keyword in seconds.
  • meta-ads-launch — launch Meta (Facebook/Instagram) ad batches through Zernio's ads API. You decide the offer, copy, audience, and budget; Claude launches ALL your creative variations in one call (1 campaign + 1 ad set + N ads), then immediately pauses the campaign so nothing spends until you review and go live. Ships with the complete verified Zernio ads API reference (multi-creative shapes, idempotency, pause/resume, targeting, lead forms, click-to-WhatsApp). Requires the Zernio Ads add-on (included on usage-based plans).

More Zernio skills (analytics, calendaring, account management) will land over time.

You hand Claude an asset (video file, Drive link, image folder, URL) and tell it where to post. Claude analyzes it, drafts platform-tailored captions and titles and hashtags, converts formats if needed, uploads the media, shows you the full package for approval, schedules the post via the Zernio API, verifies it landed, and logs the result.

Install

Open the repo as your Claude Code project, or copy .claude/skills/zernio-publish/ into any Claude Code project's .claude/skills/ folder.

git clone https://github.com/Trejon-888/zernio-library-skills.git
cd zernio-library-skills

For claude.ai web (Skills upload UI), zip the skill folder yourself:

cd .claude/skills/zernio-publish && zip -r ../../../zernio-publish.skill.zip .

Then upload that ZIP on claude.ai → Capabilities → Skills.

Configure

Edit .env at the repo root, replace the placeholder with your real key from https://zernio.com/dashboard/api-keys, and run:

git update-index --skip-worktree .env

That keeps your local key edits from being staged into commits while the file stays tracked.

Use

In Claude Code, just say what you want to ship:

"Publish this video to my YouTube, TikTok, and Reels: ./media/launch.mp4"

"Drop a carousel post to Instagram and LinkedIn from this Drive folder: <drive-url>. Angle: three lessons from launching v1."

Claude reads the skill, runs through the workflow (intake → analyze → draft → convert → upload → package → approve → post → verify → log), shows you the full package, and ships on your approval.

What's in the box

zernio-library-skills/
├── .claude/
│   └── skills/
│       ├── zernio-publish/                    ← the publishing skill
│       │   ├── SKILL.md
│       │   ├── reference/                       deep API + per-platform docs
│       │   └── templates/manifest.json
│       ├── zernio-comment-to-dm/              ← comment-to-DM automations + sequences
│       │   ├── SKILL.md
│       │   ├── reference/zernio-automations-api.md
│       │   └── templates/automation.json
│       ├── meta-ads-launch/                   ← Meta ad batches via Zernio's ads API
│       │   ├── SKILL.md
│       │   ├── reference/zernio-ads-api.md      the complete verified ads API reference
│       │   └── templates/campaign-plan.json
│       └── skill-creator/                     ← Anthropic's official skill-creator
│                                                  bundled for when you add more skills
├── .env                                       ← ZERNIO_API_KEY (placeholder until you fill it)
├── CLAUDE.md                                  ← architectural map of the repo
├── README.md
├── LICENSE                                    ← MIT
└── .gitignore

No build scripts, no helper scripts, no auto-generated artifacts. The agent has its native Claude Code toolkit (bash, curl, ffmpeg, vision, transcript extraction, web fetching) and uses those directly.

Adding more skills to the library

Drop a new folder under .claude/skills/:

mkdir .claude/skills/my-new-zernio-skill
# Create SKILL.md with YAML frontmatter (name + description) and the skill's instructions

The bundled skill-creator at .claude/skills/skill-creator/ is Anthropic's official tool for designing and improving skills — invoke it when building a new one.

Design principles

  • End-to-end agency. Claude handles intake, analysis, drafting, conversion, upload, posting, verification, logging. Not transport-only.
  • The skill informs; the agent acts. SKILL.md is reference knowledge plus a suggested workflow — not a script. The agent uses its native tools to execute.
  • Approval before publish. Every post passes through an explicit gate. No bypass, no auto-publish.
  • Verify, don't trust. Zernio's 200 OK ≠ landed on the platform. Claude hits each platform's public URL or oEmbed after scheduledFor + 60s and surfaces any field drift.
  • Schedule, don't push. Multi-platform posts always use scheduledFor 2-3 minutes ahead.
Ver o original no GitHub

Bom para